WONDERWARE 2013 Historian Concepts PDF

Wonderware
Historian Concepts
Guide
9/27/13
All rights reserved. No part of this documentation shall be reproduced, stored in a retrieval system, or
transmitted by any means, electronic, mechanical, photocopying, recording, or otherwise, without the
prior written permission of Invensys Systems, Inc. No copyright or patent liability is assumed with respect
to the use of the information contained herein. Although every precaution has been taken in the
preparation of this documentation, the publisher and the author assume no responsibility for errors or
omissions. Neither is any liability assumed for damages resulting from the use of the information
contained herein.
The information in this documentation is subject to change without notice and does not represent a
commitment on the part of Invensys Systems, Inc. The software described in this documentation is
furnished under a license or nondisclosure agreement. This software may be used or copied only in
accordance with the terms of these agreements.
© 2002-2010, 2012-2013 by Invensys Systems, Inc. All rights reserved.
Invensys Systems, Inc.

26561 Rancho Parkway South
Lake Forest, CA 92630 U.S.A.
(949) 727-3200
http://www.wonderware.com
For comments or suggestions about the product documentation, send an e-mail message to
[email protected].
All terms mentioned in this documentation that are known to be trademarks or service marks have been
appropriately capitalized. Invensys Systems, Inc. cannot attest to the accuracy of this information. Use of
a term in this documentation should not be regarded as affecting the validity of any trademark or service
mark.
Alarm Logger, ActiveFactory, ArchestrA, Avantis, DBDump, DBLoad, DT Analyst, Factelligence,
FactoryFocus, FactoryOffice, FactorySuite, FactorySuite A2, InBatch, InControl, IndustrialRAD,
IndustrialSQL Server, InTouch, MaintenanceSuite, MuniSuite, QI Analyst, SCADAlarm, SCADASuite,
SuiteLink, SuiteVoyager, WindowMaker, WindowViewer, Wonderware, Wonderware Factelligence, and
Wonderware Logger are trademarks of Invensys plc, its subsidiaries and affiliates. All other brands may
be trademarks of their respective owners.
3
Contents
Welcome .................................................. 11
Wonderware Historian Documentation Set .................................. 11
Documentation Conventions ......................................................... 12
Technical Support .......................................................................... 13
Chapter 1 Introduction .............................................. 15

The Wonderware Historian Solution ............................................ 16
Process Data ................................................................................ 16
Wonderware Historian Benefits ................................................. 17
Integration with Microsoft SQL Server ........................................ 18
Support for SQL Clients ............................................................. 18
Wonderware Historian Subsystems .............................................. 19
Chapter 2 System-Level Concepts and Functionality ...... 21

About Tags ...................................................................................... 21
VTQ for Tags ............................................................................... 23
Tag Properties ............................................................................. 24
Security ........................................................................................... 26
Windows Operating System Security ........................................ 27
SQL Server Security ................................................................... 28
Management Console Security ................................................... 33
Default Access Rights for Different Operating Systems ........... 34
Wonderware Historian Concepts Guide

4 Contents
Time Handling ............................................................................... 35

System Parameters ........................................................................ 36
System Messages ........................................................................... 41
Wonderware Historian Processes .................................................. 42
The System Driver and System Tags ............................................ 44
Error Count Tags ........................................................................ 44
Date Tags ..................................................................................... 44
Time Tags .................................................................................... 45
Storage Space Tags ..................................................................... 45
I/O Statistics Tags ....................................................................... 45
System Monitoring Tags ............................................................. 46
Miscellaneous (Other) Tags ........................................................ 47
Classic Event Subsystem Tags ................................................... 48
Event Storage System Tags ........................................................ 48
Replication Subsystem Tags ....................................................... 49
Performance Monitoring Tags .................................................... 51
Supported Protocols ....................................................................... 53
Modification Tracking .................................................................... 53
Modification Tracking for Configuration Changes .................... 54
Modification Tracking for Historical Data Changes ................. 55
Data Categories .............................................................................. 56
Original Data and Revision Data ............................................... 57
Original Streamed Data ............................................................. 57
Original Non-Streamed Data ..................................................... 59
Latest Revision Data ................................................................... 60
New-Original Revision Data ....................................................... 60
Data Quality ................................................................................... 61
Viewing Quality Values and Details .......................................... 62
Acquisition and Storage of Quality Information ....................... 65
Client-Side Quality Values ......................................................... 66
Chapter 3 Configuration Subsystem ............................ 69

Configuration Subsystem Components ......................................... 70
About the Runtime and Holding Databases ................................. 70
The Runtime Database ............................................................... 71
The Holding Database ................................................................ 72
About the Configuration Service ................................................... 72
Dynamic Configuration .................................................................. 72
Effects of Configuration Changes on the System ...................... 74
Cases in Which Configuration Changes are not Committed .... 75

Contents5
Chapter 4 Data Acquisition Subsystem ........................ 77

Data Acquisition Components ....................................................... 78
Data Acquisition from I/O Servers ................................................ 79
I/O Server Addressing ................................................................. 80
About IDASs ................................................................................ 81
I/O Server Redundancy ............................................................... 90
Redirecting I/O Servers to InTouch HMI Software ................... 91
Time Synchronization for Data Acquisition .............................. 91
Data Acquisition by Means of INSERT and UPDATE
Statements ...................................................................................... 94
Data Acquisition from the Historian Client Access Layer
(HCAL) ............................................................................................ 94
Initial HCAL Data Collection without a Connection to the
Historian ................................................................................. 95
Storing Data through HCAL ...................................................... 95
Data Adjustments Performed by HCAL .................................... 96
Data Retrieval through HCAL ................................................... 97
HCAL Security ............................................................................ 97
Data Acquisition from the Manual Data Acquisition Service
(MDAS) ........................................................................................ 98
Importing Data from a CSV File ................................................... 99
Chapter 5 Data Storage Subsystem........................... 101

Storage Subsystem Processes ...................................................... 102
Storage Modes .............................................................................. 102
"Forced" Storage ........................................................................... 103
Delta Storage ................................................................................ 103
Time and Value Deadbands for Delta Storage ........................ 104
"Swinging Door" Deadband for Delta Storage ......................... 104
Cyclic Storage ............................................................................... 113
Data Conversions ......................................................................... 114
History Blocks .............................................................................. 114
History Block Notation ............................................................. 115
History Block Creation ............................................................. 116
History Block Storage Locations .............................................. 116
Automatic Deletion of History Blocks ...................................... 118
Storage and Classic Storage ........................................................ 120
Identifying Which Tags are Stored by Which Storage
Subsystem .................................................................................. 121
Upgrading Tags from Classic Storage ..................................... 121

6 Contents
Chapter 6 Classic Data Storage Subsystem ................ 125

Differences between Storage and Classic Storage ..................... 126
Classic Storage Subsystem Processes ......................................... 126
About the Real-Time Data Window for Classic Storage ............ 127
Determining If the Real-Time Window Is Configured
Appropriately for a Swinging Door Deadband .................... 128
Disk Requirements and Performance Considerations for a
Swinging Door Deadband ..................................................... 129
Reserved Values for Classic Storage ........................................... 129
About the Active Image for Classic Storage ............................... 130
Automatic Resizing of the Active Image .................................. 130
How the Active Image Storage Option Affects Data
Retrieval .................................................................................... 132
Dynamic Configuration Effects on Classic Storage .................... 132
Memory Management for Classic Storage .................................. 133
Chapter 7 Data Retrieval Subsystem ......................... 135

Data Retrieval Features .............................................................. 136
History Blocks: A SQL Server Remote Data Source .................. 136
Retrieval Service .......................................................................... 137
About the Wonderware Historian OLE DB Provider ................. 137
Extension (Remote) Tables for History Data ........................... 139
Query Syntax for the Wonderware Historian OLE DB
Provider ..................................................................................... 140
Wonderware Historian OLE DB Provider Unsupported Syntax
and Limitations ..................................................................... 144
Linking the Wonderware Historian OLE DB Provider to the
Microsoft SQL Server ........................................................... 153
Wonderware Historian Time Domain Extensions ...................... 154
Wonderware Historian I/O Server .............................................. 156
Chapter 8 Data Retrieval Options .............................. 157

Understanding Retrieval Modes ................................................. 157
Cyclic Retrieval ......................................................................... 158
Delta Retrieval .......................................................................... 162
Full Retrieval ............................................................................ 169
Interpolated Retrieval ............................................................... 171
"Best Fit" Retrieval ................................................................... 177
Average Retrieval ...................................................................... 182
Minimum Retrieval ................................................................... 188
Maximum Retrieval .................................................................. 194

Integral Retrieval ...................................................................... 200

Slope Retrieval .......................................................................... 203
Counter Retrieval ...................................................................... 206
ValueState Retrieval ................................................................. 211
RoundTrip Retrieval ................................................................. 218
Understanding Retrieval Options ............................................... 222
Which Options Apply to Which Retrieval Modes? ................... 223
Using Retrieval Options in a Transact-SQL Statement ......... 224
Cycle Count (X Values over Equal Time Intervals)
(wwCycleCount) .................................................................... 225
Resolution (Values Spaced Every X ms) (wwResolution) ....... 228
About "Phantom" Cycles ........................................................... 230
Time Deadband (wwTimeDeadband) ....................................... 232
Value Deadband (wwValueDeadband) .................................... 236
History Version (wwVersion) .................................................... 239
Interpolation Type (wwInterpolationType) ............................. 241
Timestamp Rule (wwTimestampRule) ..................................... 244
Time Zone (wwTimeZone) ......................................................... 246
Quality Rule (wwQualityRule) ................................................. 247
State Calculation (wwStateCalc) ............................................. 255
Analog Value Filtering (wwFilter) ........................................... 257
Selecting Values for Analog Summary Tags
(wwValueSelector) ..................................................................... 263
Edge Detection for Events (wwEdgeDetection) ....................... 266
Chapter 9 Query Examples....................................... 277

Querying the History Table ......................................................... 277
Querying the Live Table .............................................................. 278
Querying the WideHistory Table ................................................ 279
Querying Wide Tables in Delta Retrieval Mode ......................... 280
Querying the AnalogSummaryHistory View .............................. 281
Querying the StateSummaryHistory View ................................. 282
Using an Unconventional Tagname in a Wide Table Query ..... 283
Using an INNER REMOTE JOIN ............................................... 283
Setting Both a Time and Value Deadband for Retrieval ........... 284
Using wwResolution, wwCycleCount, and wwRetrievalMode in the
Same Query ............................................................................... 287
Determining Cycle Boundaries ................................................... 288
Mixing Tag Types in the Same Query ........................................ 288
Using a Criteria Condition on a Column of Variant Data ......... 289
Using DateTime Functions .......................................................... 290
Using the GROUP BY Clause ..................................................... 292

8 Contents
Using the COUNT() Function ..................................................... 292

Using an Arithmetic Function ..................................................... 293
Using an Aggregate Function ...................................................... 294
Making and Querying Annotations ............................................. 296
Using Comparison Operators with Delta Retrieval ................... 296
Using Comparison Operators with Cyclic Retrieval and Cycle
Count ......................................................................................... 301
Using Comparison Operators with Cyclic Retrieval and
Resolution ..................................................................................... 305
Returning Time Between Value Changes ................................... 308
SELECT INTO from a History Table .......................................... 314
Moving Data from a SQL Server Table to an Extension Table . 315
Using Server-Side Cursors .......................................................... 316
Using Stored Procedures in OLE DB Queries ............................ 318
Querying Data to a Millisecond Resolution using SQL
Server 2005 ................................................................................... 318
Getting Data from the OPCQualityMap Table .......................... 320
Using Variables with the Wide Table ......................................... 320
Retrieving Data Across a Data "Hole" ........................................ 321
Returned Values for Non-Valid Start Times .............................. 323
Retrieving Data from History Blocks and the Active Image ..... 323
Querying Aggregate Data in Different Ways ............................. 324
Chapter 10 Replication Subsystem ............................ 327

About Tiered Historians .............................................................. 327
How Tags are Used During Replication ...................................... 329
Simple Replication ....................................................................... 331
Summary Replication .................................................................. 332
Analog Summary Replication ................................................... 333
State Summary Replication ...................................................... 334
Replication Schedules .................................................................. 335
Replication Schedules and Daylight Savings Time ................. 336
Replication Groups ....................................................................... 337
How Replication is Handled for Different Types of Data .......... 339
Streaming Replication .............................................................. 339
Queued Replication ................................................................... 340
Tag Configuration Synchronization between Tiered Historians 341
Replication Run-time Operations ................................................ 341
Replication Latency ................................................................... 341
Replication Delay for "Old" Data .............................................. 342

Continuous Operation ............................................................... 342

Overflow Protection ................................................................... 342
Security for Data Replication ...................................................... 343
Using Summary Replication instead of Event-Based
Summaries .................................................................................... 343
Chapter 11 Event Subsystem ................................... 345

Event Subsystem Components .................................................... 346
Uses for the Event Subsystem ..................................................... 347
Event Subsystem Features and Benefits .................................... 348
Event Subsystem Performance Factors ...................................... 349
Event Tags .................................................................................... 350
Event Detectors ............................................................................ 350
SQL-Based Detectors ................................................................ 351
Schedule Detectors .................................................................... 353
External Detectors .................................................................... 354
Event Actions ............................................................................... 354
Generic SQL Actions ................................................................. 354
Snapshot Actions ....................................................................... 355
E-mail Actions ........................................................................... 355
Deadband Actions ..................................................................... 356
Summary Actions ...................................................................... 356
Event Action Priorities ............................................................. 358
Event Subsystem Resource Management ................................... 358
Detector Thread Pooling ........................................................... 359
Action Thread Pooling ............................................................... 360
Event Subsystem Database Connections ................................. 361
Handling of Event Overloads and Failed Queries ................... 361
Event Subsystem Variables ......................................................... 363
Index..................................................... 365

10 Contents

11
Welcome
This guide provides information about the general architecture of the

Wonderware Historian and describes the different subsystems and
components that make up the product. This guide can be used as a
reference guide for all conceptual information about the Wonderware
Historian components.
Wonderware Historian Documentation Set

The Wonderware® Historian documentation set includes the following
guides:
• Wonderware System Platform Installation Guide
(WSP_Install_Guide.pdf). This guide provides information on
installing the Wonderware Historian, including hardware and
software requirements and migration instructions.
• Wonderware Historian Concepts Guide (HistorianConcepts.pdf).

This guide provides an overview of the entire Wonderware
Historian system and describes each of the subsystems in detail.
• Wonderware Historian Administration Guide

(HistorianAdmin.pdf). This guide describes how to administer and
maintain an installed Wonderware Historian, such as configuring
data acquisition and storage, managing security, and monitoring
the system.
• Wonderware Historian Database Reference

(HistorianDatabase.pdf). This guide provides documentation for all
of the Wonderware Historian database entities, such as tables,
views, and stored procedures.

12  Welcome
• Wonderware Historian Glossary (HistorianGlossary.pdf). This

guide provides definitions for terms used throughout the
documentation set.
In addition, the Invensys License Manager Guide (License.pdf)

describes the Invensys License Manager and how to use it to install,
maintain, and delete licenses and license servers on local and remote
computers.
A PDF file for each of these guides is available on the Wonderware
Historian installation CD. You can easily print information from the
PDF files. The Wonderware Historian documentation is also provided
as an online help file, which can be accessed from the System
Management Console management tool.
Documentation Conventions
This documentation uses the following conventions:
Convention Used for
Initial Capitals Paths and file names.

Bold Menus, commands, dialog box names, and
dialog box options.
Monospace Code samples and display text.

Welcome13
Technical Support
Wonderware Technical Support offers a variety of support options to
answer any questions on Wonderware products and their
implementation.
Before you contact Technical Support, refer to the relevant section(s)
in this documentation for a possible solution to the problem. If you
need to contact technical support for help, have the following
information ready:
• The type and version of the operating system you are using.
• Details of how to recreate the problem.
• The exact wording of the error messages you saw.
• Any relevant output listing from the Log Viewer or any other
diagnostic applications.
• Details of what you did to try to solve the problem(s) and your
results.
• If known, the Wonderware Technical Support case number
assigned to your problem, if this is an ongoing problem.

14  Welcome

15
Chapter 1
Introduction
The Wonderware Historian, formerly known as IndustrialSQL

Server™, bridges the gap between a real-time high-volume plant
monitoring environment and an open, flexible business information
environment. The historian:
• Acquires plant data from high-speed Wonderware I/O Servers,
DAServers, InTouch HMI software, Application Server, and other
devices.
• Compresses and stores data.
• Responds to SQL requests for plant data.
The historian also contains event, alarm, summary, configuration,
security, backup, and system monitoring information.
The historian is tightly coupled to Microsoft SQL Server.

16  Chapter 1 Introduction
The Wonderware Historian Solution

The Wonderware Historian is a high-capacity, high-throughput server
that stores time-series data. The historian acquires and stores process
data at full resolution or at a specified resolution and provides
real-time and historical data together with configuration, event,
summary, and associated production data to client applications on the
desktop. The historian combines the power and flexibility of Microsoft
SQL Server with high speed acquisition and efficient data
compression.
The Wonderware Historian is designed to:
• Handle the volume of time-series data.
• Handle the rapid storage rate of time-series data.
• Extend SQL to effectively handle time-series data.
Process Data
Process data is any relevant information to successfully run a process.
The following information is considered to be process data:
• Real-time data - What is the current value of this tag?
• Historical data - What was the value of this tag every second last
Monday?
• Summary data - What is the average of each of these five tags?
• Event data - When did that boiler trip?
• Configuration data - How many I/O Servers am I using and what
are their types?
To improve performance and quality while reducing cost, process data
must be available for analysis. Process data is typically analyzed to
determine:
• Process analysis, diagnostics, and optimization.
• Predictive and preventive equipment maintenance.
• Product and process quality (SPC/SQC).
• Health and safety; environmental impact (EPA/FDA).
• Production reporting.
• Failure analysis.

The Wonderware Historian Solution17
Wonderware Historian Benefits

Wonderware Historian extends the functionality of Microsoft SQL
Server, providing more than an order of magnitude increase in
acquisition speeds, a corresponding reduction in storage volume, and
elegant extensions to structured query language (SQL) to query time
series data.
• High-speed data capture
The comprehensive range of Wonderware I/O Servers and
DAServers are used to connect to over 500 control and data
acquisition devices.
Designed for optimal acquisition and storage of analog, discrete,
and string data, the Wonderware Historian outperforms all normal
relational databases on similar hardware by a wide margin,
making the storage of high-speed data in a relational database
possible.
Wonderware I/O Servers and DAServers support the SuiteLink™
protocol. SuiteLink allows for time and quality stamping at the I/O
Server and further improves the rate of data acquisition.
• Reduced storage space
The Wonderware Historian stores data in a fraction of the space
required by a normal relational database. The actual disk space
required to store plant data depends on the size and nature of the
plant and the length of the plant history required.
• Time domain extensions to SQL
The SQL language does not support time series data. In particular,
there is no way to control the resolution of returned data in SQL.
An example of resolution would be an evenly spaced sampling of
data over a period of time.
Microsoft SQL Server supports its own extensions to the SQL
language, called Transact-SQL. The Wonderware Historian
further extends Transact-SQL, allowing control of resolution and
providing the basis for time-related functions such as rate of
change and process calculations on the server.

Integration with Microsoft SQL Server

A large amount of time-series data has the same characteristics as
normal business data. For example, configuration data is relatively
static and does not change at a real-time rate. Over the life of a site,
tags are added and deleted, descriptions are changed, and engineering
ranges are altered. A Microsoft SQL Server database, called the
Runtime database, stores this type of information.
The Runtime database is the SQL Server online database for the
entire Wonderware Historian. The Runtime database is shipped with
a set of standard database entities, such as tables, views, and stored
procedures to store configuration data for a typical factory. You can
use the Configuration Editor within the System Management Console
to easily add configuration data to the Runtime database that reflects
your factory environment.
Microsoft SQL Server Object Linking and Embedding for Databases
(OLE DB) is used to access the time-series data that the historian
stores outside of the SQL Server database. You can query the
Microsoft SQL Server for both configuration information in the
Runtime database and historical data on disk, and the integration
appears seamless.
Because the historian is tightly coupled to and effectively extends a
Microsoft SQL Server, it can leverage all of the features that Microsoft
SQL Server has to offer, such as database security, replication, and
backups.
Support for SQL Clients

The client/server architecture of Wonderware Historian supports
client applications on the desktop, while ensuring the integrity and
security of data on the server. This client/server architecture provides
common access to time-series data and associated configuration, event,
and business data. The computing power of both the client and the
server is exploited by optimizing processor intensive operations on the
server and minimizing data to be transmitted on the network to
improve system performance.
The gateway for accessing any type of information in the historian is
the Microsoft SQL Server. Thus, any client application that can
connect to Microsoft SQL Server can also connect to the historian.
Two categories of client applications can be used to access and retrieve
information from the historian:

Wonderware Historian Subsystems19
• Clients developed specifically to access data from historian.

Wonderware provides a number of client tools to address specific
data representation and analysis requirements. Third-party query
tools that specifically support Wonderware Historian are also
available. These client tools remove the requirement for users to be
familiar with SQL and provide intuitive point-and-click interfaces
to access, analyze, and graph both current and historically
acquired time-series data.
• Any third-party query tool that can access SQL or ODBC data
sources. Numerous commercial query and reporting tools are
available that provide rich, user-friendly interfaces to SQL-based
data. All client tools with an interface to Microsoft SQL Server or
ODBC are suitable for historian data access and reporting.
Wonderware Historian Subsystems

The Wonderware Historian is made up of specialized subsystems,
which work together to manage data as it is acquired or generated,
stored, and retrieved, as follows.
• Configuration Subsystem
• Data Acquisition Subsystem
• Data Storage Subsystem
• Data Retrieval Subsystem
• Event Subsystem
• Replication Subsystem


21
Chapter 2
System-Level Concepts and

Functionality
Some concepts and functionality apply across the entire Wonderware

Historian, such as tags, time handling, system parameters, security,
and data quality.
About Tags
The Wonderware Historian stores and retrieves values, where every
value gets stored with some timestamp and associated quality. This
triplet of value, timestamp, and quality is called VTQ and constitutes
the smallest addressable piece of data in the Wonderware Historian
data model. Sometimes a VTQ can be referred to as a value or data
point, but even in that case it is assumed that it means the whole
triplet.
A tag is an entity that identifies a set of VTQs received from the same
data source. Usually it represents a single attribute of some physical
process or device. In Wonderware Historian, a tag is identified by a
unique tag name.

22  Chapter 2 System-Level Concepts and Functionality
Types of Tags
The following table describes the types of tags in the system.
Tag Type Description
Analog An analog value is a variable that measures a

continuous physical quantity. For example, the
temperature of a boiler would be measured as an
analog value.
Discrete A discrete value is a variable that only has two states:
'1' (True, On) or '0' (False, Off).
String A string value is a text expression treated as a single
data item. A string does not require a special format or
syntax.
Event An event tag is a name for an event definition in the
system. For example, if you wanted to detect when the
temperature of tank reached 100 degrees, you might
define an event tag and name it "TankAt100."
Analog An analog summary tag contains summarized data
Summary (minimum, maximum, average, and so on) that is
configured to be replicated from one historian to
another.
State An state summary tag contains summarized data
Summary (minimum time in state, maximum time in state,
average time in state, and so on) that is configured to
be replicated from one historian to another.
Naming Conventions for Tagnames

Tagnames may contain letters, digits, and special characters, where:
• Letter = any letter as defined by the Unicode Standard. The
Unicode definition of letters includes Latin characters from a
through z and from A through Z, in addition to letter characters
from other languages.
• Digit = any numerical character
• Special character = any graphics character except the following:
characters whose ASCII table code is 0 through 32 -(non-graphic
characters) and . + - * / \ = ( ) ` ~ ! ^ & @ [ ] { } | : ; ’ , < > ? “ space
It is highly recommended that you adhere to the rules for SQL Server
identifiers as well.

About Tags23
For "conventional" tagnames the first character may be a:

• Letter
• Digit
• Dollar sign ( $ ) or pound sign ( # )
Subsequent characters may be:
• A digit, but then the tagname must contain at least one letter
• Any of the supported special characters.
Due to storage formatting requirements, you cannot use either a
quotation mark ( " ) or one or more single quotation marks ( ' ) at the
beginning or at the end of a tag name.
Tag names that do not comply with these rules are regarded as
"unconventional."
In a SQL query against a wide table, unconventional tag names must
be delimited with brackets ( [ ] ), because the tag name is used as a
column name. For more information, see "Using an Unconventional
Tagname in a Wide Table Query" on page 283.
VTQ for Tags

Sources for tag values are as follows:
• Data from Wonderware Application Server
• Automatically acquired I/O Server data, either in real-time or later
• Replicated data from one or more tier-1 Wonderware Historians
• I/O Server data arriving late from an IDAS store-and-forward
cache
• Data generated internally for system monitoring tags
• Data inserted or updated with a Transact-SQL statement
• Data contained in a properly formatted .CSV file, which you import
• Data sent from client applications developed with the Wonderware
Historian Software Development Toolkit (SDK)
The VTQ timestamps are transmitted and stored in the Coordinated
Universal Time (UTC) format. During data retrieval they are returned
converted by default to the local time on the Historian server.

Data quality is the degree of validity for a data value. Data quality can
range from good, in which case the value is exactly what was originally
acquired from the plant floor, to invalid, in which the value is either
wrong or cannot be verified. Essentially it is a piece of metadata
present in every VTQ and provides information about how much you
can rely on the value and the timestamp of that VTQ.
Wonderware Historian VTQ's quality element consists of two 16-bit
unsigned integers called OPCQuality and QualityDetail.
OPCQuality is the actual quality information provided by the data
source, such as an I/O Server or DAServer. This is the primary
indicator of the value's quality. Wonderware Historian keeps that
quality information intact in its exact original state since the moment
it was received.
QualityDetail is an additional proprietary piece of information created
by the Wonderware Historian and attached to every VTQ it processes.
QualityDetail is used to record information about what happened with
that VTQ while it was processed by various historian subsystems. For
example, it can indicate that some VTQ was stored as the first value
after the connection to the historian server was restored after a
temporary disconnect, or it can tell that the retrieved value is not the
actual value generated by an I/O Server or DAServer, but a manually
corrected one.
When executing a SQL query, there is a special indicator returned
through the Quality column at retrieval time that provides a summary
for the related data item. The value of that indicator is derived from
the available OPCQuality and QualityDetail. For a listing of these
values, see "Client-Side Quality Values" on page 66.
For additional information regarding data quality, see "Data Quality"
on page 61.
Tag Properties
Every Wonderware Historian tag is associated with one or more tag
metadata instances. A tag metadata instance is a set of properties
identified by a unique TagId. The TagId is a 16-byte globally unique
identifier (GUID). Tag metadata properties include tag name,
description, tag type, storage type, creation time, and so on.

About Tags25
Tag metadata properties describe what the tag is, where the data for
that tag is coming from, how the VTQs of that tag should be stored,
and how they should be retrieved and displayed by client applications.
The following table describes the most important tag metadata
properties:
Property Description
TagName A Unicode string (UTF-16) of up to 256

characters.
TagId A 16-byte GUID.
DateCreated The UTC timestamp when the tag metadata
instance was created.
CreatedBy The name of the user or application created
the tag metadata instance.
TagType The type of tag: analog, discrete, string, event,
or summary. For more information, see "Types
of Tags" on page 22.
AcquisitionType The method by which the tag's values are
acquired.
StorageType The method of storing the tag's values.
AITag Indicates whether the tag's values are stored
by the classic storage subsystem. For more
information on classic storage, see Chapter 6,
"Classic Data Storage Subsystem."
Depending on values of these properties, additional properties provide

more specific details. For example if a tag is an analog tag—that is, it
represents a variable measuring a continuous physical quantity such
as the temperature of a boiler—the tag metadata property RawType
specifies what kind of numeric type is used, either float or integer. If it
is an integer, the IntegerSize property specifies the number of bits in
that integer, and so on.
Note: The ability to set up an alternate file storage location for tag
metadata is possible to ensure it is available should the primary
location become corrupt or not accessible.
For the full list of tag metadata properties, see the tag-related tables
in Chapter 2, "Tables."

Tag Configuration Versioning

Every time a tag metadata property changes, a new tag metadata
instance is created with a unique TagId. This mechanism allows
Wonderware Historian to preserve the tag configuration history when
different types of data is stored for the same tag over time.
Several different tag metadata instances can share the same tag
name, resulting in several versions of the same tag. The most recently
created tag metadata instance is referred to as the current version of
that tag.
To illustrate this, consider the following example. You create a tag
named MyTag to store 16-bit integer values received from some device.
The tag metadata instance of some TagId, say
883DDAE3-E3F5-441C-A5FD-38AD97DEC070, gets created and then
some data values get stored. Several months later the device gets
upgraded to generate 32-bit integer values so you reconfigure the
MyTag tag to store 32-bit integers. During the tag reconfiguration a
new tag metadata instance of another TagId, say
FFA0E74C-12FD-49A6-8EBA-B30AFAEF55DA, is created and newly
stored 32-bit values will be associated with that TagId. Now although
the current version of the MyTag references it as a 32-bit integer, the
older 16-bit values are still accessible because they are associated with
the older tag metadata instance preserved in the history.
Security
The Wonderware Historian uses two security mechanisms:
• Windows operating system security
• Microsoft SQL Server security
For clients to access the historian, they must pass through both of
these security levels. The historian Management Console (within the
System Management Console) in particular adds an additional level of
security checking to restrict access to functions affecting the state of
the historian to only authorized users. Also, some of the historian
components require Windows and SQL Server logins.
For more information on configuring user rights assignments for local
security policies, see the Microsoft documentation.
For information on how to manage security, see Chapter 8, "Managing
Security," in the Wonderware Historian Administration Guide.

Security27
Windows Operating System Security

To log on to Wonderware Historian as a client, the first thing a user
must be able to do is to log on to the operating system on their
computer. For the Windows operating system, a valid user account,
consisting of a login ID (username) and password, is required to log on
to the computer. This Windows user account can also be used to gain
access to network resources in the domain.
SQL Server also requires authentication in order for clients to connect
to it. You can use either Windows authentication or SQL Server
authentication. For more information, see "SQL Server Security" on
page 28.
Default Windows Login for Wonderware Historian

Services
All of the modules in the Wonderware Historian, except for the
Management Console and the Configuration Editor, run as Windows
services, and therefore require a valid Windows user account (the
ArchestrA administrative user account) to ensure proper operation.
This ArchestrA user account is specified during installation.
The ArchestrA account must be a member of the local administrators
group on the server hosting the historian, as well as on all computers
hosting a remote IDAS.
You can change the ArchestrA user account by using the ArchestrA
Change Network Account Utility.
WARNING! Changing the ArchestrA user account affects all

ArchestrA components that run as services, not just historian
services. If you change this account, you must restart the
computer.
Do not configure historian services (including remote IDASs) to run

under a specific user account. All services should be configured by
default to log on to the Windows operating system using the
LocalSystem account. The historian services will impersonate the
ArchestrA user account, and this impersonation will fail if the service
is not run using the LocalSystem account.

SQL Server Security

Because the Wonderware Historian works closely with Microsoft SQL
Server, it uses and takes advantage of the security features that
Microsoft SQL Server has to offer. The purpose of security for a SQL
Server is to control who can access the server, access specific
databases within a server, and perform certain actions within a
database.
A database user must pass through two stages of security for the
historian:
• Authentication, which validates the user’s identity to the server
itself.
• Database authorization, which controls the database(s) that user
can access, as well as the types of actions that the user can perform
on objects within the database.
User authentication and database authorization are managed from
Microsoft SQL Server Management Studio.
To access information in the Wonderware Historian databases, users
need to be granted access to the databases. The historian is shipped
with pre-configured database roles and user accounts to serve as a
starting point for your security model. Database roles, SQL Server
login IDs, and user accounts are managed using the Microsoft SQL
Server Management Studio.
Authentication
Microsoft SQL Server authenticates users with individual login
account and password combinations. After the user’s identity is
authenticated, if authentication is successful, the user is allowed to
connect to a SQL Server instance. There are two types of
authentication:
• Windows authentication, in which users must connect to the SQL
Server using a Windows user account (a Windows login ID and
password that is provided during the Windows login session).
• SQL Server authentication, in which users must connect to the
SQL Server using SQL Server login account (a SQL Server login
ID and password).

SQL Server can operate in one of two security modes, which control
the type of accounts that must be used for access to the server:
• Windows authentication mode. In this mode, the SQL Server only
uses Windows authentication.
• Mixed mode. In this mode, the SQL Server uses both Windows
authentication and SQL Server authentication. If the login name
matches the Windows network username, then validation is
handled by the Windows security mechanisms. If the user login
name does not match the Windows network username, then
Microsoft SQL Server security mechanisms are used.
SQL Server authentication is provided for backward compatibility
only. Microsoft recommends that you use Windows authentication,
when possible.
Note: A new system parameter now allows Private Namespace to

work with domain logins. Enabled, every domain user has their own
Private Namespace and Annotations entries.
For more information about authentication, see your Microsoft SQL

Server documentation.
Default Windows Security Groups

The following Windows security groups are created by default on the
Wonderware Historian computer. Use these groups to assign different
levels of database permissions to users.
• aaAdministrators
• aaPowerUsers
• aaUsers
• aaReplicationUsers
Each group is automatically configured to be a member of the SQL
Server database role with the same name. For example, the
aaAdministrators Windows security group is a member of the default
aaAdministrators SQL Server database role. If you add Windows users
to the aaAdministrators security group, they will automatically be
given permissions of the aaAdministrators SQL Server database role.

Wonderware Historian Default Logins

When the Wonderware Historian is installed, default SQL Server
logins are created that you can use for logging on to the historian from
client applications. These default logins provide "out of the box"
functionality in that you do not have to create logins to start using the
system. The following table describes the pre-configured logins:
Login
Name Password Description
aaAdmin pwAdmin A user who can access and modify all data
and create objects. Cannot drop the
database or truncate tables.
aaPower pwPower A user with full read access and the
ability to create objects and modify the
contents of the non-core tables.
aaUser pwUser A read-only user who can access all data,
but cannot modify data or consume
database resources.
aadbo pwddbo Database owner. Full permissions.
The default database for each of these logins is the historian Runtime
database. This default security model is provided as a starting point
for system security and is suitable for many types of installations.
These logins are valid if the Microsoft SQL Server is set to mixed mode
security. If only Windows authentication is used, you must configure
the access rights for each user.
Important: Never use blank passwords for logins.
The following logins are provided for backward compatibility only.

They will be deprecated in a future release. Do not use these logins.
Login Name Password Description
wwUser wwUser Same as aaUser.

wwPower wwPower Same as aaPower.
wwAdmin wwAdmin Same as aaAdmin.
wwdbo pwddbo Same as aadbo.

Database Authorization
After a user successfully connects to the Microsoft SQL Server, the
user needs authority to access databases on the server. This is
accomplished by user accounts for each database. A database user
consists of a user name and a login ID. Each database user must be
mapped to an existing login ID.
User names are stored in the sysusers table in each database. When a
user tries to access a database, the Microsoft SQL Server looks for an
entry in the sysusers table and then tries to find a match in the
syslogins table in the master database. If the Microsoft SQL Server
cannot resolve the username, database access is denied.
The types of actions the user can perform in the database are based on
authority information defined in the user account. The authority to
perform a certain action is called a permission. There are two types of
permissions: object permissions and statement permissions.
Permission Description
Object Regulates the actions that a user can perform on

certain database objects that already exist in the
database. Database objects include things such as
tables, indexes, views, defaults, triggers, rules, and
procedures. Object permissions are granted and
revoked by the owner (creator) of the object.
Statement Controls who can issue particular Transact-SQL
statements. Database statements include
commands such as SELECT, INSERT, or DELETE.
Statement permissions, also called command
permissions, can only be granted and revoked by
the system administrator or the database owner.
Users can be grouped into roles, which is a single unit against which
you can apply permissions. Permissions granted to, denied to, or
revoked from a role also apply to any members of the role.

Wonderware Historian Default Users and Roles

The Wonderware Historian is shipped with a number of pre-configured
user accounts and roles.
The following table describes the default SQL Server usernames, the
login IDs and database roles to which they belong, and the actions that
they are allowed to perform in the Runtime database. You can add
additional users and roles using SQL Server Enterprise Manager.
Username
in
Login ID Database Member of Role Permissions
aaUser aaUser aaUsers SELECT on all tables

INSERT, UPDATE, DELETE on
PrivateNameSpace and Annotation
aaPower aaPower aaPowerUsers CREATE Table
CREATE View
CREATE Stored procedure
CREATE Default
CREATE Rule
SELECT on all tables
INSERT, UPDATE, DELETE on grouping
tables
aaAdmin aaAdmin aaAdministrators CREATE Table
CREATE View
CREATE Stored procedure
CREATE Default
CREATE Rule
DUMP Database
DUMP Transaction
SELECT, INSERT, UPDATE, DELETE on
all tables
aadbo dbo db_owner Full database owner capabilities

The following users and roles are provided for backward compatibility
only. They will be deprecated in a future release. Do not use these
users and roles.
Username
in
Login ID Database Member of Role Permissions
wwUser wwUser wwUsers Same as for aaUser.

wwPower wwPower wwPowerUsers Same as for aaPower.
wwAdmin wwAdmin wwAdministrators Same as for aaAdmin.
wwdbo wwdbo db_owner Same as for aadbo.
Each default role contains the corresponding SQL Server user account,
as well as the corresponding default Windows security group. For more
information on the default Windows security groups, see "Default
Windows Security Groups" on page 29.
Default SQL Server Login for Wonderware Historian

Services
Some components of the Wonderware Historian require a SQL Server
login ID to access the master, Runtime, and Holding databases. By
default, the historian uses the ArchestrA user account to log on to the
Microsoft SQL Server, using Windows authentication.
For Microsoft SQL Server, if the Windows user account is an
administrative account on the local computer, it will map the account
to the sysadmin fixed server role. (This user will be granted the same
permissions as the sa SQL Server user account.) Because the
ArchestrA user account is always a local administrative account, it
will always have administrative permissions (sysadmin) within the
SQL Server.
Management Console Security

The Wonderware Historian Management Console (which is part of the
overall System Management Console) runs in the context of the logged
on Windows user account. To protect against unauthorized access to
the Wonderware Historian, you must specify a separate Windows user
account that the Management Console will use to connect to the
historian. You can specify this account when you set up the server
registration properties. For more information on registration, see
Chapter 1, "Getting Started with Administrative Tools," in the
Wonderware Historian Administration Guide.

If the account specified is not a member of the local administrators

group on the computer hosting the historian, the Management Console
has "read-only" access. That is, you may view all the information
shown in the Management Console, but you cannot perform any
control actions on the historian, such as starting or stopping the
system, creating new history blocks, and so on.
Important: To prevent possible unauthorized access, the password

for the Management Console login account must NOT be blank.
Default Access Rights for Different Operating

Systems
By default, in the Windows XP operating system, all users are given
the right to shut down the local computer. Therefore, for these
operating systems, all users are automatically given "level 2" access to
the historian. A "level 2" user in the Wonderware historian can shut
down the local computer. To prevent users from being able to start or
stop the historian on these operating systems, you must take away the
"Shut Down the System" right in the Windows local security policy.
For the Windows Server 2003 operating system, "power users" are
given the right to shut down the local computer. Power users are
members of "Shut Down the System" right in the Windows local
security policy. Therefore, for this operating system, power users are
automatically given "level 2" access to the historian.
For the Windows Server 2008, Windows Vista, and Windows 7
operating systems, "power users" are not given the right to shut down
the local computer. Power users are not members of "Shut Down the
System" right in the Windows local security policy. Therefore, for these
operating systems, power users are automatically given "level 1"
access to the historian.

Time Handling35
Time Handling
Timestamps for all data are stored in Coordinated Universal Time
(UTC). The current UTC time is derived from the current local time
and the time zone setting in the operating system of the computer on
which the Wonderware Historian is running. During data retrieval,
timestamps are returned in local time, by default. You can convert the
timestamps so that they are shown in local time by using a special
query parameter.
You should use the international date/time format for all timestamps
used in queries. The format is:
YYYYMMDD HH:MM:SS.000
where,
YYYY = year
MM = month
DD = day
HH = hour
MM = minutes
SS = seconds
000 = milliseconds
The format for timestamps returned from queries is controlled by the
default language settings of the SQL Server login. Make sure that you
configure the default language setting for SQL Server logins correctly,
especially in environments with mixed languages/regional settings.
If you have multiple historians and/or are using remote data sources,
it is very important that you synchronize the time between the
computers. For more information, see "Time Synchronization for Data
Acquisition" on page 91.
Make sure that you have selected the operating system setting to
automatically adjust time for daylight savings, if the time zone in
which the computer is located observes daylight savings time.

System Parameters
A system parameter is a parameter that controls some aspect of the
overall Wonderware Historian behavior. The following table describes
the default system parameters:
Name Description
AIAutoResize Controls the automatic resizing of the active image. 1

= Automatic resizing enabled; 0 = Automatic resizing
disabled. The default is 1. For more information, see
"About the Active Image for Classic Storage" on
page 130.
AIResizeInterval Interval, in minutes, that the system will resize the
active image, if the AIAutoResize parameter is
enabled.
AllowOriginals Used to allow the insertion of original data for I/O
Server tags. You must set this parameter to 1 before
importing .lgh original data. For more information,
see "Data Acquisition by Means of INSERT and
UPDATE Statements" on page 94.
AnalogSummaryTypeAbbreviation Abbreviation used when generating analog summary
tagnames. For more information, see "Specifying
Naming Schemes for Replication" in Chapter 7,
"Managing and Configuring Replication," in the
AutoStart Used to start the historian automatically when the
computer on which the Wonderware Historian is
running is started: 1 = Autostart enabled; 0 =
Autostart disabled. For more information, see
"Configuring Wonderware Historian to AutoStart" in
Chapter 1, "Getting Started with Administrative
Tools," in the Wonderware Historian Administration
Guide. If you change this parameter, you must
commit the changes to the historian system.
Note: You cannot change the Autostart system

parameter using the SMC if SQL Server is not installed
to run as an administrator. The SQL Server instance is
used to change the service settings on behalf of the
SMC.
ConfigEditorVersion (Not editable.) The minimum version number of the

Configuration Editor that can edit the Runtime
database. Used internally by the system.
DatabaseVersion (Not editable.) Current version number of the
database.

System Parameters37
Name Description
DataImportPath Path to the CSV file for an import of external data.

For more information, see "Importing Data from a
CSV File" on page 99. If you change this parameter, a
restart of the system is required.
DataIndexPath File path for storing indexing information for data.
EventStorageDuration Maximum duration, in hours, that event records are
stored in the EventHistory table.
HeadroomXXXX Number of tags for which to pre-allocate memory in
the system. For more information, see "Pre-allocating
Memory for Future Tags" in Chapter 2, "Configuring
Tags," in the Wonderware Historian Administration
Guide.
HistorianPartner The computer name of a partner historian. You can
use either the host name, fully qualified name, or an
IP address. Leading backslashes are optional. The
name/IP used must be one that can be correctly
resolved by all of the AppEngine and Historian Client
computers that will connect to the partner. For more
information, see "Using a Redundant Historian" in
Chapter 9, "Viewing or Changing System-Wide
Properties," in the Wonderware Historian
Administration Guide.
HistorianVersion (Not editable.) Current version number and build
number of the Wonderware Historian. The value for
this parameter is automatically supplied when the
system starts.
HistoryCacheSize Allocation of system memory, in MB, for tag
information. The default is 0. For more information,
see "Memory Management for Classic Storage" on
page 133. If you change this parameter, you must
commit the change and then rescan the history blocks
to flush the cache.
HistoryDaysAlwaysCached The duration, in days, for which history block
information is always loaded in memory. The default
is 0.
HoursPerBlock Duration, in hours, for history blocks. Valid values
are: 1, 2, 3, 4, 6, 8, 12, 24. The default is 24 hours. The
history block size must always be greater than the
highest scan rate. For more information, see "History
Blocks" on page 114.

Name Description
InterpolationTypeInteger The type of interpolation for data values of type

integer. 0=Stair-step; 1=Linear. The default is 0. For
more information on interpolation, see "Interpolation
Type (wwInterpolationType)" on page 241.
InterpolationTypeReal The type of interpolation for data values of type real.
0=Stair-step; 1=Linear. The default is 1.
LateDataPathThreshold Controls the store-and-forward threshold for late
data.
LicenseRemoteIDASCount (Not editable.) The number of allowed remote IDASs
for the historian. This value is determined from the
license file. Used internally by the system.
LicenseTagCount (Not editable.) The number of allowed tags for the
historian. This value is determined from the license
file. Used internally by the system.
ManualDataPathThreshold Controls the store-and-forward threshold for manual
data.
ModLogTrackingStatus Turns modification tracking on or off. The value you
specify will determine what modifications are
tracked. For more information, see "Turning
Modification Tracking On/Off" in Chapter 9, "Viewing
or Changing System-Wide Properties," in the
OldDataSynchronizationDelay Time delay, in seconds, between when changes for
“old” data (inserts, updates, and store-and-forward
data) must be sent from the tier-1 historian to the
tier-2 historian.
QualityRule Indicates whether the system should use values
having a quality of Good and Uncertain, or having
only a quality of Good. 0 = Good and Uncertain; 1 =
Good only. The default is 0. For more information on
the quality rule, see "Quality Rule (wwQualityRule)"
on page 247.
RealTimeWindow The maximum delay, in seconds, for which data is
considered real-time data for swinging door storage.
The delay is relative to the current time. Valid values
are between 30 and 300 milliseconds. The default is
60. For more information, see "About the Real-Time
Data Window for Classic Storage" on page 127.
ReplicationConcurrentOperations Limits the total number of retrieval client objects
performing calculations in a retrieval based
calculations for a time cycle.

System Parameters39
Name Description
ReplicationDefaultPrefix The default prefix for replication tags on the tier-2

historian. If you change ReplicationDefaultPrefix
system parameter, all replication tags that use the old
prefix are not updated to use the newer prefix. For
more information, see Specifying Naming Schemes for
Replication in Chapter 7, Managing and Configuring
Replication, in the Wonderware Historian
ReplicationTCPPort The TCP port number the Historian Client Access
Port (HCAP) service listens on for any
incoming-connection requests from an HCAL-enabled
client. It must match the port number the
HCAL-enabled client is using for communication with
the historian. When modifying this system parameter
on an historian node, you must also modify the port
number in the Windows Firewall exception list for the
historian replication service or another
HCAL-enabled client to the same value. This port
number must unique on the historian node; that is, no
other applications on the historian node should be
listening on this port number.
RevisionLogPath The file path to the write-ahead log for revision data
transactions.
SimpleReplicationNamingScheme The default naming scheme used for configuring
simple replication tags. For more information, see
"Specifying Naming Schemes for Replication" in
Chapter 7, "Managing and Configuring Replication,"
in the Wonderware Historian Administration Guide.
StateSummaryTypeAbbreviation Abbreviation used when generating state summary
tagnames. For more information, see "Specifying
Naming Schemes for Replication" in Chapter 7,
"Managing and Configuring Replication," in the
SuiteLinkTimeSyncInterval Frequency, in minutes, that IDASs will attempt to
synchronize the timestamping mechanism for
associated I/O Servers. If this parameter is set to 0, no
time synchronization will occur. For more
information, see "Time Synchronization for Data
Acquisition" on page 91.

Name Description
SummaryCalculationTimeout The maximum expected delay, in minutes, for

calculating summary data for replicated tags. Setting
this parameter too high will delay associated
summary calculations unnecessarily. Setting it too
low will cause the system to prematurely calculate
summaries and then later require additional
processing to correct those calculations.
SummaryReplicationNamingScheme The default naming scheme used for configuring
summary replication tags. For more information, see
"Specifying Naming Schemes for Replication" in
Chapter 7, "Managing and Configuring Replication,"
in the Wonderware Historian Administration Guide.
SummaryStorageDuration Maximum duration, in hours, that summary records
will be stored in the legacy SummaryHistory table.
SysPerfTags Used to turn on performance monitoring tags for the
Wonderware Historian system. 0 = Off; 1 = On. The
default is 1. For more information, see "Performance
Monitoring Tags" on page 51.
TimeStampRule Used to determine which timestamp within a
retrieval cycle to use for a data value. 0 = Use the
timestamp at the start of the cycle; 1 = Use the
timestamp at the end of the cycle. The default is 1.
For more information, see "Timestamp Rule
(wwTimestampRule)" on page 244.
TimeSyncIODrivers If enabled, the Wonderware Historian will send time
synchronization commands to all associated remote
IDASs. For more information, see "Time
Synchronization for Data Acquisition" on page 91.
TimeSyncMaster Name of the computer that the Wonderware
Historian will use as a time synchronization source.

System Messages41
System Messages
System messages include error messages and informational messages
about the state of the Wonderware Historian as a whole or for any of
the internal subsystems and individual processes. System messages
are logged to the:
• ArchestrA Logger.
• Windows event log, which can be viewed with the Windows Event
Viewer. Not all messages are logged to the Windows event log. In
general, only user actions and exceptional events are written to
this log. The messages are logged with the "Historian" or the name
of the Wonderware Historian service as the source.
System messages are divided into the following categories:
Category Description
FATAL The process cannot continue. An error of this

severity results in a system shutdown.
CRITICAL These types of errors will cause malfunctions in the
data storage or retrieval systems, such as data loss
or corruption.
ERROR General errors. For example, address validation
errors during system startup. These errors may
result in an orderly shutdown of the system, but will
not preclude system operation in most cases.
WARNING Messages that simply notify the operator of
parameter settings or events that have occurred.
For example, failure to link a dynamically-linked
procedure entry point for a non-obligatory function
will be logged as a warning.
INFO Messages relating to startup progress or the
commencement of active data storage.
DEBUG Debugging messages, which will not typically
appear in released versions of the system.

Wonderware Historian messages are logged to the Log Viewer as

follows:
• Critical, fatal, and error messages are logged as "Error" messages.
The appropriate indicator, either "CRITICAL," FATAL," or
"ERROR," will be prefixed to message.
• Warnings will be logged as "Warning" message, with no prefix.
• Informational messages will be logged as "Info" messages, with no
prefix.
• Debug messages will be logged as "Trace" messages, with no prefix.
For information on monitoring the system, see Chapter 10,
"Monitoring the System," in the Wonderware Historian
Wonderware Historian Processes

The following table describes the Wonderware Historian processes:
Name Executable Name Description
Wonderware Historian aahClientAccessPoint.exe Manages communication from

Client Access Point clients to the Historian Server.
(aahClientAccessPoint)
Wonderware Historian aahCfgSvc.exe Handles all configuration requests,
Configuration as well as hosts the interfaces for
(InSQLConfiguration) manual data input and retrieval.
For more information, see Chapter
3, "Configuration Subsystem."
Wonderware Historian aahStorage.exe Storage process. Accepts incoming
Storage Process plant data and stores it to disk.
Chapter 5, "Data Storage
Subsystem."
Wonderware Historian aahStoreSvc.exe Classic storage process. Accepts
Storage (InSQLStorage) incoming plant data and stores it to
disk. For more information, see
Chapter 6, "Classic Data Storage
Subsystem."
Wonderware Historian aahManStSvc.exe Classic manual storage process.
Manual Storage Accepts incoming non-realtime
(InSQLManualStorage) plant data and stores it to disk. For
more information, see Chapter 6,
"Classic Data Storage Subsystem."

Wonderware Historian Processes43
Name Executable Name Description
Wonderware Historian aahIDASSvc.exe Acquires data from local or remote

Data Acquisition IDASs and forwards it on to the
(InSQLDataAcquisition) storage subsystem. For more
information, see Chapter 4, "Data
Acquisition Subsystem."
Wonderware Historian aahEventSvc.exe Searches through history data and
Event System determines if specific events have
(InSQLEventSystem) occurred. For more information, see
Chapter 11, "Event Subsystem."
Wonderware Historian aahIndexSvc.exe Manages the internal indexing of
Indexing (InSQLIndexing) history data on disk.
Wonderware Historian aahIOSvrSvc.exe Provides realtime data values from
IOServer the historian to network clients. For
(InSQLIOServer) more information, see Chapter 7,
"Data Retrieval Subsystem."
Wonderware Historian aahReplicationSvc.exe Performs data transformations on a
Replication tier-1 historian and sends the
(HistorianReplication) results to one or more tier-2
historians. For more information,
see Chapter 10, "Replication
Subsystem."
Wonderware Historian aahRetSvc.exe Retrieves data from storage. For
Retrieval more information, see Chapter 7,
(InSQLRetrieval) "Data Retrieval Subsystem."
Wonderware Historian aahSCM.exe Provides status information
SCM (InSQLSCM) regarding the historian. Used
internally by the historian. This
service runs continuously, even if
the historian is stopped.
WonderwareHistorian aahDrvSvc.exe Generates data values for various
System Driver system monitoring tags. For more
(InSQLSystemDriver) information, see "The System Driver
and System Tags" on page 44.
Wonderware Historian aahMetadataServer.exe Maintains the tag metadata cache
Metadata Server for tags stored by the Storage
process.
For more information on Windows services, see your Microsoft

documentation.

The System Driver and System Tags

The system driver is an internal process that monitors key variables
within an operating Wonderware Historian and outputs the values by
means of a set of system tags. The system driver runs as a Windows
service and starts automatically when the storage system is started.
The system tags are automatically created when you install the
historian. Also, additional system tags are created for each IDAS and
replication server you configure.
The current value for an analog system tag is sent to the storage
subsystem according to a specified rate, in milliseconds. All date/time
tags report the local time for the historian.
Error Count Tags

The following analog tags have a storage rate of 1 minute (60000 ms).
All error counts are since the Wonderware Historian is restarted or
since or the last error count reset.
TagName Description
SysCritErrCnt Number of critical errors

SysErrErrCnt Number of non-fatal errors
SysFatalErrCnt Number of fatal errors
SysWarnErrCnt Number of warnings
Date Tags
The following analog tags have a storage rate of 5 minutes (300000
ms).
TagName Description
SysDateDay Day of the month

SysDateMonth Month of the year
SysDateYear Four-digit year

The System Driver and System Tags45
Time Tags
All of the following tags are analog tags. Each value change is stored
(delta storage).
TagName Description
SysTimeHour Hour of the day

SysTimeMin Minute of the hour
SysTimeSec Second of the minute
Storage Space Tags

The following analog tags have a storage rate of 5 minutes (300000
milliseconds). Space remaining is measured in MB.
TagName Description
SysSpaceAlt Space left in the alternate storage path

SysSpaceBuffer Space left in the buffer storage path
SysSpaceMain Space left in the circular storage path
SysSpacePerm Space left in the permanent storage path
I/O Statistics Tags

The following analog tags can be used to monitor key I/O information.
TagName Description
SysDataAcqNBadValues* Number of data values with bad quality received. This tag
has a storage rate of 5 seconds. The maximum is 1,000,000.
SysDataAcqNOutsideRealtime* The number of values per second that were discarded
because they arrived outside of the real-time data window.
This tag has a storage rate of 5 seconds. The maximum is
1,000,000.
SysDataAcqOverallItemsPerSec The number of items received from all data sources. This
tag has a storage rate of 10 seconds. The maximum is
100,000.
SysDataAcqRxItemPerSecN* Tag value update received per second. This tag has a
storage rate of 10,000 milliseconds. Updated every 2
seconds for this IDAS.
SysDataAcqRxTotalItemsN* Total number of tag updates received since last startup for
this IDAS. This tag has a storage rate of 10,000
milliseconds.

TagName Description
SysStatusRxItemsPerSec Tag value update received per second. Updated every 2

seconds for the system driver (aahDrvSvc.Exe). This tag
has a storage rate of 1,000 milliseconds.
SysStatusRxTotalItems Total number of tag updates received since last startup for
the system driver. This tag has a storage rate of 10,000
milliseconds.
SysStatusTopicsRxData Total number of topics receiving data.
*This status tag will exist for each defined IDAS. The identifying
number (N) in the is the IODriverKey from the IODriver table. The
number 0 designates MDAS and only applies to the
SysDataAcqNBadValues and SysDataAcqNOutsideRealtime tags.
System Monitoring Tags

Unless otherwise noted, for the following discrete tags, 0 = Bad; 1 =
Good.
Tag Description
SysClassicManual Status of the classic manual storage service (aahManStSvc.exe).

Storage
SysClassicStorage Status of the classic storage service (aahStoreSvc.exe).
SysClientAccessPoint Status of the Client Access Point service
(aahClientAccessPoint.exe).
SysConfiguration Status of the configuration service (aahCfgSvc.exe). This parameter
is set to 1 as long as a dynamic configuration is required or in
progress.
SysDataAcqN* Status of the IDAS service (aahIDASSvc.exe).
SysEventSystem Status of the classic event system service (aahEventSvc.exe).
SysIndexing Status of the indexing service (aahIndexSvc.exe).
SysInSQLIOS Status of the Wonderware Historian I/O Server (aahIOSvrSvc.exe).
SysManualStorage Deprecated. Use SysClassicManualStorage instead. Status of the
classic manual storage service (aahManStSvc.exe).
SysMDASServer Deprecated. Use SysClientAccessPoint instead. Status of the MDAS
service (aahMDASServerSvc.exe).
SysOLEDB Status of the OLE DB provider (loaded by SQL Server).
SysPulse Discrete "pulse" tag that changes every minute.
SysReplication Status of Replication service (aahReplSvc.exe).

Tag Description
SysRetrieval Status of the retrieval service (aahRetSvc.exe).

SysStorage Status of the storage process (aahStorage.exe).
SysSystemDriver Status of the system driver (aahDrvSvc.exe).
SysTier2Storage Deprecated. Use SysStorage instead. Status of tier-2 storage
(aahStorage.exe).
*This status tag will exist for each defined IDAS. The identifying
number (N) appended to the end of the tag is the IODriverKey from
the IODriver table.
Miscellaneous (Other) Tags

The following table describes miscellaneous tags.
Tag Description
SysConfigStatus Number of database items affected by a dynamic

configuration (that is, the number of entries in the
ConfigStatusPending table when the commit is
performed). This value is cumulative and not reset until
the system is completely restarted.
SysHeadroomXXX Used to monitor the number of "headroom" tags still
available. Analog tags are divided by byte size: 2, 4, or 8
byte. For more information, see "Pre-allocating Memory
for Future Tags" in Chapter 2, "Configuring Tags," in the
SysHistoryCacheFaults The number of history blocks loaded from disk per minute.
The maximum value is 1,000. The storage rate for this
analog tag is 60 seconds. For more information on the
history cache, see "Memory Management for Classic
Storage" on page 133.
SysHistoryCacheUsed Number of bytes used for history block information. The
maximum value is 3,000,000,000. The storage rate for this
analog tag is 30 seconds.
SysHistoryClients The number of clients that are connected to the Indexing
service. The maximum value is 200. The storage rate for
this analog tag is 30 seconds.
SysMinutesRun Minutes since the last startup. The storage rate is 60000
milliseconds for this analog tag.

Tag Description
SysRateDeadbandForcedValues The total number of values that were forced to be stored

as a result of using a swinging door storage deadband.
This number reflects all forced values for all tags since the
system was started.
SysString String tag whose value changes every hour
SysTagHoursQueried A floating point value updated every minute that
indicates the total number of "tag retrieval hours" queried
by all client applications during that minute. For example,
if a single trend queries four tags for a 15-minute period,
that is "1.0 tag retrieval hours".
All tags, including replication sync queue tags and
non-existent tags, are counted.
Unlicensed tags are not counted.
Classic Event Subsystem Tags

The following table describes the event subsystem tags.
TagName Description
SysEventCritActionQSize Size of the critical action queue. For more information, see
"Action Thread Pooling" on page 360.
SysEventDelayedActionQSize Number of entries in the delayed action queue.
SysEventNormActionQSize Size of the normal action queue.
SysEventSystem A discrete tag that indicates the status of the event system
service (aahEventSvc.exe). 0 = Bad; 1 = Good.
SysStatusEvent Snapshot event tag whose value changes every hour.
Event Storage System Tags

The following table describes the system tags for monitoring the
storage operation of alarms and events. It should be noted that both
alarms and events are stored as event records in the database. The
system tags below track some performance value related to the storage
of events into the database..
TagName Description
SysStatusRxTotalEvents Total number of events received since startup.

SysStatusRxTotalDuplicateEvents Used as a diagnostic tool. Total number of duplicate
events received through different channels since
startup (and discarded as duplicates).

TagName Description
SysStatusRxEventsPerSec Number of events received per second, calculated

every 10 seconds.
SysStatusEventCommitPending A discrete tag that indicates the status of the commit.
Commit is the number of events that have not yet
been written to storage (SQL). 0 = No commit
pending; 1 = Commit pending.
SysStatusAverageEventCommitSize Average event commit size. Size is the number of
events written to the database per minute. This value
is averaged and stored every minute.
SysStatusAverageEventCommitTime Average event commit time in seconds.
Replication Subsystem Tags

The Replication Service collects the following custom performance
counters about its own operation, where N is a primary key of the
tier-2 historian in the Runtime database of the tier-1 historian. These
values are stored cyclically every 10 seconds.
TagName Description
SysReplicationSummaryCalcQueueItemsTotal Current number of summary calculations

stored in the summary calculation queue of
all tier-2 historians.
SysReplicationSummaryClientsTotal Current number of concurrent retrieval
clients performing summary calculations on
the tier-1 historian for all tier-2 historians.
SysReplicationSyncQueueItemsN Current number of items stored in the
synchronization queue on the tier-2
historian of key N.
SysReplicationSyncQueueItemsTotal Current number of items stored in the
synchronization queue on the tier-1 for all
tier-2 historians.
SysReplicationSyncQueueValuesPerSecN Average synchronization queue values per
second sent to the tier-2 historian of key N.
SysReplicationSyncQueueValuesPerSecTotal Average values processed by the replication
synchronization queue processor for all
tier-2 historians.
SysReplicationTotalTagsN Total number of tags being replicated to the
tier-2 historian of key N.

TagName Description
SysReplicationTotalValuesN Total number of values sent to the tier-2

historian of key N since the startup of the
replication service.
SysReplicationTotalValuesTotal Total number of values sent to all tier-2
historians since the startup of the
replication service.
SysReplicationValuesPerSecN Average values per second sent to the tier-2
historian of key N
SysReplicationValuesPerSecTotal Average values per second sent to all tier-2
historians.

Performance Monitoring Tags

You use performance monitoring tags to monitor CPU loading and
other performance parameters for various Wonderware Historian
processes. (All of these values map to equivalent counters that are
used in the Microsoft Performance Logs and Alerts application.)
The following tags allow you to monitor the percentage CPU load for
each processor (up to a total of four), as well as the total load for all
processors:
• SysPerfCPU0
• SysPerfCPU1
• SysPerfCPU2
• SysPerfCPU3
• SysPerfCPUTotal
The remaining system tags are used to monitor performance for each
historian process that runs as a Windows service and for the Microsoft
SQL Server service. For more information on services, see
"Wonderware Historian Processes" on page 42.
There are six system performance tags per each service. These tags
adhere to the following naming convention, where XXX designates the
service (ClientAccessPoint, ClassicManualStorage, ClassicStorage,
Config, DataAcq, EventSys, Indexing, InSQLIOS, Replication,
Retrieval, SQLServer, MetadataServer, EventStorage, Storage, or
SysDrv):
SysPerfXXXCPU
SysPerfXXXHandleCount
SysPerfXXXPageFaults
SysPerfXXXPrivateBytes
SysPerfXXXThreadCount
SysPerfXXXVirtualBytes
These tags have a cyclic storage rate of 5 seconds.
Note: The six performance tags will exist for each defined IDAS. The
identifying number (N) appended to the end of the "DataAcq" portion of
the tagname is the IODriverKey from the IODriver table. For
example, 'SysPerfDataAcq1CPU'.

The following table describes the suffixes assigned to the names of

system performance tags:
Suffix Description
CPU Current percentage load on the service.

HandleCount Total number of handles currently open by each
thread in the service. A handle is a identifier for
a particular resource in the system, such as a
registry key or file.
PageFaults Rate, per second, at which page faults occur in
the threads executing the service. A page fault
will occur if a thread refers to a virtual memory
page that is not in its working set in main
memory. Thus, the page will not be fetched from
disk if it is on the standby list (and already in
main memory) or if it is being used by another
process.
PrivateBytes Current number of bytes allocated by the service
that cannot be shared with any other processes.
ThreadCount Current number of active threads in the service.
A thread executes instructions, which are the
basic units of execution in a processor.
VirtualBytes Current size, in bytes, of the virtual address
space that is being used by the service.
Important: You need to ensure that the memory that SQL Server
reserves for the Wonderware Historian is adequate for the expected
load. Based on your particular environment, you may need to adjust
the SQL Server MemToLeave allocation. For more information on
MemToLeave, see the Microsoft documentation.

Supported Protocols53
Supported Protocols
Wonderware Historian supports the following protocols:
Protocol Description
DDE DDE is the passage of data between applications,

accomplished without user involvement or
monitoring. In the Windows environment, DDE is
achieved through a set of message types,
recommended procedures (protocols) for processing
these message types, and some newly defined data
types. By following the protocols, applications that
were written independently of each other can pass
data between themselves without involvement on the
part of the user. For example, InTouch HMI software
and Excel.
SuiteLink SuiteLink is a protocol that provides substantially
more application level functionality than that
provided by DDE. The SuiteLink protocol allows for
the passing of time stamp and quality information
with process data in the data packets.
System The system protocol is an internal means of passing
protocol data from system variables to the historian.
Note: DDE is not supported if the historian is running on the Windows

Server 2003, Windows Server 2008, Windows Vista, or later operating
system.
Modification Tracking
Wonderware Historian tracks modifications (inserts, updates, and
deletions) to columns in the Runtime database. If your plant tracks
changes for compliance with regulatory agencies, you can configure
the historian to use modification tracking.
Modification tracking is system-wide; it is controlled by the
ModLogTrackingStatus system parameter. You cannot turn
modification tracking on or off at a table level. Enabling modification
tracking decreases the historian's performance when making changes
to the system. This is due to the extra work and space required to
track the changes. However, there is no performance degradation
during run-time operation.
Information in the modification tracking tables are stored in the data
files of the Microsoft SQL Server database. If modification tracking is
turned on, the amount of data that is stored in these files is greatly
increased.

All of the objects for which modifications can be tracked are stored in
the HistorianSysObjects table.
You can track two types of modifications:
• Changes to configuration data. For example, additions or changes
to tag, I/O Server, and storage location definitions. For more
information, see "Modification Tracking for Configuration
Changes" on page 54.
• Changes to history data. For example, data inserts and updates by
Transact-SQL statements or CSV imports. For more information,
see "Modification Tracking for Historical Data Changes" on
page 55.
The types of changes that will be tracked is controlled by the
ModLogTrackingStatus system parameter. You can track inserts,
updates, and deletes, as well as various combinations. For more
information, see "Turning Modification Tracking On/Off" in Chapter 9,
"Viewing or Changing System-Wide Properties," in the Wonderware
Historian Administration Guide.
Modification Tracking for Configuration Changes

For configuration data, when a modification is made to a table in the
database, a record for the modification is inserted into the
ModLogTable table. One row will be inserted for each separate type of
modification, either an insert, update, or delete.
The actual value changes are recorded in the ModLogColumn table.
Each column that is modified will result in a row inserted into the
ModLogColumn table. The entry in the ModLogColumn table includes
both the column value before the change and the new column value.
For example, if you added (inserted) a single analog tag to the system,
the following changes would be reflected in the modification tracking
tables:
• Two rows would be added to the ModLogTable table, one to track
the change to the Tag table and one to track the change to the
AnalogTag table.
• One row for each of the columns in both of the Tag and AnalogTag
tables will be added to the ModLogColumn table.

Modification Tracking55
As another example, if you updated for a single analog tag the

StorageType column in the Tag table and the ValueDeadband and
RateDeadband columns in the AnalogTag table, the following changes
would be reflected in the modification tracking tables:
• Two rows would be added to the ModLogTable table, one to track
the change to the Tag table and one to track the change to the
AnalogTag table.
• Three rows would be added to the ModLogColumn table to record
the changes to the StorageType, ValueDeadband, and
RateDeadband columns.
• Important things to note:
For a tier-2 historian, modification tracking for a replicated tag
appears as the being made by the system account that the
configuration service is running under, which is typically NT
AUTHORITY\SYSTEM. To find out who modified a tag, examine the
ModLogTable of the tier-1 historian.
If modification tracking is enabled and tags are configured in offline
mode and modified, only the last known modifications are recorded in
the modification tables.
Modification Tracking for Historical Data Changes

Modifications to history data can be performed by either executing
Transact-SQL statements or by using the CSV import functionality. In
the case of Transact-SQL statements, the Wonderware Historian OLE
DB provider provides the change information to the modification
tracking tables by means of a stored procedure. This stored procedure
is also used by the storage subsystem to communicate changes that are
the result of a CSV import.
Although the history data that is changed is physically stored on disk
in the history blocks, for the purposes of modification tracking, the
data is considered to reside in the History_OLEDB extension table.
For more information on extension tables, see "Extension (Remote)
Tables for History Data" on page 139.
When a modification is made to history data, a record for the
modification is inserted into the ModLogTable table. One row will be
inserted for each separate type of modification, either an insert or an
update, for each tag.
The ModLogColumn table is used to store details for the column
modification in the History_OLEDB table. The modified column will
always be the vValue column. The total count of consecutive value
changes attempted per tag is stored in the NewValue column of the
ModLogColumn table.

The OldValue column contains the value stored in the column before
the modification was made, if the modification was to a configuration
table. For modifications to history data using SQL INSERT and
UPDATE statements, this column contains the timestamp of the
earliest data affected by the INSERT or UPDATE operation. If
multiple changes are made to the same data, then only the most recent
change will be contained in this column. This column is not used for
modifications made to history data using a CSV file.
For example, if you insert 20 data values into history for the
ReactTemp analog tag using a CSV import, the following changes
would be reflected in the modification tracking tables:
• One row would be added to the ModLogTable table, to track the
change to the History_OLEDB table. The UserName column will
contain the name of the user as contained in the CSV file header.
• One row would be added to the ModLogColumn table to record that
the value change occurred. A value of 20 will be stored in the
NewValue column to indicate that 20 values were inserted.
Data Categories
There are various ways to store data in Wonderware Historian and
there are many operations that can be performed on that data along
the way. To provide a consistent framework for data operations in the
Wonderware Historian data model, internally the data is divided into
several categories and organized into the following hierarchy.
Each category of data has a separate set of characteristics and is

handled differently by the historian. Also, some data categories have
differences based on whether the data is being handled by storage or
classic storage. For more information on storage, see Chapter 5, "Data
Storage Subsystem." For more information on classic storage, see
Chapter 6, "Classic Data Storage Subsystem."

Data Categories57
Original Data and Revision Data

Original data is the data received from the data source
originally—that is, for the first time. For example, a real-time stream
of data from an I/O Server represents original data.
The opposite kind of data is called revision data, when you already
have some original data stored for some tags and time interval, but
would like to correct it by adding some new revision values, and maybe
"masking" some original ones.
Usually the original data arrives to a historian server in high data
volumes for many tags where VTQs have timestamps close to each
other. In contrast, the revision data operations are performed on a
per-tag basis and are expected to be much less in volume compared to
the original data.
While the distinction between original and revision data happens from
the nature of the data, the distinction between streamed or
non-streamed original data depends from the ability of the data source
to enforce the time order of its output.
Note: For the Historian SDK, if you enable store-and-forward and

send streamed data, then the data is cached. However, for revision
data, the data is not cached and the server must be available. For more
information, see "Storing Data through HCAL" on page 95.
Original Streamed Data

Original streamed data is expected to be in time order where every
subsequent VTQ of the same tag has a timestamp not older than the
preceding one. An example of the original streamed data could be data
received from an I/O Server or DAServer sending a new device reading
with the current timestamp every few seconds.
Original streamed data:
• Can be efficiently replicated at storage time.
• Is exposed through the Historian I/O Server and the Live table.
• Is handled by store-and-forward. For more information, see
"Storing Data through HCAL" on page 95.
• Supports advanced storage configuration algorithms. For tags that
use classic storage, storage rules are applied by the historian
server, instead of by the historian client components.
If the data source configured for sending original streamed data starts
violating the time order, automatic timestamp correction occurs. For
more information, see "Data Adjustments Performed by HCAL" on
page 96.

Make sure that the Historian Server computer has its timestamp
configured properly and is synchronized with other computers serving
data as data inserted with timestamp in the future will be
timestamped again to the Historian time at the time of processing.
There are three sub-types of original streamed data: real-time data,
late data, and replication data.
Real-Time Data
Real-time data is expected to be in time order, where the timestamp is
in the past relative to the current Wonderware Historian server time.
Sources of real-time data values are:
• I/O Servers and DAServers sending data with current timestamps
• System tags
• Transact-SQL INSERT statements where wwVersion =
REALTIME
• Wonderware Application Server
If the tag is stored using classic storage, then the streamed data must
fall within the real-time window boundaries to be considered real-time
data. For more information on the real-time window, see "About the
Real-Time Data Window for Classic Storage" on page 127.
All real-time data contains store-and-forward indicators in the
QualityDetail.
If a value has timestamp that is beyond 1 second in the future, relative
to the Historian Server time, it is later overwritten by another data
value, and the QualityDetail value will reflect the overwrite.
For more information on storage, see Chapter 5, "Data Storage
Subsystem."
How Disconnects are Handled for Real-time Data

If a disconnect occurs between the client and the historian server, for a
real-time tag, a NULL is injected into the data stream at the moment
of the disconnect.
There is a ChannelStatus column in the Tag table that is used for tags
from Wonderware Application Server 2012 R2 or later or the
Wonderware Historian SDK 2012 R2 or later. This tag property
controls how disconnects between these sources and the Wonderware
Historian are reflected in the data until the disconnect period can be
backfilled with store-and-forward data.

Data Categories59
• If the channel status is enabled (set to 1), NULL values are

injected into the data stream for the disconnect period. For a trend,
this means that a line gap appears during the period of NULL
values. The tag remains in store-and-forward mode until the
timestamps become greater than the startup time of the server or
the time that the connection was restored.
• If the channel status is not enabled (set to 0), NULL values are not
injected and no gap is shown in client-side trends.
The ChannelStatus column value is ignored for tags that use classic
storage.
Late Data
Late data is expected to be in time order, where the timestamp is far in
the past compared to the current Wonderware Historian server time.
Typical sources of late data values are:
• I/O Servers (RTUs) that send data far in the past compared to the
historian server time
• Wonderware Application Server that has been configured to send
late data to the historian
Late data contains store-and-forward indicators in the QualityDetail
value, but does not have the channel status enabled, so no NULL
values are injected if a disconnect occurs between the HCAL-enabled
client and the historian.
If the late data values are not in time order, then the value with the
later timestamp is overwritten with a manufactured value, and the
QualityDetail value reflects the overwrite.
Replication Data
Replication data is data that has been replicated from a tier-1
historian to a tier-2 historian. Replication data does not contain
store-and-forward indicators if a disconnect and a subsequent
store-and-forward occurs between the tier-1 and tier-2 historians.
Original Non-Streamed Data

Original non-streamed data can be in any time order. Typical sources
of original non-streamed data values are:
• Transact-SQL statements with wwVersion = 'ORIGINAL'
• .CSV imports with an operation type of 'ORIGINAL VALUE'
• Fastload .CSV imports

Omitting the time order requirements means that original

non-streamed data:
• Cannot have storage rules applied, unless you are performing a
"fast load" import.
• Can be replicated only through a less efficient queued replication
mechanism.
• Cannot be published through the Historian I/O Server and the
Live table.
• Cannot be handled by store-and-forward.
Original non-streamed data with a timestamp in the future is
discarded.
Latest Revision Data

Latest revision data is the addition of new a version of data that has
already been stored.
Typical sources of latest revision data values are:
• Transact-SQL statements such as INSERTs or UPDATEs with
wwVersion = 'LATEST '
• .CSV imports with operation types of 'INSERT', 'UPDATE', and
'MULTIPOINT UPDATE'
For example, a set of original data values collected from an I/O Server
contains a series of VTQs that are known to be incorrect. An automatic
aggregation report built based on those values could produce an
unexpected outcome unless those already-stored original VTQs are
corrected. To correct the data, you can perform a latest revision data
operation (in this case an T-SQL UPDATE or a .csv file import)
resulting in retrieving the expected results when requested.
The revision data operations never physically delete or alter
already-stored original data. Instead, revision data modifications are
performed in a separate data stream allowing the original data to be
still retrievable in its intact form. The latest revision data has the
same limitations as for original non-streamed data.
New-Original Revision Data

The new-original data category is used internally exclusively for
replicated data.

Data Quality61
Data Quality
There are three aspects of quality handling for the system:
• Data quality assignment by data acquisition devices (OPCQuality)
• Storage subsystem quality for the Wonderware Historian
(QualityDetail)
• Client-side quality definitions (Quality
For more information on quality for tags, see "VTQ for Tags" on
page 23
The historian uses three distinct parameters to indicate the quality of
every historized data value: Quality, QualityDetail, and OPCQuality.
OPCQuality is strongly related to QualityDetail. Quality is a derived
measure of the validity of the data value, while QualityDetail and
OPCQuality indicate either a good quality for the data value, or the
specific cause for degraded quality of the data value.
The historian persists four bytes of quality information for every data
value that is historized. (This does not imply that four bytes of quality
data are actually stored for every data value, but it guarantees that
every data value can be associated with 4 bytes of quality information
when the value is retrieved.) The 4 bytes of quality information
comprises 2 bytes for QualityDetail and 2 bytes for OPCQuality.
QualityDetail and OPCQuality are related, but may have different
values.
In essence, OPCQuality is provided as for client applications that are
fully aware of the OPC data quality standard, while QualityDetail
(and Quality) are maintained to ensure proper operation of existing
and legacy client applications. OPCQuality is sent by the data source,
and QualityDetail is added by the Wonderware Historian.
If you import history blocks from IndustrialSQL Server 7.1 or earlier
(that do not use OPC data quality), the OPCQuality is determined by
using the top 2 bits of the lower 8 bits of the QualityDetail.

Viewing Quality Values and Details

The currently supported OPCQuality values are stored in the
OPCQualityMap table of the Runtime database. To view OPCQuality
values and descriptions, execute the following query:
SELECT OPCQuality, Description FROM OPCQualityMap
The currently supported QualityDetail values are stored in the

QualityMap table of the Runtime database. To view the complete list
of QualityDetail values and descriptions, execute the following query:
SELECT QualityDetail, QualityString FROM QualityMap
The following notes apply to the QualityDetail values:

• The boundary point for a QualityDetail of 448 is 1 second.
QualityDetail of 448 is a concatenation of quality states of good
quality 192 and the bit flag indicating a value in the future. For
more information, see "QualityDetail Bit Layout" on page 65.
• Although a quality detail of 65536 is used to indicate block gaps for
tier-2 tags, NULL values are not produced for block gaps for tier-2
tags.
Basic QualityDetail Codes

Quality detail codes are metadata used to describe data values stored
or retrieved using the Wonderware Historian. Quality details are 2
byte values that use the low order byte for the basic quality detail and
the high order byte to store quality detail flags. The basic quality
detail and quality detail flags are OR'd together.
The following table lists the basic quality detail codes. These codes use
the low order byte, and the individual bit values do not have
independent significance. For information on how to retrieve the
complete list of quality detail codes, see "Viewing Quality Values and
Details" on page 62.
Created
QualityDetail Quality String Description By
0 Bad Quality of Bad quality. Storage

undetermined state
1 No data available, tag did For cyclic, delta, full, Retrieval
not exist at the time interpolated, and best-fit,
retrieval started before the tag
was created (NULL value).
10 Communication loss Disconnect between a client Classic
and the historian server. storage,
HCAP

Created
17 Duplicate time stamp; Slope calculation of a vertical Retrieval

infinite slope line.
20 IDAS overflow recovery An IDAS data buffer overflow IDAS
was detected, and all overflow
values were discarded to
recover.
24 IOServer communication I/O Server communication Classic
failed failure. Application Server storage,
sends quality of 24 or 32. HCAL,
MDAS
33 Violation of History You have queried beyond the Retrieval
Duration license feature licensed time range limit.
44 Pipe reconnect First VTQ sent after the client Classic
has reconnected to the storage,
historian. For tags with HCAL,
AITag=0 it is always an MDAS
artificially injected VTQ
created from the last known
VTQ on the client side at the
moment of the reconnect.
64 Cannot convert The calculated point was based Retrieval
on points of different
QualityDetails: Good,
Doubtful, or Bad. This is, for
example, used in the integral
and average retrieval modes.
150 Storage startup Storage startup. Classic
storage
151 Store forward storage Store-and-forward storage Classic
startup startup. storage
152 Incomplete calculated Incomplete calculated value. Replication
value
192 Good Good value. HCAL,
MDAS
202 Good point of version Current value is the latest Classic
Latest update which “masks” a storage,
previous original value. retrieval
212 Counter rollover has Counter count reached the Retrieval
occurred roll-over value.

Created
248 First value received in First VTQ stored in HCAL,

store forward mode store-and-forward mode after MDAS
the client has disconnected
from the historian.
249 Not a number Value received was not a Classic
number, infinite value (NaN). storage,
HCAL,
MDAS
252 First value received from First value received after an Classic
IOServer I/O Server reconnect. storage
65536 No data stored in history, NULL value. Retrieval
NULL
Note: Although this quality
detail is used to indicate block
gaps on Tier 1 tags, NULL
values are not produced for
block gaps for Tier 2 tags.
QualityDetail Flags
The following bit flags are stored in the high order byte of the quality
detail. Each bit is a flag that carries specific meaning.
0x0000 - Basic QualityDetail.
The following hi-byte flags are or'ed with the basic QualityDetail. The
three flags cannot coexist; only one is set at any point in time.
0x0100 - Value received in the future.
0x0200 - Value received out of sequence.
0x0400 - Configured for server time stamping.
The following hi-byte flag can coexist with any of the prior four
combinations.
0x0800 - Point generated by Rate Deadband filter.
The following hi-byte flag can coexist with any of the prior eight
combinations.
0x1000 - Partial cycle, advance retrieval results.
The following hi-byte flag can coexist with any of the prior eight
combinations.
0x2000 - Value has been modified by filter.

This bit is set during retrieval to indicate that retrieval has modified a
value or time stamp due to applying a filter such as SnapTo() or
ToDiscrete().
QualityDetail Bit Layout

The QualityDetail bit layout is as follows:
Acquisition and Storage of Quality Information

When a data value is acquired for storage in the Wonderware
Historian, it is usually accompanied by a quality indicator generated
at the source of the data. In general, the historian passes the
source-supplied quality indicator to the storage subsystem unmodified,
but there are instances where the data acquisition subsystem actually
modifies the quality of the data value.
Quality for Data Acquired from I/O Servers

A data value acquired from a Wonderware I/O Server by an IDAS
always has 2 bytes of quality information attached to it by the I/O
Server. The IDAS presents the 2 bytes of quality information supplied
by the I/O Server to the Wonderware Historian storage subsystem as
OPCQuality, and the QualityDetail is set to 192. An exception to this
is if the quality value sent by the I/O Server is 24, in which case both
OPCQuality and QualityDetail are set to 24.
However, in some instances, IDAS will overwrite QualityDetail with a
special value to indicate a specific event or condition. For more
information on these values, see "Viewing Quality Values and Details"
on page 62. When IDAS overwrites QualityDetail with one of the
reserved values, it does not modify the OPCQuality value. Thus,
QualityDetail contains the reserved value set by IDAS, while
OPCQuality retains the value provided by the data source.
The SysDataAcqNBadValues system tag tracks the number of data
values with bad quality that are coming from an IDAS. If this value is
high, you should make sure that there is not problem with the IDAS or
I/O Server. For more information on system tags, see "The System
Driver and System Tags" on page 44.

Quality for Data Not Acquired from I/O Servers

The Wonderware Historian supports data acquisition from a variety of
sources other than Wonderware I/O Servers. These sources include
properly formatted CSV files, SQL queries, and the Application
Server.
Quality for data acquired from these sources is handled the same as
for data received from Wonderware I/O Servers, with a few exceptions:
• For data acquired from CSV files, the specified quality is always
interpreted as OPCQuality, and the QualityDetail is set to 192,
unless the specified quality is 24, in which case both OPCQuality
and QualityDetail are set to 24.
• If MDAS is responsible for acquiring the data (as is the case for
SQL queries and Wonderware Application Server), the quality
value presented by the source is preserved in OPCQuality, and
QualityDetail is set to 192. Exceptions to this are:
• If a quality value of 32 presented by the source to MDAS,
OPCQuality is set to 32 and QualityDetail is set to 24.
• If the data value presented by the source is infinite or NaN (not
a number), OPCQuality contains the actual quality presented
by the source, and QualityDetail is set to 249.
• If a special condition or event occurs, MDAS will substitute a
reserved value for QualityDetail. For more information, see
"Viewing Quality Values and Details" on page 62.
Client-Side Quality Values

Three quality indicators are exposed to clients:
• A 1-byte short quality indicator (Quality)
• A 4-byte long quality detail indicator (QualityDetail)
• A 4-byte long OPC quality indicator (OPCQuality)
Quality contains summary information that essentially falls into the
categories of Good, Bad, Doubtful, or InitialValue. This implicit
information is derived from the data source as well as the Wonderware
Historian.

Quality is a 1-byte indicator that provides a summary meaning for the

related data item. Quality is an enumerated type with the following
values:
Hex Dec Name Description
0x00 0 Good Good value.

0x01 1 Bad Value marked as invalid.
0x10 16 Doubtful Value is uncertain.
0x85 133 Initial Value (Good) Initial value for a delta request.
Initial Value and Doubtful are derived from QualityDetail. The Initial
Value is dependent on the type of query that is executed. For more
information, see "Using Comparison Operators with Delta Retrieval"
on page 296, "Using Comparison Operators with Cyclic Retrieval and
Cycle Count" on page 301, and "Using Comparison Operators with
Cyclic Retrieval and Resolution" on page 305.
QualityDetail contains detailed information that is potentially specific
to the device which supplied it. Quality values may be derived from
various sources, such as from I/O Servers or from the Wonderware
Historian storage system.
To retrieve a listing of the quality values that are used by the
historian, see "Viewing Quality Values and Details" on page 62.


69
Chapter 3
Configuration Subsystem
Configuration data is information about elements that make up the

Wonderware Historian, such as tag definitions, I/O Server definitions,
and storage locations for historical data files. Configuration data is
relatively static and does not change frequently during normal plant
operation. The configuration subsystem stores and manages
configuration data.
Setting up the required databases and included entities (such as
tables, stored procedures, and views) to support a typical factory
environment would take countless hours. However, when you install
the historian, all of these entities are defined for you, allowing you to
quickly start using Wonderware Historian.
Configuration data is stored in SQL Server tables in the Runtime
database. If you are already using InTouch® HMI software, you can
easily import much of this information from existing InTouch
applications, thus preserving your engineering investment. If you are
using Application Server, much of the Wonderware Historian
configuration is handled automatically by Application Server. You can
also use the System Management Console to manually add definitions
and configure the system. You can make bulk modifications to your
historian configuration or migrate the configuration from one
historian to another using the Wonderware Historian Database
Export/Import Utility.
You can reconfigure the system at any time with no interruption in the
acquisition, storage, and retrieval of unaffected tags. Configuration
data can be stored with a complete revision history.
For information on how to configure your system, see Chapter 1,
"Getting Started with Administrative Tools," in the Wonderware

70  Chapter 3 Configuration Subsystem
Configuration Subsystem Components

The components of the configuration subsystem are:
Component Description
Runtime database SQL Server database that stores all

configuration information.
Configuration and Consists of the System Management
management tools Console client application, the
Wonderware Historian Database
Export/Import Utility, and the
configuration tools shipped with
Microsoft SQL Server. For more
information, see Chapter 1, "Getting
Started with Administrative Tools," in
the Wonderware Historian
Configuration Service Internal process that handles all status
(aahCfgSvc.exe) and configuration information
throughout the system. This process
runs as a Windows service.
About the Runtime and Holding Databases

A relational database management system (RDBMS) such as
Microsoft SQL Server can contain many databases. A database is a
collection of objects such as:
• Tables
• Stored procedures
• Views
• User-defined data types
• Users and groups
The Wonderware Historian is shipped with two pre-configured
databases: the Runtime and Holding databases.

About the Runtime and Holding Databases71
The historian embeds a full-featured Microsoft SQL Server. The

historian supports all system tables associated with SQL Server. For
more information on the Microsoft SQL Server tables, see your
Microsoft documentation.
Note: When installed on a case-sensitive Microsoft SQL Server, the

Runtime and Holding databases are case-sensitive. Be sure that you
use the correct case when performing queries.
The Runtime Database

The Runtime database is the online database against which the
Wonderware Historian runs. The tables within the Runtime database
store all configuration information, such as:
• System configuration.
• Tag definitions.
• InTouch integration information.
• System namespaces and grouping information.
• Event configuration information.
• User-entered annotations.
Runtime database tables are usually used as references or lookup
tables by the historian and client applications. Any changes to the
historian are reflected in these configuration tables. The configuration
tables exist as normal SQL Server tables and data within them can be
modified by using the Microsoft Transact-SQL query language. For
more information on Transact-SQL, see your Microsoft documentation.
The Runtime database also stores some types of history data:
• Modification tracking data
• Event subsystem data
Tables that store modification tracking and event data are also normal
SQL Server tables.
Finally, the Runtime database is used to logically store historized tag
values. Although the tag values are stored in the history block files on
disk, the values appear to be saved to tables in the Runtime database.
For more information on history blocks, see "History Blocks" on
page 114. For more information on retrieving historized tag values, see
Chapter 7, "Data Retrieval Subsystem."
Note: You cannot change the name of the Runtime database.

The Holding Database

The Holding database temporarily stores topic and configuration data
imported into Wonderware Historian from an InTouch node. When you
import configuration data from an InTouch application, the data is
first mapped to table structures in the Holding database. Then, the
data is moved into the Runtime database.
Important: Do not modify any entities in the Holding database.
For more information about importing configuration information from

an InTouch application, see Chapter 3, "Importing and Exporting
Configuration Information," in the Wonderware Historian
About the Configuration Service

The Configuration Service is an internal process that accepts
configuration changes and updates the Runtime database. Thus, the
Configuration Service is the only component that interacts with the
configuration store.
The Configuration Service runs as a Windows service and accepts and
distributes configuration information to and from different parts of the
system by a set of interfaces. The Configuration Service also serves as
a gateway for all information pertaining to the status of the different
components of the Wonderware Historian.
Dynamic Configuration
The Wonderware Historian supports dynamic configuration; that is,
you can modify the configuration of tags and other objects in the
historian database while the system is running. The historian
automatically detects and applies the modifications to its internal
run-time state without requiring the system to be restarted. In
addition, clients do not experience interruptions due to configuration
changes.
The dynamic configuration feature in the historian caters for all
possible database modifications that affect the run-time operation of
the system. For some types of configuration modifications, the system
automatically creates a new history block. The configuration
subsystem is designed to ensure that no loss of data occurs for tags
that are not affected by the modifications being applied. However, tags
that require a change in data acquisition configuration will obviously
lose data during the reconfiguration.

Dynamic Configuration73
In most cases, the system continues to run uninterrupted. In the

following cases, a restart of the system is required:
• When you change the main historization path in the system, a
parameter that is rarely modified after installation.
• When you modify the DataImportPath system parameter.
For a description of the effect of various types of modifications made
while the system is running, see "Effects of Configuration Changes on
the System" on page 74.
Dynamic configuration is usually a two-step process:
1 Add, modify, or delete one or more objects in the database, using
the System Management Console, Transact-SQL statements, or
the database modification tool of your choice.
As soon as you make a change, the Runtime database is updated to
reflect the change. For example, when you add an analog tag using
the wizard within the Configuration Editor, the database is
updated as soon as you click Finish.
2 After making all of the modifications, you must commit the
changes, which triggers the dynamic configuration process in the
server. Modifications made to the system are done in a
transactional fashion.
The database modifications are not reflected in the running system
until you commit the changes. You are committing changes to the
system, not to the database.
You can commit changes to the configuration of the system as often as
you want. You can also commit changes in batches or individually.
There is no limit on the number of changes that may be committed to
the system. Configuration changes typically take effect within 10
seconds under maximum data throughput conditions, unless a new
history block is required, in which case the changes take longer (3 to 5
minutes) to take full effect.
For information on cases in which a commit is prevented, see "Cases in
Which Configuration Changes are not Committed" on page 75.

Effects of Configuration Changes on the System

Different types of dynamic changes to the database affect the system
in different ways.
Some types of modifications require a new history block to be created.
To reduce the need to create new history blocks, the system includes
"headroom" tags, which allocate extra space in the blocks to
accommodate new tags. For more information on headroom tags, see
"Pre-allocating Memory for Future Tags" in Chapter 2, "Configuring
Tags," in the Wonderware Historian Administration Guide.
A summary of typical changes and their effect on the system after a
commit is as follows.
• Modifying system parameters
A modification to system parameters usually takes effect
immediately (a new history block will not be created). The
exception is adding headroom for one or more tag types, which
requires a new history block. Also, if you change the
HistoryCacheSize parameter and commit the change, the cache is
not immediately flushed to bring the cache size to less than or
equal to the new value. You must rescan the history blocks to flush
the cache.
• Modifying storage locations
Modifying the circular storage location requires a shutdown and
restart of Wonderware Historian. Changes to the other storage
locations take effect immediately.
• Adding, deleting, and modifying tags
Adding one or more tags to the system generally results in a new
history block being created, unless sufficient headroom is available
for that particular tag type, in which case a new block is not
required. If the headroom is exceeded, a new block is created and
the headroom is replenished to the amount specified in the
SystemParameter table.
Deleting one or more tags takes effect immediately.
Certain modifications to tags result in a new history block. Those
include changing the integer size, changing the raw type, changing
strings tags from fixed length to variable length or vice versa,
changing storage type from "Not stored" to "Stored," changing a
string tag from ASCII to Unicode or vice versa, and changing tag
acquisition type from"IOServer" to "Manual" or vice versa.
As a general guideline, modifications to a tag that changes its
footprint on disk will result in a new history block. If only data
acquisition or retrieval characteristics of a tag are modified, the
changes take effect without requiring the system to create a new
history block.

Dynamic Configuration75
Modifying data acquisition characteristics of a tag could result in a

brief period of data loss, for that tag. As a guideline, any change to
the source of data for the tag (for example, modifying the item
name, topic name, or I/O server name of the tag) will result in a
short gap in data for the tag, while the system disconnects from the
old data source and connects to the new data source.
• Adding, deleting, and modifying IDASs
Adding a new IDAS to the system results in a new set of system
tags being added (the status and performance system tags
associated with that IDAS). While adding an IDAS in itself does
not require a new history block, the new system tags will result in
a new block, unless sufficient headroom is available.
Deleting an IDAS takes effect immediately. Modifying an IDAS
never requires a new history block, but may result in data loss for
the tags serviced by that IDAS (for example, moving an IDAS to
another computer causes a disconnect from the data sources).
• Adding, deleting, and modifying I/O Servers and topics
Adding or deleting I/O Servers and/or topics does not require a new
history block. Modifying I/O Server or topic characteristics may
result in data loss for their tags, if the modification implies a
disconnect from the data source.
Cases in Which Configuration Changes are not

Committed
If the system is not running, or storage is stopped, any commit is
ignored and the contents of the ConfigStatusPending table are cleaned
up. The exceptions are changes to the following fields in the
SystemParameter table: HistoryCacheSize,
HistoryDaysAlwaysCached, and AutoStart.
If the system is running, a commit will be disallowed:
• While a previous dynamic configuration is still in progress.
• While a new history block is in progress (whether the block is the
result of a scheduled block changeover, a dynamic configuration, or
a user request). A block changeover is in progress for five minutes
after it has been created, and also ten minutes before the next
scheduled block changeover.
For each case, a message appears, indicating that the commit is
disallowed.


77
Chapter 4
Data Acquisition Subsystem
The Wonderware Historian has been designed for high-speed

acquisition of data, acquiring and storing process data many times
faster than a traditional relational database.
Application Server, DA Servers, and I/O Servers are the main sources
of plant data. The historian can acquire data from over 500
Wonderware and third-party I/O Servers, ensuring access to the
industry's most comprehensive list of data acquisition and control
devices. I/O Servers that use the SuiteLink protocol can provide time
and quality stamping at the I/O Server level. Data can be acquired
simultaneously from multiple I/O Servers over a variety of physical
links, with a remote store-and-forward capability to prevent data loss
in the event of failed network connection.
Custom client applications can be another source for real-time
historical data. Clients that use the Wonderware Historian Client
Access Layer (HCAL) or Manual Data Acquisition Service (MDAS) can
send historical tag values directly to the system.
You can batch import historical data formatted in a comma-separated
values (CSV) file, allowing you to migrate existing data from other
historians. You can also use the InTouch History Importer to easily
import history data from InTouch HMI applications.
Finally, the historian generates data for key internal status variables,
which allow you to monitor the health of the system.

78  Chapter 4 Data Acquisition Subsystem
Data Acquisition Components

The following table describes the components of the data acquisition
subsystem. Many of the components run as Windows Services.
I/O Server Wonderware-compatible software application

(DAServer) that reads values from PLCs and other factory
devices and forwards the real-time data to
Wonderware applications.
IDAS Service Process that accepts real-time data from one or
more I/O Servers and forwards it to a single
Wonderware Historian.
Query Tools Any database query tool capable of issuing
Transact-SQL INSERT or UPDATE
statements. For example, Microsoft SQL
Server Query Analyzer.
Data Import Defined file folder to batch import tag values to
Folder the historian.
InTouch History Utility to import data from one or more
Importer InTouch history files (.lgh). For more
information, see "Importing Data from an
InTouch History File" in Chapter 6,
"Importing, Inserting, or Updating History
Data," in the Wonderware Historian
Historian Client Process that can send non-I/O Server data to
Access Layer the historian to be historized. HCAL is used by
(HCAL) Application Server 2012 R2 or later and custom
client applications built with Historian SDK
2012 R2 or later.
Historian Client Process that can accept non-I/O Server data
Access Point and send it to the historian to be historized.
(HCAP) HCAP is used by Application Server 2012 R2 or
later and custom client applications built with
Historian SDK 2012 R2 or later.
Manual Data Process that can accept non-I/O Server data
Acquisition and send it to the historian to be historized.
Service (MDAS) Data is passed to MDAS through a COM
DLL interface. MDAS is used by Application Server
2012 or earlier, the Wonderware Historian
OLE DB provider, the event subsystem, and
custom client applications built with Historian
SDK 9.0.

Data Acquisition from I/O Servers79
Manual Data Windows service that can accept non-I/O

Acquisition Server data and send it to the historian to be
Service (MDAS) historized. Data is passed to MDAS through a
Server data layer.
System Driver Internal process that monitors the entire
Service historian system and reports the status with a
set of system tags. The system driver also
sends data values to the storage subsystem for
the current date and time, as well as for
pre-defined "heartbeat" tags, such as a discrete
system pulse. For more information, see "The
System Driver and System Tags" on page 44.
Data Acquisition from I/O Servers

An I/O Server provides data to Wonderware Historian from the DDE
or SuiteLink protocols. I/O Servers accept data values from
programmable logic controllers (PLCs), Remote Telemetry Units
(RTUs), and similar devices on the factory floor and can forward them
on to other software applications, such as InTouch HMI software or
the Wonderware Historian.
Note: FastDDE is not supported. DDE is not supported if the historian

is running on the Windows Server 2003, Windows Server 2008,
Windows Vista, or Windows 7 operating system.
"Server" and "client" are relative terms. In order for data to be

transmitted from the factory floor to a client application on the
business LAN, it must be "served up" through a chain of applications.
Any application that accepts data from a lower source is a client. Any
application that sends data to a higher source is a server. The
historian is both a client and a server; it is a client to the I/O Servers
and a server to the desktop clients.

To acquire data from an I/O Server, you must add the I/O Server
addressing information to the historian database and then associate
the I/O Server with an IDAS.
For information on configuring I/O Servers and IDASs, see Chapter 4,
"Configuring Data Acquisition," in the Wonderware Historian
I/O Server Addressing

All Wonderware-compatible I/O Servers use DDE addressing, which
includes the following distinct parts:
• Computer name. This is the node name of the computer running I/O
Server software.
• Application name. This is the name of the application supplying

data. The application name can include the name of the computer
on which the application is running.
• Topic name. A topic is an application-specific subgroup of data
elements.
• Item name. An item is a data value placeholder.
The format for the addressing is as follows:
\\<computername>\<applicationname>\<topicname>!<itemname>
The following table provides some examples of DDE addressing.
Address
Information I/O Server InTouch Microsoft Excel
application \\Computer1\Modbus \\Computer1\VIEW \\Computer1\Excel

name
topic name ModSlave5 Tag name Spreadsheet1
item name Status ReactLevel A1 (cell name)
For the Wonderware Historian to acquire data from an I/O Server, the
I/O Server addressing information must be added to the overall system
configuration. You can use the System Management Console to
manually add I/O Server definitions to the system, or you can import
I/O Server definitions from existing InTouch applications.
For more information about manually adding I/O Server definitions,
see Chapter 4, "Configuring Data Acquisition," in the Wonderware
For more information on importing I/O Server definitions from
InTouch HMI software, see Chapter 3, "Importing and Exporting
Configuration Information," in the Wonderware Historian

About IDASs
A Wonderware Historian Data Acquisition Service (IDAS) is a small
software application that accepts data from one or more I/O Servers or
DAServers. An IDAS runs as a Windows service named SysDataAcq.
The IDAS processes the acquired data, if necessary, and then sends
the data to the Wonderware Historian, where it is subsequently
stored.
Note: An IDAS was previously called an I/O Driver. IDAS configuration

information is stored in the IODriver table in the Runtime database.
When you add an I/O Server definition to the historian, a topic object is
created in the associated IDAS. A separate topic object exists for each
unique combination of I/O Server computer, application, and topic.
Each topic object maintains its own state: idle, connecting, connected,
disconnecting, disconnected, overloaded, or receiving. Also, each topic
object is assigned a data time-out value based on your assessment of
how often data changes for that particular topic.
An IDAS can accept data from one or more I/O Servers, but only sends
data to a single historian.
An IDAS can run on the same physical computer as the historian, or
on a remote computer. However, only one instance of an IDAS can run
on any single computer. Both the IDAS and the historian use NetBIOS
network names for communication. Computers must be accessible by
those names. Use the Ping command to check the availability of the
remote IDAS or historian computers.
IDAS seamlessly handles data values, irrespective of their time. For
each data point acquired by IDAS, the timestamp, value, and quality
are historized in accordance with the storage rules for the tag to which
the data value belongs.
For information on configuring an IDAS, see Chapter 4, "Configuring
Data Acquisition," in the Wonderware Historian Administration
Guide.
IDAS Configuration
During normal operation, when the historian is started, it configures
an IDAS by sending it information about the tags (including their data
sources) for which the IDAS is to acquire data. When the historian
storage subsystem is ready to accept data, IDAS automatically
connects to its data sources, starts acquiring data, and sends the data
to the historian storage subsystem for historization.
The primary purpose for IDAS configuration files is to minimize
network traffic and provide information for IDASs configured for
autonomous startup. For more information on autonomous startup,
see "IDAS Autonomous Startup" on page 86.

The IDAS saves configuration information to a file on the local hard

drive in the following folder of the IDAS computer: Document and
Settings\All Users\Application
Data\ArchestrA\Historian\IDAS\Configurations.
The IDAS configuration file is named as follows:
idatacfg_SERVERNAME_IDASKEY.dat
where:
• SERVERNAME is the NetBIOS name of the historian computer
• IDASKEY is the value of the IODriverKey column for the IDAS in
the Runtime database
You can change the IDAS configuration from the System Management
Console. The historian dynamically reconfigures itself. If the IDAS is
on a remote computer, the historian sends the updated configuration
information to the IDAS. The IDAS reconfigures itself and updates the
local configuration file. The IDAS continuously acquires and sends
data during the reconfiguration process. The historian saves its copy of
the updated IDAS configuration file in the following folder of the
historian computer: Document and Settings\All Users\Application
Data\ArchestrA\Historian\Configuration\IDAS Configurations.
After a successfully configuring IDAS, a copy of the IDAS
configuration file is stored on the historian computer. The IDAS
configuration file stored on the IDAS computer is identical.
Important: IDAS configuration files have a proprietary binary format.

Do not modify these files.
If there is more than one autonomous configuration file on the IDAS

computer (for example, if you deleted an IDAS on a node while it was
disconnected and then added one again), only the newest file is used. A
warning is logged on the IDAS computer. For more information on
autonomous startup, see "IDAS Autonomous Startup" on page 86.
IDAS Data Processing

An IDAS performs minimal processing on the data values it receives
from an I/O Server. The IDAS converts data values into storage data
types, depending on the type of the tag associated with the value. For
example, if the data value is associated with a floating point analog
tag, the incoming value is converted to a floating point value before
transmission to the storage subsystem. However, no timestamp
conversion is applied because both the I/O Servers and the
Wonderware Historian storage subsystem base time on Universal
Time Coordinated (UTC).

An IDAS does not apply any storage rules (for example, delta or cyclic
storage) unless it is disconnected from the historian and is operating
in store-and-forward mode. If the IDAS is operating in
store-and-forward mode, all storage rules are applied before the data is
stored locally in the store-and-forward history blocks.
Data Transmission to the Storage Subsystem

Data received by the IDAS are stored in a series of 64 KB buffers and
are periodically sent to the storage subsystem in packets.
The IDAS sends data from a buffer after the buffer is full, or every
second, whichever comes first. The number of pre-allocated buffers is
configurable for each IDAS. If you have high data rates for an IDAS
and you see log messages indicating the buffers are full, you may need
to increase the default buffer count.
For more information, see "Editing Advanced Information for an
IDAS" in Chapter 4, "Configuring Data Acquisition," in the
IDAS Security and Firewalls

A remote IDAS uses the network account specified at the time of the
IDAS installation to communicate with the historian. The historian
uses the network account specified at the time of the historian
installation to communicate with remote IDASs. You can also change
this account after installation using the ArchestrA Change Network
Account utility.
To communicate with the historian, the IDAS relies on the security
provided by the Microsoft Windows operating system and does not
send/receive any user names or passwords over the network.
An IDAS uses named pipes to communicate with the historian and
Microsoft file sharing for transmission of store-and-forward history
blocks. If a firewall exists between a remote IDAS and the historian
computer, the firewall must allow communication using ports from 135
through 139 (TCP/UDP) and port 445 (TCP/UDP).
For more information on IDAS file sharing requirements, see "IDAS
Store-and-Forward Capability" on page 84.

IDAS Error Logging

An IDAS logs all errors to the ArchestrA Logger Service. If the IDAS is
installed on a remote computer, the ArchestrA Logger Service will also
be installed on the remote computer. During normal operation, remote
IDAS errors are logged to both the local logger and the logger on the
Wonderware Historian computer.
If the network connection between the remote IDAS and the historian
fails, no error messages are sent to the logger on the historian
computer. Therefore, you should periodically use the System
Management Console to check the log on the remote IDAS computer to
ensure that no problems occurred. After the network connection is
restored, error messages are not forwarded to historian computer.
IDAS Store-and-Forward Capability

IDAS includes "store-and-forward" capability, which protects against a
temporary loss of data in the event that a remote IDAS cannot
communicate with the Wonderware Historian.
Note: The store-and-forward option is not available if you have

specified a failover IDAS.
If the remote IDAS cannot communicate with the historian, all data
currently being processed can be stored (cached) locally on the
computer running IDAS. This hard drive location is called the
store-and-forward path and is configurable using the System
Management Console.
If the IDAS is unable to send the data to the historian, data is written
to this path until the minimum threshold for the cache is reached, at
which point no more data is stored. An error message is logged.
Remote store-and-forward paths are not supported.
The following actions occur after the historian becomes available
again:
• The historian verifies that the IDAS configuration information did
not change while the IDAS was disconnected. The historian
attempts to restore data transmission from the IDAS. The IDAS
stops local data caching and resumes sending data acquired from
its data sources to the historian.

• If historian detects a difference between its version of the IDAS

configuration, and the IDAS version, it dynamically reconfigures
the IDAS to synchronize configuration information. The IDAS
applies the changes and updates its local IDAS configuration file.
Then, the historian requests restoring data transmission from the
IDAS.
• When the IDAS detects availability of the running historian, it
sends the store-and-forward data to the historian at the same time
it is sending real-time data.
The store-and-forward data is copied to the InSQL8SF$ network share
on the historian computer. The data is sent to the historian in chunks,
pausing for a configurable amount of time between chunks. The
remote IDAS uses the network account specified during the IDAS
installation to access the network share.
If the data stored on the remote IDAS node does not constitute a
complete history block (for example, due to a power failure), the data is
still forwarded to the historian and historized. A message is logged
that an incomplete block was processed.
If an error occurs when a store-and-forward block is processed, the
block is moved to the \Circular\Support directory. A message is
logged. Blocks are not automatically deleted from the \Support
directory. You must manually delete blocks to prevent them from
consuming disk space allocation for the circular storage location. If you
upgraded from the Wonderware Historian 8.0.x, then the block is
moved to the existing \Log directory.
After data from the store-and-forward cache is sent to the historian,
the cache is deleted from the IDAS computer.
Enabling IDAS store-and-forward mode increases system resources
used by the IDAS service because the store-and-forward subsystem
must be initialized and then maintained in standby mode, ready to
accept data.
If the historian computer has sufficient system resources, you can
configure the local IDAS to continue store-and-forward data collection
even if the storage subsystem is stopped.

IDAS Redundancy
For each IDAS that you define for the system, you can specify a
"failover" IDAS. If the Wonderware Historian stops receiving data
from the primary IDAS, it automatically switches to the failover IDAS.
The switch may take some short period of time, and some data may be
lost during the transition.
Note: You cannot specify a failover IDAS for an IDAS that has
store-and-forward functionality enabled. These two features are
mutually exclusive. Applications that require both failover and
store-and-forward functionality should use a redundant Application
Server with RedundantDIObjects.
IDAS Autonomous Startup

Normally, the Wonderware Historian Configuration Service starts an
IDAS. However, a remote IDAS that is enabled for store-and-forward
can be configured to start independently of the historian. Autonomous
startup is useful when the historian is unavailable due to a network
failure or when the historian is not running when the remote IDAS
computer starts. Using autonomous startup, the IDAS starts caching
store-and-forward data without waiting for a command from the
historian.
For an IDAS to autonomously start, it must be configured to acquire
data from at least one data source. During the configuration process,
the IDAS must be connected to the historian to ensure that the
configuration file is created on the local IDAS computer. An
autonomous startup requires an existing local IDAS configuration file
on the IDAS computer, so that it has all of the information it needs to
begin acquiring data. For more information, see "IDAS Configuration"
on page 81.
An IDAS can start autonomously by being started manually from the
Windows Services console or by starting automatically after a
computer restart.
When an IDAS starts, it attempts to load the configuration
information from the local configuration file. If it is able to do so, the
IDAS uses that information to connect to its data sources and start
acquiring data. As soon as the internal IDAS data buffers are full, the
IDAS switches to store-and-forward mode and stores data to the local
hard drive.
If there is more than one autonomous configuration file on the IDAS
computer (for example, if you deleted an IDAS on a node while it was
disconnected and then added one again), only the newest file is used. A
warning is logged on the IDAS computer.

If the local configuration information cannot be loaded, the IDAS

remains in an idle state until contacted by the historian. If the IDAS is
not contacted by the historian within the default start time-out of 60
seconds, the IDAS shuts down. Note that the IDAS startup time-out is
different than the time-out used by the IDAS during autonomous
startup. Information on changing the default IDAS startup time-out is
provided in a TechNote, which is available from technical support.
When the historian becomes available, data transmission from the
IDAS will be restored. For more information, see "IDAS
Store-and-Forward Capability" on page 84.
Even if the IDAS is configured for autonomous startup, under certain
circumstances it may be started by the Wonderware Historian
Configuration service. The autonomous startup time-out is the time, in
seconds, that an autonomous IDAS waits for configuration commands
when started by the Configuration service before switching to
autonomous mode. The autonomous startup time-out does not apply
when you start the IDAS either manually using the Windows Services
console or by configuring the IDAS to automatically start (autostart)
using the Windows Services console.
If an IDAS is configured as autonomous, the startup type for the
SysDataAcq service are changed to Automatic, and the IDAS starts
every time the IDAS computer is restarted. If the IDAS is then
changed to be non-autonomous, the startup type will be changed back
to Manual. Information on changing this default behavior is provided
in a TechNote, which is available from technical support.
IDAS Late Data Handling

"Late" data arrives at the Wonderware Historian with a timestamp
later than 30 seconds of the current historian time. The storage
subsystem can cater for processing different types of late data from an
I/O Server. For example:
• Data that is late, but is sent in steady stream. For example,
because of communications delays, a topic defined in an I/O Server
might send a stream of data values that is consistently two to three
minutes behind the Wonderware Historian time. Although this
type of data is late, it is handled by the real-time data storage
process if the topic is configured for late data.
• Data that is sent in periodic bursts. For example, a topic defined in
an I/O Server might represent an RTU that sends a block of data
values every few hours. Late data of this type is handled by the
"manual" data storage process.
• Late data must not encompass more than six history blocks at any
given point of time."

The late date option must be enabled for the topic in order for late data
to be processed. If the late data option is not enabled, any data values
from that topic later than 30 seconds behind the Wonderware
Historian are discarded. Any discarded data is reflected in the
SysPerfDataAcqNOutsideRealtime system tag for the IDAS.
For more information on enabling late data, see "Editing Advanced
Information for an IDAS" in Chapter 4, "Configuring Data
Acquisition," in the Wonderware Historian Administration Guide.
Important: Support for late data is not intended to accommodate

time synchronization problems between the IDAS, I/O Servers, and the
historian. It is imperative that you properly synchronize the IDAS, I/O
Servers, and historian time. If the IDAS is sending data in a steady
stream outside of the real-time window, it is likely there is a time
synchronization problem. For more information, see "Time
Synchronization for Data Acquisition" on page 91.
Regarding data throughput, the following rules apply:

• If the late data option has been enabled for a topic and the data is
within the real-time window, the system will support a late data
throughput equal to its real-time throughput capability.
• If the late data option has been enabled for a topic and the data is
outside the real-time window, the system will support a late data
throughput of one percent of its real-time throughput capability.
If you enable late data for a topic, you need to configure the following
two parameters:
• Idle duration. The idle duration is a delay in processing the data
from the topic. For example, an idle delay of 60 seconds means that
data from the topic is cached and only processed by the historian
storage subsystem after no more data has been received from the
topic for at least 60 seconds. By default, the idle duration is set to
60.
The idle duration is important if you anticipate bursts of late data
being sent to the IDAS. The setting you choose depends on your
specific historian implementation and application requirements,
because the higher you set the idle duration, the longer it will take
to see the data, for example, in a historian client application. Also,
if you are trending blocks of late data, there will be no gap (NULL
values) to indicate the end of the data burst. Instead, the trend pen
will "flat line" until the next block of late data starts.
• Processing interval. The processing interval is a safety
precaution. In case the nature of the data is such that the idle
duration is never satisfied, the historian storage subsystem will
process data from the topic at least once every processing interval.
By default, the processing interval is set to twice the idle duration.
The processing interval cannot be set to a value less than the idle
duration.

The processing interval is important if you anticipate a steady stream

of late data being sent from the IDAS. The higher you set the
processing interval, more memory will be used by the historian, but no
more than 32 MB in total. When that limit is reached, all accumulated
data is processed.
Whenever possible, it is better to set up a remote IDAS on the I/O
Server computer to handle the data processing, instead of using the
late data settings for the topic.
If you are using a remote IDAS with several late data topics, the
lowest settings for any topics configured for the IDAS will apply.
All data values received as late data from an autonomous IDAS are
stored in delta mode, regardless of whether a tag is configured for
cyclic storage. Therefore, you should configure the tag to be stored in
delta mode, if you expect the data to be late. Also, if late data is
enabled for a topic, the timestamp for the initial value will never be
overwritten by the storage subsystem, no matter how early it is. This
differs from how real-time data is handled. For real-time data, if the
storage subsystem receives an initial data value that has a timestamp
that is older that the system start time, it will change the timestamp
of the first value to be the current time of the historian.
If you enable late data for an autonomous IDAS, the topic time-out will
automatically be set to 0 and disabled.
Important: If the topic is configured to receive late data, the storage

subsystem never receives disconnect/reconnect values from the IDAS
for tags belonging to that topic. You may see "flat line" on a trend even
if the data source is disconnected from the historian.
Support for Slow and Intermittent Networks

You can configure an IDAS to accommodate data transfer over a slow
and/or intermittent network. A network is considered intermittent if it
fails for short periods (a few seconds at a time) at random intervals. A
slow network can have transfer rates as low as 56Kb/s or periods of
large data transfers that use most of the available bandwidth.
For an intermittent network, you may need to set the minimum
store-and-forward duration to a value higher than the default. This
prevents the IDAS from frequently going in and out of
store-and-forward mode due to the brief periods of network
unavailability. Typically, the default duration is appropriate in all but
the worst cases.

For a slow network, you can make the following adjustments:

• Set the file chunk size to a smaller value. The file chunk is the
block of data that is sent from the IDAS to the Wonderware
Historian at one time. Sending smaller chunks reduces the
network load at any given time. A good indication that the file
chunk size is too big is that when the forwarding operation starts,
the IDAS immediately goes back into store-and-forward mode; the
forwarded chunk is large enough to overload the network, causing
the IDAS to detect a problem and switch into store-and-forward
mode.
• Increase the forwarding delay. The delay is the interval at which
the IDAS sends chunks of the store-and-forward data to the
historian. Increasing the delay further spreads out the network
load for the forwarding operation.
• Increase the autonomous start time-out for IDASs configured to
autonomously start. An indicator that the time-out period is too
small is that when the remote IDAS starts by the Wonderware
Historian Configuration service, it goes into store-and-forward
mode, and then later restores the connection to the historian.
• Increase the connection time-out. One symptom that the time-out
is too small is that the network connection is physically fine, but
you see error messages in the log regarding IDAS time-out
problems.
A remote IDAS can sustain a network interruption as long as there is
disk space available to hold the local store-and-forward data.
I/O Server Redundancy

You can edit an I/O Server definition to include a "failover" I/O Server.
This alternate I/O Server can be installed on the same computer as the
primary IDAS or on another computer. If the network connection
between the primary I/O Server and the IDAS fails, the IDAS
automatically switches to the alternate I/O Server, provided that the
alternate I/O Server is running. The switch may take some short
period of time, and some data points may be lost during the transition.

Redirecting I/O Servers to InTouch HMI Software

When you redirect an I/O Server to InTouch HMI software, you are
specifying to acquire tag values from a particular InTouch node that is
using an I/O Server, instead of acquiring them directly from the I/O
Server. This feature is useful when you need to reduce the loading on
the I/O Server, or if the InTouch node is more accessible on the
network.
When you redirect the I/O Server, the computer name and I/O Server
type will reflect the InTouch node as the I/O Server from which data is
acquired. For example, suppose you were using the a Modicon Modbus
I/O Server on computer "I23238." The application name for the I/O
Server address appeared as \\I23238\modbus. If you redirect this
I/O Server to the InTouch node "InTouchNode1," then the address will
be modified to reflect \\InTouchNode1\view.
Time Synchronization for Data Acquisition

All I/O Servers that support SuiteLink timestamp plant data as it is
acquired. It is important to understand how synchronization is
handled between the timestamps for I/O Server, the computer clock for
the IDAS(s), and the computer clock for the Wonderware Historian(s).
An overview of time synchronization is as follows:
1 If you have multiple historians on your network, you should
synchronize all computer clocks to a single historian designated as
a master time server. You could then synchronize this master
historian to an external time source. Designate the master
historian using the TimeSyncMaster system parameter.
2 Periodically, a historian automatically synchronizes the computer
clock of any remote IDASs to its own computer clock. The IDAS
synchronization is enabled by means of the TimeSyncIODrivers
system parameter.

3 Every hour, an IDAS automatically synchronizes the timestamping

mechanism of any associated I/O Servers with its own computer
clock. This does not actually change the system clocks of any I/O
Server computers. Instead, the difference in the system clocks on
the two computers (I/O server and Historian) are determined, and
a bias is calculated that is then applied to all values from that I/O
server computer. For example, if the historian clock is seven
seconds ahead of the I/O Server computer's clock, SuiteLink adds
seven seconds to every timestamp from the I/O Server. If a topic is
disconnected/reconnected due to a topic time-out or other
communications failure, the I/O Server timestamping is not
updated until the time synchronization interval has passed. You
can change the frequency of the synchronization using the
SuiteLinkTimeSyncInterval system parameter.
The SuiteLink protocol also does some time adjustments to keep
timestamps consistent across nodes. SuiteLink bases this
adjustment on the time difference detected at startup and each
hour. For example, NodeA and NodeB have a time difference of 17
seconds. The I/O Server is on NodeA, and the IDAS is on NodeB
(either local to the historian or a remote IDAS for a historian on
another NodeC). When the I/O Server on NodeA timestamps a
value at 12:00:00.000, it is transmitted to NodeB with an adjusted
timestamp of 12:00:17.000. If the historian is configured to
timestamp at the source, this value is stored with a timestamp of
12:00:17.000. If, instead, the historian is configured to timestamp
at the server, and there is a two-second communications latency,
then the value is stored with a timestamp of 12:00:19.000.
For normal operations on systems with synchronized clocks, there
is no adjustment made by SuiteLink and everything operates as
expected. However, when either the systems are out of synch, or
even were out of synch when SuiteLink communications between
the nodes started, the timestamps will be adjusted. Because of the
way SuiteLink adjusts timestamps, it is easy to produce
misleading results if system tests involve adjusting system clocks
on the systems, because SuiteLink does not immediately update its
time skew.
Note: Time synchronization does not apply to I/O Servers that use
DDE because these servers do not perform timestamping. The time of
the IDAS computer is always used for data coming from DDE from I/O
Servers.
For more information on setting system parameters, see "Editing

System Parameters" in Chapter 9, "Viewing or Changing System-Wide
Properties," in the Wonderware Historian Administration Guide.

The following diagram shows an example of how computers can be

synchronized to a single time:
Historian 2 Historian 1: Historian 3

Master Server IDAS 3
I/O Server 3
IDAS 1 IDAS 2
I/O Server 2
I/O Server 1
For an MDAS-enabled or HCAL-enabled client application, you can

use the net time command (for the Windows operating system) to
synchronize the client computer's clock to your master historian.
A good indicator that a time synchronization problem is occurring is if
the value of the SysDataAcqNOutsideRealtime system tag is high.
This tag indicates how many values received from a particular IDAS
are outside the limits of the real-time window. For more information
on system tags, see "The System Driver and System Tags" on page 44.
For more information on the real-time window, see "About the

Data Acquisition by Means of INSERT and

UPDATE Statements
You can insert or update history data in the Wonderware Historian
extension tables using Transact-SQL INSERT and UPDATE
statements.
For more information, see Chapter 6, "Importing, Inserting, or
Updating History Data," in the Wonderware Historian Administration
Guide.
Data Acquisition from the Historian Client

Access Layer (HCAL)
The Historian Client Access Layer (HCAL) is used by the following
clients to send data to and communicate with the historian:
• Wonderware Application Server 2012 R2 (version 3.6) or later
• Clients built using the Wonderware Historian SDK 2012 R2
(version 11.0) or later
• Other historian components, such as the OLE DB provider and the
replication service
Wonderware Application Server 2012 (version 3.5) or earlier and the
Wonderware Historian SDK version 9.0 use the Manual Data Access
Server (MDAS) and MDAS2 to send data to the Wonderware
Historian. These versions continue to work with Historian 2012 R2.
For more information on MDAS, see "Data Acquisition from the
Manual Data Acquisition Service (MDAS)" on page 98.
For Wonderware Historian 2012 R2 or later, HCAL replaces MDAS
and MDAS2 on the client side for sending data to the Wonderware
Historian.
Important: Wonderware Application Server version 2012 R2 or later

is not backward compatible with versions of Wonderware Historian
earlier than 2012 R2.
The Historian Client Access Point (HCAP) is a process that resides on

the Historian Server node and receives data from HCAL. HCAP
replaces the MDAS Server from versions of Wonderware Historian
earlier than 2012 R2. HCAP gets installed and runs on each
AppEngine object running in an Application Server redundant setup.
By default, HCAL uses port 32568 to communicate with HCAP. This
listening port is used by HCAL on the client node to establish
communications.

Data Acquisition from the Historian Client Access Layer (HCAL)95
Initial HCAL Data Collection without a

Connection to the Historian
HCAL can establish and maintain a connection to one or more
Historians either synchronously or asynchronously. If a disconnection
from the Historian occurs, HCAL attempts to restore the connection.
Tags that are configured using an HCAL client, such as Application
Server or the Historian SDK, do not require a connection to a historian
to be established before local data storage begins. The tags are cached
in memory and their data stored locally until HCAL can establish the
connection. After the tags are configured successfully on the historian,
current data will start to be sent, along with the locally-stored data.
HCAL synchronizes tags on each reconnection with Historian.
Tag configuration can be changed dynamically while data is being sent
for the tag.
Storing Data through HCAL

HCAL can store data from the following categories:
• Original streamed data, which includes real-time data, late data,
and replication data
• Original non-streamed data
• Revision data
For more information on data categories, see "Data Categories" on
page 56.
HCAL uses a local instance of the storage subsystem for
store-and-forward data. For more information on data storage, see
Chapter 5, "Data Storage Subsystem."
HCAL can go into store-and-forward mode even if it is not connected to
the historian. Store-and-forward is only supported for streamed
original data if the server is not available.
Following parameters are used to configure store-and-forward for
HCAL:
• Store-and-forward directory. This is the folder on disk where the
local data is stored.
• Minimum store-and-forward duration. When the network
connection is unstable, HCAL will go into store-and-forward mode
and stay in store-and-forward mode for the duration specified. This
avoids the situation where the client goes into and comes out of
store-and-forward too often due to an intermittent network
connection.

• Minimum free disk space. When the free disk space on the disk
configured for store-and-forward reaches the minimum free disk
space configured, the storage subsystem engine switches to
read-only mode to prevent total disk consumption and periodically
logs error messages in the ArchestrA Logger.
• Bandwidth usage. If an HCAL client will be operating in a shared
network environment, you can limit the bandwidth usage for
HCAL communications to minimize the impact on other
applications. This bandwidth limit applies to all types of data sent
from HCAL. If the limit is set below what is required for the
current load, HCAL goes into store-and-forward mode.
Data Adjustments Performed by HCAL

HCAL performs certain data adjustments prior to the data being sent
to the historian.
• Storage rules are applied as specified in the tag configuration. For
more information about storage rules, see Chapter 5, "Data
Storage Subsystem."
• If an input value's timestamp is in the future relative to the
Historian Server, the timestamp is adjusted by HCAP and a
QualityDetail bit is adjusted. For more information, see
"QualityDetail Bit Layout" on page 65.
• If an input value's timestamp is out of order with the previous
input value, the timestamp is adjusted and a QualityDetail bit is
adjusted. For more information, see "QualityDetail Bit Layout" on
page 65.
• Channel information is injected for streamed original data by
HCAL and HCAP. For more information on the channel status, see
"How Disconnects are Handled for Real-time Data" on page 58.
• If an Application Server AppEngine goes offscan or
disconnects, HCAL injects a manufactured value with the last
value's timestamp and a quality detail of 24.
• When a tag first goes into store-and-forward mode, HCAL
injects a manufactured value with a QualityDetail of 248.
• When a tag comes out of store-and-forward mode, HCAL
injects a manufactured value with a QualityDetail of 44. This
value's timestamp is re-adjusted by the HCAP for real-time
streamed data.
• After the last ping from HCAL exceeds the specified timeout,
HCAP injects a manufactured value with a QualityDetail of 10.

Data Acquisition from the Historian Client Access Layer (HCAL)97
Data Retrieval through HCAL

HCAL is used internally by OLE DB or SDK applications for data
retrieval from the server. The client issues a retrieval request by
calling HCAL with specific information about the query, such as the
names of tags for which to retrieve data, start and end time, retrieval
mode, resolution, and so on.
Data cannot be retrieved from the HCAL store-and-forward cache.
HCAL Security
A connection from an HCAL client is authenticated before replication,
storage, or retrieval tasks can be performed on the historian. HCAP
uses the security credentials supplied by the HCAL client.
To store data, the user must belong to one the following Windows
groups on the historian server computer:
• aaPowerUsers
To retrieve data, the user must belong to one of the following Windows
groups on the historian server computer:
• aaPowerUsers
• aaUsers
To replicate data, the user must belong to the following Windows
group on the historian server computer:
By default, the ArchestrA user specified during installation is included
in the aaAdministrators and aaReplicationUsers groups. If more than
one computer is included in the setup and each has different
ArchestrA users, you will need to manually add the remote account
that HCAL is using to the appropriate group on the Historian node.
Integrated Windows authentication is currently not supported.

Data Acquisition from the Manual Data

Acquisition Service (MDAS)
The manual data acquisition service (MDAS) accepts data (real-time
data, as well as inserts and updates) from its host and forwards it to
the Wonderware Historian storage subsystem. For example,
Application Server uses MDAS to send history data to the historian.
Replicated data from a tier-1 historian is sent to the historian using
MDAS.
The storage subsystem merges data acquired from MDAS to existing
historized data. All data can be accessed from the History,
WideHistory, and Live extension tables.
MDAS is implemented in two ways within the system: one as a
client-side Windows DLL and one as a Windows service.
The DLL version of MDAS uses DCOM and file shares to send data to
the historian. For both the MDAS and Wonderware Historian
computers, make sure that DCOM is enabled (not blocked) and that
TCP/UDP port 135 is accessible. The port may not be accessible if
DCOM is disabled on either of the computers or if there is a firewall
between the two computers that blocks the port. For information on
enabling DCOM communication through a firewall, see your Microsoft
Windows operating system documentation.
The Windows service version of MDAS is used only for the replication
subsystem. The MDAS service manages data transmission between
tier-1 and tier-2 servers. The Windows service version of MDAS
communicates using TCP-based communication layer through an open
port that you specify when you set up the replication server. The
MDAS service also handles security-related error reporting on the
tier-2 historian.
Note: Data acquired through the manual data acquisition service is

NOT stored in the ManualAnalogHistory, ManualDiscreteHistory, or
ManualStringHistory tables. Prior to Wonderware Historian version 8.0,
these tables were the only mechanism to store manually acquired data.
Currently, these tables are provided for backward compatibility only.

Importing Data from a CSV File99
Importing Data from a CSV File

Legacy data can be batch imported into the Wonderware Historian
from a CSV file. A batch import allows you to import legacy data from
other historians. If you want to import history data from an InTouch
application, you can use the InTouch History Importer, which converts
InTouch history files to the required CSV format. Imported data is
integrated with data currently stored in history blocks, providing you
with seamless access to all your data.
For more information, see Chapter 6, "Importing, Inserting, or
Updating History Data," in the Wonderware Historian Administration
Guide.


101
Chapter 5
Data Storage Subsystem
The Wonderware Historian storage subsystem saves plant data from

various sources to disk. The storage subsystem stores data for analog,
analog summary, discrete, state summary, string, and system tags in
sets of files on disk called history blocks.
For more information on types of data, see "Data Categories" on
page 56.
Historical data can be retrieved by sending SQL queries through the
Wonderware Historian OLE DB provider, which is part of the data
retrieval subsystem. At retrieval, the historized tag data is presented
as if it resided in SQL Server tables. For more information, see
Chapter 7, "Data Retrieval Subsystem."
The storage subsystem processes only historical data; it does not
handle configuration data. For more information about saving
configuration data, see Chapter 3, "Configuration Subsystem."
Also, the storage subsystem does not handle event data, including
summarized data created by the event subsystem. For more
information about event data, see Chapter 11, "Event Subsystem."
You use the System Management Console to configure all aspects of
the storage subsystem. For more information, see Chapter 5,
"Managing Data Storage," in the Wonderware Historian

102  Chapter 5 Data Storage Subsystem
Storage Subsystem Processes

The storage subsystem runs in a process named aahStorage.exe, which
is not a system service, but a child process of the Historian
Configuration Service. The same executable is used as a
store-and-forward storage engine for storing data locally before
forwarding it to the destination historian server, so in the Windows
Task Manager you can see several instances of the aahStorage.exe
process that are different by their command lines.
The storage subsystem is available in 32 and 64-bit versions. If you
install Wonderware Historian on a 64-bit operating system, the
storage subsystem always runs as native 64-bit application. However,
if a Wonderware Application Server engine was deployed in a 32-bit
mode, the store-and-forward storage engine on that computer will also
be 32-bit if there was no prior historian installation on the computer.
The storage subsystem also includes the Metadata Server process
(aahMetadataServer.exe) responsible for caching Storage tag
metadata persisted in the Runtime database.
The classic storage subsystem consists of two system services
aahStoreSvc.exe and aahManStSvc.exe that function as they did in
Wonderware Historian 10.0 and earlier versions. For more
information, see "Classic Storage Subsystem Processes" on page 126.
Storage Modes
When you define a tag, you need to specify a storage method, which
specifies how the tag’s data is saved as historical records. For example,
the system may be receiving 100 values per millisecond for an I/O
Server tag, but you might not want to store all of these values. How
you configure the storage method affects the number of values written
to disk and the resolution of the data that will be available later for
retrieval.
The following types of storage modes are available:
• No data values are stored.
• All data values are stored (forced storage).
• Only changed data values are stored (delta storage).
• Only data values that occur at a specified time interval are stored
(cyclic storage).

"Forced" Storage103
"Forced" Storage
Forced storage is the storage of each value as it comes in from the
plant floor or from a client application, without filtering applied (that
is, no delta or cyclic storage mode is used).
Forced storage is useful, for example, if you have an I/O Server that
collects data by exception from the plant floor device (instead of using
a polling interval). You can allow the I/O Server to filter the values by
exception, instead of the Wonderware Historian.
If the I/O Server uses a polling interval, then storage with no filtering
is similar to cyclic storage. However, the cyclic storage rate determines
the values that are stored, and that rate can only go down to a one
second resolution. If you use storage with no filtering, the I/O Server
controls which values are stored, and these can be at a finer resolution
than one second. For forced storage, if the data source is sending the
same values at each internal scan cycle, then each value is stored.
Delta Storage
The Delta storage mode stores data based on a change in a value.
Delta storage writes a historical record only if the current value
changes from the previous value. Delta storage is also called "storage
by exception." Delta storage is typically used to store discrete values,
string values, and analog values that remain constant for long periods
of time. For example, you don’t want to store the value of the discrete
tag "PUMPON" every ten seconds if the pump usually stays on for
months at a time. The value is stored along with a timestamp of when
the change occurred, to an accuracy of 1 ms.
The following types of deadbands can be applied for delta storage:
• Time deadband
• Value deadband
• Rate of change (swinging door) deadband
Note: You can retrieve down to the millisecond by using a CONVERT

clause in an OPENQUERY statement with specific style specifiers of the
Wonderware Historian. For more information, see "Querying Data to a
Millisecond Resolution using SQL Server 2005" on page 318.

Time and Value Deadbands for Delta Storage

To further decrease the resolution of tag values stored in delta mode,
use a time deadband or a value deadband.
• A time deadband is the minimum time, in milliseconds, between
stored values for a single tag. Any value changes that occur within
the time deadband are not stored. The time deadband applies to
delta storage only. A time deadband of 0 indicates that the system
will store the value of the tag each time it changes.
• A value deadband is the percentage of the difference between the
minimum and maximum engineering units for the tag. Any data
values that change less than the specified deadband are not stored.
The value deadband applies to delta storage only. A value of 0
indicates that a value deadband will not be applied.
"Swinging Door" Deadband for Delta Storage

A "swinging door" deadband is the percentage of deviation in the
full-scale value range for an analog tag. The swinging door (rate)
deadband applies to delta storage only. Time and/or value deadbands
can be used in addition to the swinging door deadband. Any value
greater than 0 can be used for the deadband. A value of 0 indicates
that a swinging door deadband will not be applied.
The swinging door deadband is essentially a rate of change deadband,
based on changes in the slope of the incoming data values. For
example, specifying a swinging door deadband value of 10 percent
means that points will be stored if the percentage change in slope of
the consecutive data values exceeds 10 percent. The percentage of
allowable "swing" in the data values gives this type of deadband its
name.
&
2
5
LPDJLQDU\
GDWD
VZLQJLQJ
SRLQW
GRRU
9DOXH
5
2
&
7LPH

Delta Storage105
Benefits of the Swinging Door Deadband

One benefit of using a swinging door deadband is that it reduces the
disk space required to store data. Another benefit of the swinging door
deadband is that it captures the data value before the rate change,
which is something that a value deadband does not do. If you trend
data, the peaks and valleys of the trend curve are more defined to
provide a more accurate picture of what is happening in your plant.
Generally, using a swinging door (rate) deadband provides better
representation of the value change curve with the same or less number
of values stored than regular value or time deadbands for delta
storage.
The following graphics compare the trend curves of the same raw data,
but with different deadbands applied.
The following graph shows the trend of the actual raw data values:
10 11
9 12
13
14
8
15
7
16
6
18 19
4 17
1 2 3 5

The following graph shows the trend of the data values with a value
deadband applied. Notice how only the first data value that deviates
by the deadband from the previous value will be stored, and not any of
the values between the starting value and the first deviating value.
9 10 11 12
13
14
8
15
7
16
6
18 19
4 5 17
2 3
1
Value Deadband
The following graph shows the data values that will be stored for both
a value deadband and a swinging door deadband. Notice how the
swinging door deadband captures data before the deadband change,
allowing for a more complete view of the data.
.
9 10 11
12 13
14
8
15
7
16
6
18
4 19
2 3 5 17
1
Value Deadband
Swinging Door

A swinging door deadband is most useful for tags that have a steady
increase and decrease in slope, such as a tank level or tank
temperature that rises and falls. A swinging door deadband may not
be appropriate for "noisy" signals, in which the value of the tag
constantly fluctuates around a certain point for long periods of time.
Also, the reduction in storage requirements offered by the swinging
door deadband may not have much of an impact if you have an
application with a small tag count (for example, 500 tags). In this case,
it may not be necessary to use a deadband at all.
A swinging door deadband is applicable for analog tags that receive
data from the following sources:
• Real-time data values from I/O Servers, MDAS, or HCAL
• Store-and-forward data from a remote IDAS
• Late data from an I/O Server topic that was configured for late
data
• A "fast load" .CSV import
• Real-time inserts of data using a Transact-SQL statement
A swinging door deadband is not applicable for manual inserts of data
through a .CSV import of a Transact-SQL statement.
To best visualize the tag that uses swinging door storage, plot a trend
using the Historian Client Trend application and set the plot type from
to "line" (rather than "step-line").
Additional Options that Affect the Swinging Door

Deadband
The swinging door deadband (the rate deadband) can optionally be
combined with a value deadband and/or a deadband override period,
the combination of which will affect which values are actually stored.
If a tag is configured to be stored by classic storage, the behavior of the
swinging door algorithm also depends on the value of the real-time
window in the Wonderware Historian, as specified by the
RealTimeWindow system parameter.
• Value deadband
When combined with rate deadband (with or without a deadband
override period), the value deadband is always applied first,
followed by the other deadbands. For the value deadband, the
system checks the difference in value between the received point
from the value of the last stored point. Only when this difference
exceeds the value deadband does the system consider the point for
rate evaluation.

• Deadband "override" period

If the elapsed time since the last stored point exceeds the
deadband override period, the last received point before the time at
which the deadband override period expired is stored, regardless of
value and rate deadband.
• Real-time window
The real-time window setting for classic storage (RealTimeWindow
system parameter) allows for the expansion of the time window for
which the storage system considers data to be "real-time." The
real-time window is important for swinging door deadbanding
because it determines the maximum length of time that a point
will be "held" by the storage system without storing it, while
waiting for the next point. For more information, see "About the
Real-time window and deadband override periods are two
independent modifiers that force the storage of received points that
may have otherwise been discarded due to the setting of either the
rate deadband or the value deadband.
• The real-time window specification is more likely to select points
for storage when the time period between points received from the
source is less than the real-time window, but the slope of the
incoming data values is such that the rate deadband excludes the
points from being stored.
• The deadband override period is more likely to select points for
storage if the rate at which points are received from the data
source is slow (slower than the real-time window) and the rate
deadband excludes the points from being stored.
For an illustration of how these factors work together to determine the
actual values to be stored, see "Swinging Door Deadband Examples"
on page 108.
Whatever the combination of rate deadband, value deadband, and
deadband override period specified, only points actually received from
the data source are stored on disk. That is, points to be stored on disk
are never "manufactured" by the swinging door algorithm. This is
particularly relevant in understanding the behavior implied by
specifying the real-time window and the deadband override period.
Swinging Door Deadband Examples

The following examples illustrate the effects of the different swinging
door options.

All of the examples are based on the following raw data. The
numbered points represent actual values received from a data source.
10 18
9
17
0
8 16
7 15
6
14
1 2 3 4 5 11 12 13 20
19
Swinging Door Deadband: Rate Only

The following diagram depicts an ideal case, where the incoming
signal is noise-free and with a proper rate deadband specification only
(no value deadband or deadband override period).
18
10
9
17
0
8 16
7 Real time 15
window +30.0
seconds
6
14
1 2 3 4 5
11 12 13 19 20
Assume point 0 has been stored on disk. The system waits for point 2
to arrive before making its next storage decision. When point 2 is
received, the storage engine calculates the change in slope as follows:
Slope0_1 is considered the base slope, and Slope1_2 is considered
the current slope.
Slope0_1 = (Value1 - Value0) / (Time1 - Time0)
Slope1_2 = (Value2 - Value1) / (Time2 - Time1)
Slope_Change_Percent = 100* | (Slope1_2 - Slope0_1) / Slope0_1 |
If
Slope_Change_Percent > Rate_Deadband_Specified

In other words, if the percentage change in slope is greater than the

specified rate deadband, the storage engine goes ahead and stores
point 1 on disk. Next, it receives point 3. The base slope for point 2 will
be the slope between points 1 and 2. The current slope will be the slope
between points 2 and 3 only if point 1 was stored. If point 1 was not
stored, then the base slope for point 2 will be the slope between points
0 and 1, and the current slope will be the slope between points 2 and 3.
The base slope for an evaluation point is not changed unless the
previous point is stored; otherwise, the base slope will be the last
known current slope that caused a point to be stored on disk.
Assuming point 1 is stored, because the slope between points 2 and 3 is
about the same as the slope between points 1 and 2, the rate deadband
criterion is not satisfied, and point 2 is discarded. When point 4 is
received, the slope change calculation results in point 3 being
discarded, and so on until point 6 arrives. Now the rate deadband
criterion is satisfied (slope change between points 5 and 6 and points 1
and 2 is greater than the rate deadband specified), and point 5 is
stored on disk.
The arrival of point 7, likewise, discards point 6 even though the
actual slope between point 6 and point 7 may be quite high, and may
even be higher than the rate deadband specified, it is not sufficiently
different from the slope between points 5 and 6 to qualify point 6 to be
stored. Following this logic through until point 12 is received results in
the storage on disk of points 10 and 11, discarding all the other points
in between.
Point 13 illustrates the effect of the real-time window setting. Under
normal circumstances, point 12 would not qualify to be stored. If,
however, the elapsed time between receiving point 12 and point 13
exceeds the time window in which the storage engine is able to store
point 12 as a real-time point, point 12 is stored anyway, and the value
of the SysRateDeadbandForcedValues system tag is incremented. In
other words, if, while the system waits for point 13 to arrive, the
timestamp of point 12 becomes so old that it reaches the limit for the
real-time window, point 12 is stored regardless of whether it is outside
the deadband.
The SysRateDeadbandForcedValues system tag counts the number of
"extra" points stored as a result of an insufficient real-time window for
swinging door storage.
When point 14 arrives, the base slope for evaluating point 13 is
between points 11 and 12, and not between points 12 and 13, because
point 12 was stored due to the real-time window expiration. A point
stored due to the real-time window does not re-establish the base
slope; only points stored due to exceeding the rate change causes the
base slope to be re-established. Then "normal" rate change evaluation
resumes, resulting in point 13 being stored, and so on.

Swinging Door Deadband: Rate and Value

In the following diagram, a signal with some "noise" is shown. The
effect of applying both a rate and value deadband to swinging door
storage is illustrated. The value deadband is indicated by two
horizontal dashed lines.
15
14
1
16
13
3 5 7 9 11
Value
deadband
2 4 6 8 10 12 17
Assume that point 1 has been stored to disk. Point 3 passes the value
deadband check, allowing points 2 and 3 to be evaluated for rate
change. Assuming that the point exceeds the rate change requirement,
then point 2 is stored. Until point 13 is received, all intermediate
points are discarded by the value deadband filter. In this example, it is
assumed that the change in slope between points 2 through 3 and
points 12 through 13 is greater than the rate deadband, so point 12 is
stored on disk. When point 14 is received, the normal operation begins.
If a rate deadband is applied without a value deadband, all of the
"noisy" points (3 through 11) would have been stored, because the
slope of the signal changes radically between successive points. The
value deadband removes the noise, but also introduces some amount of
distortion in the resultant signal.

Swinging Door Deadband: Rate, Value, and

Deadband Override Period
The following graphic illustrates the effect of a rate deadband
combined with a value deadband and a deadband override period.
15
14
1
16
Deadband
override
period
13
3 5 7 9 11
Value
deadband
2 4 6 8 10 12 17
Assume point 1 is stored to disk. Point 3 makes it through the value

deadband check, allowing points 2 and 3 to be evaluated for rate
change. Assuming the point exceeds the rate change requirement,
then point 2 is stored.
Adding a value deadband alone could result in distortion of the stored
data.
For example, suppose that the rate deadband is specified such that
point 12 does not get stored. That is, the change in slope between
points 2 through 3 and points 12 through 13 is not greater than the
rate deadband. In that case, the data representation (points 1, 2, and
15) is grossly distorted because the value deadband is discarding key
points.
To allow for better representation, a deadband override period may
optionally be specified. If the elapsed time between the last stored
point and the currently received point is more than the specified
deadband, then the point immediately prior to the currently received
point is stored. In this example, the elapsed time between point 2 and
point 10 is more than the deadband, so point 9 is stored. The data
actually stored to disk (points 1, 2, 9, and 15) is a better approximation
of the original data.
It is important to note that after point 9 is stored, subsequent rate
calculations use the slope between points 2 and 3 as the baseline for
subsequent storage decisions because point 2 was last point that was
stored normally by storage.
The deadband override period can have any value and is not related to
the real-time window value.

Cyclic Storage113
Cyclic Storage
Cyclic storage is the storing of analog data based on a time interval.
Cyclic storage writes a record to history at the specified interval, only
if a data changes during the time interval. For example, you could
store the value of an analog tag every five seconds.
The time interval for cyclic storage is called the storage rate. Each
analog tag has its own storage rate. The storage rate you should select
depends on the minimum timestamp resolution with which you want
to be able to retrieve the data sometime in the future. Storing data
using a very low time interval will result in a very high resolution of
data being stored over a small time span. Storing data using a very
high time interval, however, may result in significant data values
being missed. An exception to this will be values received or generated
due to a connect or disconnect event.
Available storage rates for analog tags are:
• 1, 2, 3, 5, 6, 10, 15, 30 seconds
• 1, 2, 3, 5, 6, 10, 15, 20, 30 minutes
• 1 hour
The timestamp for the first data value received during the cyclic time
span will be used. For example, you might have specified a 10 second
storage rate for a particular analog tag. The last data value received
during the 10 second lapse will be stored, along with the corresponding
timestamp.
The storage subsystem physically stores cyclically stored values in the
same manner as it does value stored by delta. That is, it keeps track of
repeated values in time in a logical manner and will "fill in" the
missing values upon retrieval. For example:

Data Conversions
The Wonderware Historian can store values up to 32-bits. If the data
source sends a 64-bit real number to Wonderware Historian, the
number is converted to a 32-bit real number and, consequently, loses
precision to six or seven decimal places. (The range depends on the
number of binary digits, which does not exactly map to a fixed number
of decimal places.) As a result, values that differ in the data source
system beyond the seventh digit will not register as a change for delta
storage and will not be re-stored. For example, a value of 1.0449991
followed by a 1.0450001 would both round to 1.045000, and the second
point would not be stored.
The maximum length of a string value that can be stored is 513
characters. Any characters over this limit are truncated. This limit
does not apply to variable-length strings.
History Blocks
The Wonderware Historian stores data in files of a proprietary file
format in units called history blocks. A history block is essentially a
sub-folder of the main historian data storage folder, defined during
installation or by subsequent dynamic configuration changes. A
history block is self-contained, containing all the information
necessary to retrieve data for the period represented by the history
block. You specify the duration of history blocks. The default duration
of a history block is one day, and the minimum allowed duration is one
hour. The historian automatically creates a new history block at
system startup, at scheduled block changeover times, at your request,
or in response to certain dynamic configuration actions.
Note: Configuration data and event history are not stored in the
history blocks; they are stored in the Runtime database file.
As data is acquired from the plant, the size of these history blocks
grows on a continual basis, being limited only by the size of the hard
disk on which the historian resides.
By storing plant data in the history blocks instead of in normal SQL
Server tables, the historian can store data in a fraction of the space
that would be required by a normal relational database. Compact
storage formats reduce the storage space requirements than would be
required in a standard relational database. Upon retreival, historical
data is presented by the Wonderware Historian OLE DB provider as if
it were stored in SQL Server tables.

History Blocks115
History Block Notation

Each history block is contained in a single sub-directory in the circular
storage directory. The sub-directory name includes the date stamp of
the Wonderware Historian computer at the time the block was
created. Also, a numerical suffix is used to differentiate blocks of
historical data created on the same day as a result of, for example,
multiple restarts of the system, or where the block time interval
configured is less than one day. The block duration can be configured
by changing the HoursPerBlock system parameter. For more
information on editing system parameters, see Editing System
Parameters in Chapter 9, Viewing or Changing System-Wide
Properties, in the Wonderware Historian Administration Guide.
In the following example, the "010912" portion of the directory name is
the date that the history block was created. The "_001" portion is the
numerical suffix that identifies this history block as the first block
created that day.
EORFNFUHDWLRQGDWH
<<00''
$B
V\VWHPXVH EORFNQXPEHU
By default, history blocks are 24 hours in size. In the following

example, the system was restarted on September 12, 2012, resulting in
two blocks existing for that date. The two blocks together span the 24
hours for that period.

History Block Creation

A single history block is created at system startup. After that, new
history blocks are automatically created upon expiration of a
designated time interval as stipulated in the database. The default
value is 24 hours. You can change the time interval using the
HoursPerBlock system parameter. For more information on editing
system parameters, see "Editing System Parameters" in Chapter 9,
You can manually start a new history block at any time by executing a
menu command from within the System Management Console. For
more information, see "Starting a New History Block" in Chapter 5,
"Managing Data Storage," in the Wonderware Historian
A new block is also created automatically if any of the storage data
files becomes larger than 1.5 GB.
Note: The system supports 999 history blocks per day. If this limit is
reached, the system begins overwriting data in the first block for the
day. A warning message will be issued in the Wonderware Historian
Console when the limit is about to be reached. To reduce the number of
history blocks created per day, increase your standard history block
size (if it is less than 24 hours).
History Block Storage Locations

There are four types of storage locations for history blocks: circular,
alternate, buffer, and permanent. The paths to the circular, buffer,
and permanent storage locations are initially defined during
installation. The alternate storage location can be defined later using
the System Management Console.
Certain restrictions apply when specifying a path to the storage
location. The circular storage location must be a local drive on the
server machine, and the path must be specified using normal drive
letter notation (for example, c:\Historian\Data\Circular). While the
alternate, buffer, and permanent storage locations can be anywhere on
the network, it is strongly recommended to have the alternate storage
location configured on a dedicated physical drive locally attached by a
high-speed interface to the Historian Server or configured to be on a
different internal hard drive. If you use a network location, then the
ArchestrA user must have full access to the network location. The
locations must be specified using UNC notation. Mapped drives are
not supported.
When planning your storage strategy, be sure to allow enough disk
space for storing your plant data for the required length of time.

Circular Storage Location

Circular storage is used for the main historical data storage. When the
storage subsystem starts for the first time, the first history block is
created, and data starts being written to that history block. The block
of historical plant data is saved as a subdirectory in the circular
storage directory.
The circular storage location consists of a single location, written to in
a "circular buffer" fashion. When the free disk space on the disk
containing the circular storage location drops below a minimum
threshold or when the data is of a specified age, the oldest data is
deleted out of this storage location and replaced with the new data to
be stored. You can also limit the size of the circular storage location.
When the contents of the circular storage location reach or exceed this
limit, the oldest data will be deleted. "Automatic Deletion of History
Blocks" on page 118.
Instead of data being deleted from the circular storage location, it can
be moved into the alternate storage location, if this location is defined.
As long as the free disk space for the circular storage location is below
the threshold, the oldest data will be continuously migrated to the
alternate storage location to make room for the new historical data.
It is the responsibility of the system administrator to monitor disk
space and back up history blocks to storage media (such as DAT tape)
on a periodic basis.
Alternate Storage Location

When the free disk space in the circular storage location goes below
the defined threshold, the circular directory exceeds the specified
maximum size, or the blocks reach a certain age, the storage
subsystem will start moving the oldest history blocks to the alternate
locations, if configured.
History blocks in the alternate storage location are managed in the
same way as the blocks in the circular storage location. However,
blocks will not be deleted based on age until the sum of the specified
ages for both the circular and alternate storage has passed.
If the alternate storage location reaches its deletion threshold limit, or
configured maximum size, or age, the oldest history blocks are deleted.
A physical drive is strongly recommended. This storage location is
optional.

Permanent Storage Locations

Permanent storage locations are used to store critical data (for
example, reactor trips) that must not be overwritten. The storage
subsystem will never attempt to delete data in this location. Data in a
permanent storage location can be accessed and viewed along with the
data stored in the circular storage location.
Buffer Storage Locations

Buffer locations are used for temporary purposes, such as retrieval
from a data archive. This storage location can reside on the same hard
disk as the circular storage location or on a different disk. Data stored
in the buffer storage location can be accessed and viewed along with
the data stored in the circular storage location. Data is never deleted
from this location by the storage subsystem.
Automatic Deletion of History Blocks

History blocks in the circular and alternate storage locations may be
automatically deleted to make room for new history blocks. Whether or
not the blocks are deleted is determined by the minimum threshold
and the maximum size and/or age specified for the storage location.
The history block management system will check for available space in
the circular and alternate locations if it detects any changes made by
other subsystems or the user in controlled directories. If the real-time
storage service is running, this check is performed every 20 seconds.
The history block management system computes the sum of the sizes
of all history blocks (including the current one) in the circular storage
location and determines if there is enough space on the drive to hold
all of the blocks.
If the space available on the storage location drive is below a certain
threshold, the storage subsystem will delete enough of the oldest
history blocks to bring the available disk drive space back to a positive
value and then move the new history block in.
If an alternate storage location exists, the older block(s) will be moved
there instead of being deleted. The alternate storage location functions
exactly like the circular storage location. However, when the blocks
exceed the set limits (minimum threshold, maximum size, or
maximum age), the oldest blocks will be deleted from disk.
To avoid this loss of data, it is important that the system
administrator regularly monitors the disk availability and periodically
backs up old blocks to long term storage before they are deleted.

For example, a history block is stored in the circular storage location.

The maximum size of the disk drive for circular storage is 120 MB. In
addition to the circular storage location, an alternate location with a
maximum disk drive size of 120 MB is defined. For both locations, the
minimum threshold value is 50 MB. Essentially, this means that there
is 70 MB of actual storage space.
Note: The sizes in this example are purposely small; the disk drives
for storage locations should be much larger.
You should typically set the minimum threshold to a value that is 3

times larger than the size of the biggest history block. This will
provide the history block management system enough time to copy
oldest history block from the circular location to the alternate, and
then delete block from the circular location.
If you monitor the disk drive space available in the circular or
alternate storage location over time, the value will fluctuate between
the threshold value and the maximum size of the location, with sharp
increases when blocks are moved out. While the system is moving a
block(s) out, the space available will dip just below the threshold value
before the increase.
If the maximum threshold is reached before the age of the block
reaches the specified limit, the block is moved or deleted. A block will
be moved or deleted within one history block duration of it reaching
the age limit. If, for any reason, the system is unable to move a block
that is past the age limit, the block will not be deleted until the size or
space limit is reached.

Storage and Classic Storage

The storage system introduced in Wonderware Historian 2012 R2 is
capable of higher sustained data rates and much higher tag counts,
and processing large volumes of data with timestamps significantly
behind the current system time. This storage system allows you to
implement production systems of higher tag counts and higher data
rates with better retrieval performance and communication flexibility
that was previously possible.
The storage subsystem from Historian 10.0 and earlier remains in
place and is called “classic” storage. All historical data collected by the
classic storage subsystem remains fully accessible. The Wonderware
Historian 2014 storage subsystem and its associated communication
layer are backward-compatible, while at the same time are faster,
more scalable, easier to use, and consume less system resources than
the 10.0 version.
The Wonderware Historian storage subsystem is based on the
replication architecture and data model and offers the following
benefits over classic storage:
• Native 64-bit execution on a 64-bit operating system, thus allowing
use of more than 2 GB of memory
• Utilization of multicore hardware when executing various actions
in parallel, such as importing forwarded snapshots and processing
CSV files and late data
• Data retrieval from snapshots are not flushed to disk, eliminating
the need for the active image
• Data filtering on the client side to reduce unnecessary data
transfer
• Support for adding new tags and reconfiguring existing ones
without waiting for a switch to a new history block
• Reduced retrieval response time
• Storing data equally fast regardless of its proximity to the current
system time, thus eliminating the need for the real-time window
concept
Note that timestamping at the Historian Server is not supported for
non-classic storage tags.
Important: In Wonderware Historian, data values cannot be stored

with their original timestamps if those timestamps exceed the current
system time of the historian computer. Such timestamps will be
overwritten by the current system time and the corresponding
QualityDetail set accordingly.

Storage and Classic Storage121
Identifying Which Tags are Stored by Which

Storage Subsystem
The storage subsystem stores data for the following tags:
• Tier-2 tags
• Tags created by Wonderware Application Server 2012 R2 or later
• Tags created by Wonderware Historian SDK 2012 R2 or later
• Manual tags created using the Wonderware Historian 2012 R2
Configuration Editor
• Manual tags created using Wonderware Historian 2012 R2 stored
procedures where the AITag parameter is set to 0
• Manual tags created using previous versions of Wonderware
Historian and that are manually upgraded. For more information
about upgrading tags, see "Upgrading Tags from Classic Storage"
on page 121.
The classic storage subsystem stores data for the following tags:
• System tags
• I/O tags, that is, belonging to local or remote IDASes
• Tags created by Wonderware Application Server 2012 and earlier
versions, where Wonderware Application Server has not been
upgraded
• Tags created using the Wonderware Historian SDK 9.0
The AITag column of the Tag table identifies which storage system
handles a particular tag, where “AI” stands for the active image. The
Wonderware Historian 2012 R2 storage subsystem does not use the
active image, so:
• AITag = 0 means that the tag data is stored by the storage
subsystem
• AITag = 1 means that the tag data is stored by the classic storage
subsystem
Upgrading Tags from Classic Storage

Some tags that were originally created as classic storage tags can be
upgraded to storage system tags to take advantage of improved
scalability and more flexible data acquisition. When you upgrade
Wonderware Application Server from version 2012 or earlier to 2012
R2 or later, its tags will be automatically upgraded. Manually created
tags (such as those used for data imported from a CSV file), must also
be manually upgraded.

WARNING! Downgrading tags back from storage (AITag=0) to

classic storage (AITag=1) is not supported.
When tags are upgraded from classic storage, all tag data stored prior
to the upgrade remains in the classic storage subsystem, and any new
data values after the upgrade are stored in the storage subsystem.
The AIHistory column of the Tag table indicates whether data exists
for a tag in both storage and classic storage:
• AIHistory = 0 means that there is no data was previously collected
by classic storage
• AIHistory = 1 means that the tag may have data previously
collected by classic storage
If you query data for some tag where the query time interval goes
across the time of the tag upgrade, the retrieval subsystem fetches the
data from both storage and classic storage subsystems and presents it
in a seamless manner.
Consider the following steps of an upgrade for a system that uses both
the Wonderware Historian and Wonderware Application Server:
1 The Wonderware Historian version 10.0 is collecting data from
Wonderware Application Server 2012 or earlier
2 The Wonderware Historian is stopped and upgraded to version
2012 R2. The Application Server engines continue collecting data
in store-and-forward mode.
3 The Wonderware Historian is started. The Application Server
engines reconnect to the historian and continue storing data.
At this point, Application Server is storing data in the classic
storage subsystem. Both the AITag and AIHistory column values
will be 1 for all Application Server tags stored in the historian.

Storage and Classic Storage123
4 When the Application Server is upgraded to 2012 R2 or later, the

engines are undeployed and then redeployed.
5 When the Application Server engines start up, they automatically
initiate the historian tag upgrade, after which all Application
Server tags stored on the historian are upgraded from the classic
storage subsystem. The AITag column value is changed to 0, but
the AIHistory column value remains as 1, because those tags were
upgraded and data still exists in classic storage.
Another supported upgrade scenario is where you want to upgrade
tags that were created manually and the associated data was
manually imported using .csv files or Transact-SQL statements. You
can upgrade the tags from classic storage by executing an UPDATE
statement that sets AITag=0 for the tags. Within 30 seconds the
historian should detect that there are no more changes being made
and perform an automatic commit of the tag configuration change.
After the commit completes, the data from .csv files and
INSERT/UPDATE statements referencing those tags are handled by
the storage subsystem.
All new manual tags are assigned to the storage subsystem
automatically by default.


125
Chapter 6
Classic Data Storage

Subsystem
Classic storage is used for data coming from the following:

• An IDAS or a Historian System Driver
• Wonderware Application Server 2012 (version 3.5) or earlier
• Applications that use the Wonderware Historian SDK version 9.0
or earlier

126  Chapter 6 Classic Data Storage Subsystem
Differences between Storage and Classic

Storage
If the AITag column in the Tag table set to 1, values for the tag are
stored by classic storage. The following table describes the differences
between storage and classic storage.
Applies to Applies to
Storage Classic Storage
Storage data categories Yes. Yes.

Real-time window No. Yes.
Data revisions and versioning Yes. Yes.
Storage modes Yes. Yes.
History blocks Yes. Yes.
Storage locations Yes. Yes.
Reserved values No. Yes.
Active image No. Yes.
Dynamic configuration effects on No. Yes.
storage
Memory management No. Yes.
Classic Storage Subsystem Processes

The classic storage subsystem includes the Realtime Data Storage
Service (aahStoreSvc.exe), which is an internal process that stores
real-time data to disk. This process runs as a Windows service.
Manual Data Storage Service (aahManStSvc.exe) is an internal
process that processes non-real-time data and stores it to disk. This
process runs as a Windows service.
The Realtime Data Storage Service (aahStoreSvc.exe) and the Manual
Data Storage Service (aahManStSvc.exe) work together to store all of
data to disk, organize the data in such a way that, upon retrieval, the
data is as seamless and integrated as possible.
Late data can be processed by either the Realtime Data Storage
Service (aahStoreSvc.exe) or the Manual Data Storage Service
(aahManStSvc.exe), depending on whether or not the timestamps fall
within the block of data that is currently being processed. Data is
always processed by Realtime Data Storage Service (aahStoreSvc.exe)
first.

About the Real-Time Data Window for Classic Storage127
About the Real-Time Data Window for Classic

Storage
The real-time "window" is the maximum delay, relative to current time
of the server, in which data is considered real-time by class storage.
The real-time window can range from -30 seconds to +999 milliseconds
of the current server time.
• For real-time data, the timestamp for the data value must fall
within the time window.
• For late data, the timestamp for the data value can fall either
inside or outside of the time window, depending on whether the
late date setting is enabled for the topic. Note that late data can be
processed very slowly if the timestamp falls outside of the
real-time window.
• For non-streamed data, the data can have any timestamp.
The following rules apply when storing data with timestamps relative
to the real-time window:
• If the late data setting for a tag topic is not enabled and the
received data is more than 30 seconds late, the value is discarded,
and a warning is logged. If the received data value is within 30
seconds of the server time, it is stored as received.
• If the late data setting for a tag topic is enabled and the received
data value is within the real-time window, then the value is stored
by the real-time storage service with no changes. If the received
data value is outside of the real-time window, then the value is
passed to the alternate storage services and stored without
changes.
• If the late data setting for a tag topic is enabled, the received data
value is stored in delta mode, even if the tag is configured for cyclic
storage and the received data value is within the real-time
window.
For more information on the late data setting for a topic, see "IDAS
Late Data Handling" on page 87.
You can adjust the real-time window for “late” data topics by
configuring the RealTimeWindow system parameter. The real-time
window for regular real-time data (not "late") is fixed as 30 seconds.
For more information, see "Editing System Parameters" in Chapter 9,

You must balance the need for a larger real-time window with the
amount of memory available on your computer. If your system has a
large tag count or high data throughput, increasing the real-time
window increases the memory requirements for storage because the
storage system must process more data as real-time data, which is
more resource-intensive than the storage of late or old data.
Adjusting the real-time window also has implications if you are using
delta storage with a swinging door deadband. For more information,
see ""Swinging Door" Deadband for Delta Storage" on page 104.
If the system does not have enough memory to adequately process
real-time data, the window is adjusted internally. An appropriate
message is logged. The value of the RealTimeWindow system
parameter, however, remains unchanged.
Important: The real-time window is not intended to accommodate

time synchronization problems between an IDAS and the historian. It is
imperative that you properly synchronize the IDAS and historian. If the
IDAS is sending data in a steady stream outside of the real-time
window, it is likely there is a time synchronization problem. For more
information, see "Time Synchronization for Data Acquisition" on
page 91.
If a data value is discarded because it did not fit the requirements of

the real-time window, the historian logs a warning message. Warning
messages are logged at one intervals during the period when data is
being discarded.
Determining If the Real-Time Window Is

Configured Appropriately for a Swinging Door
Deadband
To determine if the real-time window is configured correctly for a
swinging door deadband, look at the number of data values that are
forced to be stored while the system waits for the next valid data point
to make the filtering calculation.
The SysRateDeadbandForcedValues system tag counts the number of
"extra" points forced to be stored as a result of an insufficient real-time
window for swinging door storage. Also, you can determine the number
of points forced to be stored for an individual tag by executing a query
that uses the full retrieval mode and specifies a quality detail of 2240,
which indicates that these points were stored because of an
insufficient real-time window.

Reserved Values for Classic Storage129
If you find a large number of forced storage points, you can either
reconfigure the tag to use delta storage or increase the real-time
window.
Note: The first two points received for a tag configured for swinging
door storage are always stored.
Disk Requirements and Performance

Considerations for a Swinging Door Deadband
One of the benefits of using the swinging door deadband is better data
compression. However, because the storage system already provides a
good compression ratio, the amount of disk space that is saved by
applying this type of deadband for slow-changing tags (changing less
than twice in a 15-minute interval) is negligible. For example, a tag
that changes 12 times per hour will use 2K bytes of disk space in a
24-hour period. Even if only every fifth point is stored, the savings is
only 1.5K bytes per day.
Also, use caution when setting the real-time window to accommodate a
swinging door deadband.
• If your system has a large tag count or high data throughput,
increasing the real-time window will increase the memory
requirements for storage, because the storage system will have to
process more data as real-time data, which is more
resource-intensive than the storage of late data.
• If you increase the real-time window and you apply a swinging
door deadband to a slow-changing tag, the amount of storage space
required increases because the tag value is forced to be stored more
often than if you used delta storage with no deadband.
Reserved Values for Classic Storage

Reserved values for classic storage are:
• -2,147,483,648, which represents an inserted NULL value for
32-bit tags.
• -32767, which represents an inserted NULL value for 16-bit tags.
• The last stored value for all types of tags.
• The last received value for analog tags stored cyclically.
• The last received value for tags that are configured for swinging
door storage.
• Invalid values for all tag types except for variable length strings.

About the Active Image for Classic Storage

The active image is an allocation of memory in which copies of values
of acquired data are temporarily held so as to service client requests
while the actual data is being written to disk. The rate at which values
are acquired into the active image depends on the rate of change of the
incoming values for a particular tag. Values will collect in the active
image until the default number of values to be held is reached. This
value limit is initially set to 65 values (samples) per tag.
Note: The rate at which values are acquired by the active image is
NOT related to the value that is stored in the AcquisitionRate column of
the Tag table.
When the sample limit for the active image is reached, the oldest tag
values will start to be overwritten with new tag values. Overwriting
the older tag values makes room for new tag values, with a default
number of 65 values for each tag being held in the active image at any
given time. Also, all of the values stored in the active image may not be
stored to disk: it depends on the storage rate (from the StorageRate
column in the Tag table). If the samples in the active image are
acquired at a rate that is faster than the storage rate, tag values will
be acquired into the active image at a higher resolution than they will
be stored.
To prevent data values in the active image from being overwritten, the
number of samples held would need to be increased to cover the time
gap required for the storage subsystem to store the actual values to
disk.
Automatic Resizing of the Active Image

The system continuously recalculates the optimum number of samples
for each tag based on the data rates received. Beginning at one minute
after startup and every five minutes thereafter, the system will
perform a test to see if the amount of memory allocated for the active
image needs to be increased. The system will calculate an average
time span for a sample for a tag, based on the timestamp of the first
and last samples in the active image and the number of samples. If
necessary, the system will increase the number of samples. However,
if the calculated time span is greater than 65 seconds, the system will
not change the value from the default 65 values.
In short, if you receive more than 65 changes in a value within a
65-second interval, the system will increase the number of
SamplesInActiveImage in the Tag table and increase the size in
memory accordingly to accommodate the additional samples for the
fast-changing tag values.

About the Active Image for Classic Storage131
The new calculated sample number is the number of values required

in the active image to hold data for 1 min (+15%), as calculated by the
system. This value is updated only if the AIAutoResize system
parameter is set to 1 and the number of required samples is greater
than 65. This value is written to the SamplesInActiveImage column of
the Tag table at system startup.
To turn off the automatic increasing of the active image, set the value
of the AIAutoResize system parameter to 0. Also, you can change the
rate at which the system recalculates the samples to be used for each
tag in the active image using the AIResizeSecInterval system
parameter. For more information, see "System Parameters" on
page 36.
The number of samples in the active image never automatically
decreases, but you could manually decrease it using the
ww_SetAISamples stored procedure. However, the changes will not
take effect until the next system startup; the running system will not
automatically detect that the new number specified is lower than
number of samples required to hold values in active image for one
minute. For information, see Chapter 4, "Stored Procedures," in the
Historian Database Reference.
Resizing the active image is a very resource extensive operation. To
prevent unnecessary "noise," after the active image adjusts the
samples for a particular tag, it will change this number only if the
increase in samples more than 10 percent of the previous number of
samples.
You could use the active image to optimize your system, because the
retrieval subsystem will first check the active image to see if the start
time for all tags is in the active image. If so, only the active image is
queried. If a tag does not exist in the active image at the start time,
data is retrieved in parallel from both the active image and disk and
then merged together. If duplicate values are encountered during the
merge, the value from disk will be used.
For example, if you need to create hourly reports for a week of
production based on the "SysTimeHour" tag, then you might want to
increase the number of samples in the active image for that tag to 144
(24 * 7 = 144). This way, the system will never go to disk to retrieve
data for "SysTimeHour," and both the overall time for retrieving data
and the CPU load will decrease.
Important: Although you can manually set the active image samples
for a variable length string to a value other than 0, NULL, or 65, this is
not recommended. The performance impact will be extremely high,
because each sample for a variable length string is 1038 bytes
allocated in memory.

How the Active Image Storage Option Affects

Data Retrieval
You can configure whether to see the data in the active image as it
comes into the system or based on the storage algorithm. You can
specify this when you configure the storage options for the tag. For
more information, see Chapter 2, "Configuring Tags," in the
If you configure the active image to hold all received values, you may
see a discrepancy between tag values for the same time period,
depending on when you ran the query. This is true for tags that are
stored cyclically or by exception (delta) with value or time deadbands.
For example, if you run a query for data between 20011206
1:00:00:000 and 20011206 1:02:00:000, and data during this time
period is being held in the active image, data will be returned at the
resolution it was acquired.
However, if you run the same query later, and the values are now
stored in the history blocks on disk, you may not see such a high
resolution of data. Some values may be discarded if the storage rate is
slower than the acquisition rate. It may appear that you have "lost"
some of the values when, in fact, they were never configured to be
stored in the first place.
Dynamic Configuration Effects on Classic

Storage
During dynamic configuration, the Configuration Service will
determine if the changes require the creation of a new history block.
The storage of existing tags will not be interrupted during dynamic
reconfiguration; however, storage for new or affected tags may not
begin until five to ten minutes after the reconfiguration was
committed, unless you have already allocated memory for the new
tags. For more information, see "Pre-allocating Memory for Future
Tags" in Chapter 2, "Configuring Tags," in the Wonderware Historian
If the storage subsystem cannot complete the reconfiguration for some
reason, a critical or fatal error is generated and written to the error
log. An error is also generated if the reconfiguration process takes so
long that the storage buffer overflows and data is lost.

Memory Management for Classic Storage133
Memory Management for Classic Storage

By default, the Wonderware Historian loads all tag information for the
history blocks into memory so that it can more efficiently service
requests for data. This tag information includes tag properties and
indexing information and allows for quick navigation through the files
containing real-time data. The process that manages the tag
information in memory is the Wonderware Historian Indexing Service
(aahIndexSvc.exe).
For large systems, it is possible that loading the tag information from
all of the history blocks will require more memory than the 2 GB limit
that is imposed by the Windows operating system for a single process.
The actual limit may be even be less than 2 GB, if the amount of
installed RAM is insufficient.
The total amount of tag information for the history blocks depends not
only on the total number of tags, but also on the number of tag
versions, which are created during modifications to old data.
Therefore, it is recommended that you monitor the memory
consumption for all systems, large and small, if you are regularly
performing data inserts, updates, or CSV file imports.
To avoid excessive memory consumption by the Wonderware Historian
Indexing Service, tune and monitor the service for your system using
the following system parameters and system tags.
• HistoryCacheSize and HistoryDaysAlwaysCached system
parameters
You can limit the maximum amount of memory the Indexing
Service can use for tag information by adjusting the value of the
HistoryCacheSize system parameter. When this parameter is set 0
(default), the Indexing Service selects a default cache value
automatically by taking into account the amount of installed
physical memory and the maximum available address space for the
process. In some rare cases when some specific performance tuning
is needed, you may want to set the HistoryCacheSize parameter
manually. In this case, the Indexing Service uses the specified
value, but still may automatically change the effective
HistoryCacheSize if the specified value is too low or two high.
Regardless of whether the effective HistoryCacheSize was selected
automatically (default) or specified by you, the Indexing Service
manages the cache using a "least-recently used" algorithm. In this
algorithm, when there is a request to access a history block that is
not currently cached, the Indexing Service unloads the tag
information from the least-recently used history block and then
loads the tag information from the requested block.

All of these operations are performed automatically in the

background, but you may notice a slowness data retrieval if the
data is retrieved from a block that is not currently loaded into
memory. Keep in mind that the smaller the amount of memory
that you allocate for the cache, the potentially longer it may take to
service data requests.
To guarantee the maximum retrieval performance for the newest
history blocks (for example, if you a running a trend application for
the last week), you can "lock" a certain number of the most recent
history blocks in the cache. To do this, set the number of days to be
locked in the cache by changing the HistoryDaysAlwaysCached
system parameter.
• SysHistoryCacheFaults and SysHistoryCacheUsed system tags
To determine if you need to clamp the memory used by the
Indexing Service, use the Windows Task Manager application or
the Performance console to see how much memory is used by the
aahIndexSvc.exe process. Also, you can monitor the
SysHistoryCacheFaults and SysHistoryCacheUsed system tags. A
high number of cache faults may be indicating that the cache size
is insufficient. The SysHistoryCacheUsed system tag shows the
number of bytes currently used for keeping the tag information.
This tag may be helpful to see how much memory is consumed by
the tag information, even if the memory management is not
enabled.
At any time, you can observe the current status of the history blocks in
the Wonderware Historian Management Console. When the tag
information from a history block is not loaded into memory, the history
block icon is dimmed. You can manually refresh the console window to
see changes in the status for the history blocks.

135
Chapter 7
Data Retrieval Subsystem
The Wonderware Historian data retrieval subsystem receives SQL

queries from clients, locates the requested data, performs necessary
processing, and then returns the results. For configuration and event
data, retrieval is made possible by normal SQL queries, because these
types of data are stored in standard SQL Server database tables.
Historical data, however, must be retrieved from history blocks and
then sent to clients as if it is stored in SQL Server tables.
To accomplish retrieval from both data repositories, the retrieval
subsystem includes:
• An implementation of a SQL Server data provider, which
determines whether the requested data is saved in normal SQL
Server tables or in history blocks.
• A retrieval service, which is responsible for extracting the
requested data from the history blocks and presenting to the
Wonderware Historian OLE DB provider as "virtual" history
tables.
• A set of SQL Server extensions, which are implemented as
columns in the history tables. You can use these extensions to
specify the nature of the rowset that is returned, such as the
number of rows returned, the resolution of the data, or the
retrieval mode.
For more information on data storage, see Chapter 5, "Data Storage
Subsystem."

136  Chapter 7 Data Retrieval Subsystem
Data Retrieval Features

Some of the main features of the data retrieval subsystem are:
• All tag types can be included in the same query when retrieving
from the History table. Any combination of tags can be submitted
in a single query.
• Both fixed length and variable length strings are supported.
• All internal time computation and manipulation is done using the
Win32 FILETIME type. The resolution of FILETIME is 100
nano-seconds. The resolution exposed in queries depends on the
version of SQL Server used.
• All times are handled internally as absolute time (UTC).
Conversions to and from local time are handled going in and out of
retrieval so the external interface is local time.
• Non-real-time data is supported (for example, store-and-forward
data or data imported from a comma-separated values (CSV) file.
• Retrieval of different versions is supported.
History Blocks: A SQL Server Remote Data

Source
Remote data sources are data repositories that exist outside of a SQL
Server database file (.MDF). Microsoft sometimes refers to these types
of data sources as "non-local data stores." In the case of the
Wonderware Historian, a remote data source is the set of history block
files. All tag data is stored in history blocks. For more information on
history blocks, see "History Blocks" on page 114.
OLE DB technology can be used to access data in any remote data
store. This access is accomplished though a software component called
an OLE DB provider.

Retrieval Service137
Retrieval Service
The retrieval service (aahRetSvc.exe) retrieves history data from both
the active image and history blocks on disk. The retrieval service:
• Formats data so that it can be passed up through the system to the
Wonderware Historian OLE DB provider or other MDAS-enabled
or HCAL-enabled client applications.
• Returns information regarding the history blocks, such as the start
and end dates and the location.
Linear scaling for analog tags is performed within retrieval using the
following formula.
V_out = (V_in - MinRaw)*(MaxEU - MinEU)/(MaxRaw - MinRaw)
+ MinEU
where:
V_out = scaled output value
V_in = stored raw value
MinRaw = minimum raw value for the tag
MaxRaw = maximum raw value for the tag
MinEU = minimum engineering unit value
MaxEU = maximum engineering unit value
About the Wonderware Historian OLE DB

Provider
Object Linking and Embedding for Databases (OLE DB) is an
application programming interface (API) that allows COM-based
client applications to access data that is not physically stored in the
SQL Server to which they are connecting.
The benefit of using OLE DB is that it provides access to different
types of data in a broader manner. By using OLE DB, you can
simultaneously access data from a variety of sources, such as from a
SQL Server database, an Oracle database, and a Microsoft Access
database. A query that accesses data from multiple, dissimilar data
sources such as these is called a "heterogeneous query," with
"heterogeneous" meaning "dissimilar." A heterogeneous query can also
be called a "distributed query," because the data can be distributed
across various data sources.

SQL Server uses OLE DB to process heterogeneous queries and makes

linking data between the data sources much easier. Through OLE DB,
Microsoft SQL Server supports Transact-SQL queries against data
stored in one or more SQL Server and heterogeneous databases
without any need for specialized gateway server applications.
The interface required to access data in a non-local data store (such as
the Wonderware Historian history blocks) is provided by a "virtual"
server, called an OLE DB provider. OLE DB providers allow you to use
the power of the SQL Server query processor to make linking data
stored in the SQL Server databases and from history blocks much
easier and more robust. Also, the Wonderware Historian OLE DB
provider has a rich set of query capabilities.
The name of the Wonderware Historian OLE DB provider is "INSQL."
The Wonderware Historian OLE DB provider is installed during
Wonderware Historian installation and then associated, or linked,
with the Microsoft SQL Server. For information on the syntax for
linking the Wonderware Historian OLE DB provider, see "Linking the
Wonderware Historian OLE DB Provider to the Microsoft SQL Server"
on page 153.
To access Wonderware Historian historical data using OLE DB, any
COM-based client application must connect directly to the SQL Server
and then specify to use the Wonderware Historian OLE DB provider in
the syntax of the query.
When you execute a query and specify the Wonderware Historian OLE
DB provider in the syntax, the Microsoft SQL Server parser will pass
the appropriate parts of the data request to the Wonderware Historian
OLE DB provider. The Wonderware Historian OLE DB provider will
then interface with the retrieval service to locate the data store,
extract the requested information, and return the data to the Microsoft
SQL Server as a rowset. Microsoft SQL Server will perform any other
processing required on the data and return the data to the client
application as a result set and a set of output parameters, if
applicable.
The Wonderware Historian OLE DB provider must be present on the
server running Microsoft SQL Server. The set of Transact-SQL
operations that can be used to retrieve data in the history blocks
depends on the capabilities of the Wonderware Historian OLE DB
provider. The Wonderware Historian OLE DB provider is SQL-92
compliant.
For more information on OLE DB, see your Microsoft documentation.

About the Wonderware Historian OLE DB Provider139
Extension (Remote) Tables for History Data

Some of the history tables are rowset representations that provide a
means for handling acquired plant data. These tables are not part of
normal SQL Server functionality. A normal SQL Server table stores
data directly in the database's data device file (.mdf). An extension
table, however, presents data as if it were a real table, but it does not
physically exist in the data device. An extension table is a logical table
that is populated from other types of data files; thus, the data is stored
"remotely" from SQL Server. In the case of the Wonderware Historian,
the data files are the history blocks generated by the storage system.
Note: Extension tables are also called remote tables.
Data access from the history blocks is made possible by SQL Server's
OLE DB provider technology. Client applications must connect
directly to the Microsoft SQL Server and then specify to use the
Wonderware Historian OLE DB provider in the syntax of the query.
The extension tables are:
• AnalogSummaryHistory
(INSQL.Runtime.dbo.AnalogSummaryHistory)
• History (INSQL.Runtime.dbo.History)
• HistoryBlock (INSQL.Runtime.dbo.HistoryBlock)
• Live (INSQL.Runtime.dbo.Live)
• StateSummaryHistory
(INSQL.Runtime.dbo.StateSummaryHistory)
• StateWideHistory (INSQL.Runtime.dbo.StateWideHistory)
The AnalogHistory, DiscreteHistory, StringHistory, AnalogLive,
DiscreteLive, StringLive, AnalogWideHistory, DiscreteWideHistory,
StringWideHistory, and v_SummaryData tables are provided for
backward compatibility. For more information, see Chapter 6,
"Backward Compatibility Entities," in the Historian Database
Reference.
The AnalogHistory, DiscreteHistory, StringHistory, and History tables
are the only tables which are updateable. The remaining tables are
read-only.
For more information on the history extension tables, see "History
Tables" in Chapter 1, "Table Categories," in the Historian Database
Reference.

Query Syntax for the Wonderware Historian OLE

DB Provider
The most common Wonderware Historian query is a SELECT
statement:
SELECT select_list
FROM table_source
WHERE search_condition
[ GROUP BY group_by_expression ]
[ HAVING search_condition ]
[ ORDER BY order_expression [ ASC | DESC ] ]
A WHERE clause is mandatory when issuing a SELECT query against

any extension table except HistoryBlock.
There are four variations for issuing a SELECT statement to the
Wonderware Historian OLE DB provider to retrieve history data:
• Using the Four-Part Naming Convention
• Using a Wonderware Historian OLE DB Provider View
• Using the OPENQUERY Function
• Using the OPENROWSET Function
You should use the four-part name or a provider view to specify the
extension table, whenever possible. However, there are instances
when the OPENQUERY or OPENROWSET function must be used,
such as for queries on wide tables.
For general information on creating SQL queries, see your Microsoft
SQL Server documentation.
Using the Four-Part Naming Convention

The linked server name is simply a name by which the Wonderware
Historian OLE DB provider is known to the Microsoft SQL Server. In
order for a query to be passed on to the Wonderware Historian OLE
DB provider, you must specify the linked server name and the
extension table name as part of a four-part naming convention.
For example, this query specifies to retrieve data from the History
extension table in the Wonderware Historian OLE DB provider:
SELECT * FROM INSQL.Runtime.dbo.History
WHERE TagName = 'SysTimeSec'
AND DateTime >= '2001-09-12 12:59:00'
AND DateTime <= '2001-09-12 13:00:00'

The four-part naming convention is described in the following table:
Part Name Description
linked_server Linked server name. By default, INSQL.

catalog Catalog in the OLE DB data source that contains
the object from which you want to retrieve data.
For Microsoft SQL Server type databases, this is
the name of the database. To use the Wonderware
Historian OLE DB provider, the catalog name will
always be "Runtime."
schema Schema in the catalog that contains the object. For
Microsoft SQL Server type databases, this is the
name of the login ID for accessing the data. To use
the Wonderware Historian OLE DB provider, the
catalog name will always be "dbo."
object_name Data object that the OLE DB provider can expose
as a rowset. For the Wonderware Historian OLE
DB provider, the object name is the name of the
remote table that contains the data you want to
retrieve. For example, the History table.
In the case of four-part queries, SQL Server produces the statement

that is sent to the Wonderware Historian OLE DB provider from the
statement that the user executes. Sometimes this produced statement
is incorrect, too complex, or lacks portions of the WHERE clause
required for the Wonderware Historian OLE DB provider to return
data.
A typical error message when executing unsupported syntax is:
Server: Msg 7320, Level 16, State 2, Line 1
Could not execute query against OLE DB provider 'INSQL'.
[OLE/DB provider returned message: InSQL did not receive a WHERE

clause from SQL Server. If one was specified, refer to the
InSQL OLE DB documentation]

For four-part queries against non-English SQL Servers running on

non-English operating systems, the default date format might differ
from the English versions. For example, for a French or German SQL
Server running on the corresponding operating system, the date/time
in a four-part query must be:
yyyy-dd-mm hh:mm:ss.fff
For example:
2003-28-09 09:00:00.000
The default SQL date format is dependent on SQL Server and not on
the operating system used. However, you can modify the format using
the SQL Server Convert() method. The output of this method can be
determined by the regional settings configured for the operating
system.
Using a Wonderware Historian OLE DB Provider

View
Microsoft SQL Server views have been provided that will access each
of the extension tables, eliminating the need to type the four-part
server name in the query. These views are named the same as the
provider table name.
Note: Backward compatibility views are named according to the

v_ProviderTableName convention.
For example:
SELECT * FROM History
AND DateTime >= '2001-09-12 12:59:00'
AND DateTime <= '2001-09-12 13:00:00'
Using the OPENQUERY Function

You can use the linked server name in an OPENQUERY function to
retrieve data from an extension table. The OPENQUERY function is
required for retrieving from the wide table. For example:
SELECT * FROM OPENQUERY(INSQL, 'SELECT * FROM History
WHERE TagName = "SysTimeSec"
AND DateTime >= "2001-09-12 12:59:00"
AND DateTime <= "2001-09-12 13:00:00"
')

The following example retrieves data from a wide table:

SELECT * FROM OPENQUERY(INSQL, 'SELECT DateTime, SysTimeSec
FROM WideHistory
WHERE DateTime >= "2001-09-12 12:59:00"
AND DateTime <= "2001-09-12 13:00:00"
')
The OPENQUERY portion of the statement is treated as a table by

SQL Server, and can also be used in joins, views, and stored
procedures. SQL Server sends the quoted statement, unchanged and
as a string, to the Wonderware Historian OLE DB provider.
Consequently, only the syntax that the Wonderware Historian OLE
DB provider can parse is supported. Also, be sure that you do not
exceed the 8000 character limit for the statement. Consider the
following example:
SELECT * FROM OpenQuery(INSQL, 'XYZ')
where "XYZ" is the statement to pass. You should be sure that the
value of "XYZ" is not more than 8000 characters. This limit is most
likely to cause a problem if you are querying many tags from a wide
table.
Also, you should supply the datetime in an OPENQUERY statement
in the following format:
yyyy-mm-dd hh:mm:ss.fff
For example:
2001-01-01 09:00:00.000
You cannot use variables in an OPENQUERY statement. For more
information, see "Using Variables with the Wide Table" on page 320.
Using the OPENROWSET Function

The linked server name can be used as an input parameter to an
OPENROWSET function. The OPENROWSET function sends the
OLE DB provider a command to execute. The returned rowset can
then be used as a table or view reference in a Transact-SQL statement.
For example:
SELECT * FROM OPENROWSET('INSQL',' ', 'SELECT DateTime, Quality,
QualityDetail, Value
FROM History
WHERE TagName in ("SysTimeSec")
AND DateTime >= "2001-09-12 12:59:00"
AND DateTime <= "2001-09-12 13:00:00"
')

Syntax Options Supported

The following table indicates the syntax options that are available for
queries that use either the four-part naming convention (or
corresponding view name) or the OPENQUERY function.
Four-Part
Syntax Element Query OPENQUERY
ORDER BY Yes No. Does not work within the

OPENQUERY function. However,
will work if used outside of the
function.
GROUP BY Yes No
TagName IN ( ..) Yes Yes
TagName LIKE '..' Yes Yes
Date and time functions (for Yes Yes
example, DateAdd)
MIN, MAX, AVG, SUM, STDEV Yes MIN, MAX, AVG, SUM only
Sub-SELECT with one normal SQL Yes, with No
Server table and one extension table restrictions
Sub-SELECT with two extension No No
tables
Wonderware Historian OLE DB Provider

Unsupported Syntax and Limitations
The Wonderware Historian OLE DB provider does not support certain
syntax options in queries. In general, these limitations are due to
underlying limitations in the current Microsoft SQL Server OLE DB
Provider implementation.

No Notion of Client Context

The OLE DB provider has no notion of a client context. The OLE DB
provider is entirely stateless, and there is no persistence across
queries in the same connection. This means that you must set the
value of a Wonderware Historian time domain extension (for example,
cycle count) each time you execute a query.
Also, the OLE DB provider cannot continuously return data (similar to
a "hot" link in InTouch HMI software). The OLE DB specification (as
defined by Microsoft) does not permit a provider to return rows to a
consumer without a request from the consumer.
Limitations on Wide Tables

Wide tables do not have a fixed schema, but a schema which varies
from query to query. They are transient tables, existing for the
duration of one query only. For this reason, they must be accessed
using the OPENQUERY function, which bypasses many of the tests
and requirements associated with fixed tables. Wide tables support up
to 1024 columns.
For more information on wide tables, see “Wide” History Table Format
in Chapter 1, Table Categories, in the Historian Database Reference.
LIKE Clause Limitations

The LIKE clause is only supported for the TagName and Value
columns. The syntax " ... Value LIKE 'a string' ... " is only
supported for a string table. For example:
SELECT TagName, Value FROM History
WHERE TagName LIKE 'Sys%'
AND DateTime > '1999-05-24 14:30:00'
AND DateTime < '1999-05-24 14:32:00'
IN Clause Limitations
If you are querying analog, discrete, or string tags from the
AnalogTag, DiscreteTag, or StringTag tables (respectively), you cannot
use the LIKE clause within an IN clause to condition the tagname
unless you are returning the vValue column. This restriction applies if
you are using the four-part naming convention or an extension table
view.

For example:
SELECT DateTime, TagName, vValue, Quality, QualityDetail
FROM History
WHERE TagName IN (SELECT TagName FROM StringTag WHERE
TagName LIKE 'SysString')
AND DateTime >='2001-06-21 16:00:00.000'
AND DateTime <='2001-06-21 16:40:00.000'
AND wwRetrievalMode = 'Delta'
However, it is more efficient to use an INNER REMOTE JOIN to

achieve the same results. For more information, see "Using an INNER
REMOTE JOIN" on page 283.
OR Clause Limitations
You cannot use the OR clause to specify more than one condition for a
time domain extension. For more information, see "Wonderware
Historian Time Domain Extensions" on page 154.
Using Joins within an OPENQUERY Function

Joins are not supported within a single OPENQUERY statement. For
example, the following query contains an implicit join between the Tag
and Live tables, and will fail:
SELECT * FROM OPENQUERY(INSQL, 'SELECT v.DateTime, v.TagName,
v.Value, t.Description
FROM Tag t, Live v
WHERE t.TagName LIKE "%Date%"
AND v.TagName = t.TagName
')
A work-around is to place the join outside of the OPENQUERY. For

example:
SELECT v.DateTime, v.TagName, v.Value, t.Description
FROM OPENQUERY(INSQL, 'SELECT DateTime, TagName, Value
FROM Live
WHERE TagName LIKE "%Date%"
') v, Tag t
WHERE v.TagName = t.TagName
Explicit joins are also not supported within OPENQUERY. For

example, the following query will fail:
SELECT * FROM OPENQUERY(INSQL, 'SELECT v.DateTime, v.TagName,
v.Value, e.Unit
FROM Live v
JOIN AnalogTag t ON v.TagName = t.TagName
JOIN EngineeringUnit e ON t.EUKey = e.EUKey
WHERE v.TagName LIKE "%Date%"
')

A work-around is to place the join outside the OPENQUERY. For

example:
SELECT v.DateTime, v.TagName, v.Value, e.Unit
FROM OPENQUERY(INSQL, 'SELECT DateTime, TagName, Value FROM
Live
WHERE TagName LIKE "%Date%"
') v
ORDER BY t.TagName
In general, use four-part syntax wherever possible. All of the previous

queries are more conveniently expressed in four-part syntax. For
example, the syntax for the preceding query would be:
SELECT v.DateTime, v.TagName, v.Value, e.Unit
FROM INSQL.Runtime.dbo.History v
WHERE v.TagName LIKE '%Date%'
ORDER BY t.TagName
Using Complicated Joins

You can only use simple joins between SQL Server tables and the
Wonderware Historian OLE DB extension tables. Joins typically
require use of the INNER REMOTE JOIN syntax.
For an example of the INNER REMOTE JOIN syntax, see "Using an
INNER REMOTE JOIN" on page 283.
Using a Sub-SELECT with a SQL Server Table and

an Extension Table
Using a sub-SELECT with a query on a normal SQL Server table and
an extension table should be avoided; it is very inefficient due to the
way SQL Server executes the query. For example:
SELECT TagName, DateTime, Value
FROM INSQL.Runtime.dbo.History
WHERE TagName IN (select TagName FROM SnapshotTag WHERE
EventTagName = 'SysStatusEvent')
AND DateTime = '2001-12-20 0:00'

Instead, it is recommended that you use the INNER REMOTE JOIN

syntax:
SELECT h.TagName, DateTime, Value
FROM SnapshotTag st INNER REMOTE JOIN
INSQL.Runtime.dbo.History h
ON st.TagName = h.TagName
AND EventTagName = 'SysStatusEvent'
AND DateTime = '2001-12-20 0:00'
The results are:

TagName DateTime Value
SysPerfCPUTotal 2001-12-20 00:00:00.000 15.0
SysSpaceMain 2001-12-20 00:00:00.000 1302.0
In general, use the following pattern for INNER REMOTE JOIN

queries against the historian is:
<SQLServerTable> INNER REMOTE JOIN
<HistorianExtensionTable>
For more information on INNER REMOTE JOIN, see your Microsoft
documentation.
WHERE Clause Anomalies

In some rare cases, the SQL Server query processor truncates the
WHERE clause in an attempt to optimize the query. If you execute a
query with a WHERE clause, but an error message is returned stating
that no WHERE clause was received by the SQL Server, simply add
another condition clause to the query.
For example, in the following query, the SQL Server query processor
optimizes out the WHERE clause, because it is superfluous.
SELECT DateTime, Value, QualityDetail
FROM History
WHERE TagName LIKE '%'
A workaround is to add another condition clause. For example:

SELECT DateTime, Value, QualityDetail
FROM History
WHERE TagName LIKE '%'
AND wwRetrievalMode = 'delta'

CONVERT Function Limitations

The CONVERT function is not supported on the vValue column in an
OPENQUERY statement. If you are using OPENQUERY on the
History table, you must filter on the vValue column outside of the
query.
In the following example, the value of the vValue column is converted
to a float. Note that no string tags are included in the query.
SELECT * FROM OpenQuery(INSQL, 'SELECT DateTime, Quality,
OPCQuality, QualityDetail, Value, vValue, TagName
FROM History
WHERE TagName IN ("SysTimeMin", "SysPulse")
AND DateTime >= "2001-12-30 04:00:00.000"
AND DateTime <= "2001-12-30 09:00:00.000"
AND wwRetrievalMode = "Delta"
')
WHERE convert(float, vValue) = 20.0
You can also use the following formats:

WHERE convert(float, vValue) = 0
WHERE convert(float, vValue) = 2.0000e01
The following example includes a string tag and converts the vValue
value to a char or varchar datatype. All values returned can be
converted to a string.
SELECT * FROM OpenQuery(INSQL, 'SELECT DateTime, Quality,
OPCQuality, QualityDetail, Value, vValue, TagName
FROM History
WHERE TagName IN ("SysString", "SysTimeMin", "SysPulse")
AND DateTime >= "2001-12-30 04:00:00.000"
AND DateTime <= "2001-12-30 09:00:00.000"
AND wwRetrievalMode = "Cyclic"
AND wwCycleCount = 300
')
WHERE convert(varchar(30), vValue) = '2001-12-30 14:00:00'
You can also use the following formats:

WHERE convert(varchar(30), vValue) = '20'

SQL Server Optimization of Complex Queries

The SQL Server query optimizer may incorrectly parse a complex
query and not send certain query criteria to the Historian OLE DB
provider for handling. This can cause unexpected results for the data.
If you suspect that this is happening, use SQL Server Management
Studio tools to examine the query plan that the optimizer is using and
then modify your query so that the needed criteria gets directed to the
Historian OLE DB provider.
For example, the following query will be incorrectly parsed:
SELECT GETDATE()
DECLARE @TagList TABLE (TagName nvarchar(256))
INSERT @TagList
SELECT 'SysTimeSec' UNION
SELECT 'SysPerfCPUTotal'
-- Prevent the TagName criteria from being sent to the Historian

OLE DB provider (incorrect)
SELECT DateTime, h.vValue, h.TagName

FROM History h
INNER REMOTE JOIN @TagList l
ON h.TagName = l.TagName
WHERE DateTime >= DATEADD(hour,-1,GETDATE())
AND DateTime < GETDATE()
AND wwRetrievalMode = 'AVG'
AND wwCycleCount=1
GO

To correct this issue, rewrite the query so that the tagname criteria is
passed to the Historian OLE DB provider correctly.
SELECT GETDATE()
DECLARE @TagList TABLE (TagName nvarchar(256))
INSERT @TagList
SELECT 'SysTimeSec' UNION
SELECT 'SysPerfCPUTotal'
-- Force the TagName criteria to be sent to the InSQL OLE DB

Provider (correct)
SELECT DateTime, h.vValue, h.TagName

FROM @TagList l
INNER REMOTE JOIN History h
ON h.TagName = l.TagName
WHERE DateTime >= DATEADD(hour,-1,GETDATE())
AND DateTime < GETDATE()
AND wwRetrievalMode = 'AVG'
AND wwCycleCount=1
GO
Using Columns of a Variant Type with Functions

If you use a column of a variant type as the parameter for some
functions, SQL Server returns a syntax error. However, the error is
not passed to the Historian OLE DB provider to return to clients.
For example, in the following query, the rounding is specified for the
vValue column, which is of type variant. The query does not work, but
no error is returned by the Historian OLE DB provider.
SELECT DateTime, round(vValue, 2)
FROM History
WHERE TagName IN ('SysTimeSec')
AND DateTime = getdate()
AND wwRetrievalMode = 'Cyclic'

Using StartDateTime in the Query Criteria

You cannot use StartDateTime in the query criteria instead of
DateTime. For example, the following query works, except that it does
not apply the StartDateTime >= @StartDate clause.
SET NOCOUNT ON
DECLARE @StartDate DateTime
DECLARE @EndDate DateTime
SET @StartDate = DateAdd(mi,-30,GetDate())
SET @EndDate = GetDate()
SET NOCOUNT OFF
SELECT History.TagName, DateTime = convert(nvarchar, DateTime,

21), Value, vValue, StateTime, StartDateTime
FROM History
WHERE History.TagName IN ('Reactor1Level')
AND wwRetrievalMode = 'RoundTrip'
AND wwStateCalc = 'AvgContained'
AND vValue = convert(SQL_VARIANT, '1')
AND wwTimeStampRule = 'Start'
AND wwQualityRule = 'Good'
AND wwFilter = 'ToDiscrete(5.0,>)'
AND wwVersion = 'Latest'
AND DateTime >= @StartDate
AND DateTime <= @EndDate
AND StartDateTime >= @StartDate
Comparison Statements and NULL Values

SQL Server returns an error for a query that contains a comparison
statement like 'Value > 0' whenever a NULL is returned. Be sure that
you always include 'AND Value IS NOT NULL', so that the NULL
values are filtered out.
OPENQUERY and Microsoft Query

Microsoft Query is not able to process an OPENQUERY statement.

Linking the Wonderware Historian OLE DB

Provider to the Microsoft SQL Server
Because the Wonderware Historian OLE DB provider retrieves data
from the history blocks and presents it to Microsoft SQL Server as a
table, it can be thought of as a type of server. The Wonderware
Historian OLE DB provider must be added to the Microsoft SQL
Server as a "linked" server before it can be used to process queries.
This linking is performed automatically during the Wonderware
Historian installation. If, for some reason, you need to re-link the
Wonderware Historian OLE DB provider to the Microsoft SQL Server,
the statements for linking are as follows:
sp_addlinkedserver
@server = 'INSQL',
@srvproduct = '',
@provider = 'INSQL'
go
sp_serveroption 'INSQL','collation compatible',true
go
sp_addlinkedsrvlogin 'INSQL','TRUE',NULL,NULL,NULL
go
"INSQL" is the name of the Wonderware Historian OLE DB provider

as the linked server. Use this name to specify the Wonderware
Historian OLE DB provider in a query.
To perform joins between the legacy analog history tables and discrete
history tables, the installation program also creates an alias for the
same Wonderware Historian OLE DB provider:
sp_addlinkedserver
@server = 'INSQLD',
@srvproduct = '',
@provider = 'INSQL'
go
sp_serveroption 'INSQLD','collation compatible',true
go
sp_addlinkedsrvlogin 'INSQLD','TRUE',NULL,NULL,NULL
go
For example, if you want to execute a query that performs this type of
join, use the normal alias in specifying the first table (the analog
history table), and use the second alias in specifying the second table
(the discrete history table, hence the "D" added to the alias name).

Wonderware Historian Time Domain Extensions

Data in the extension tables can be manipulated by using normal
Transact-SQL code, as well as the specialized SQL time domain
extensions provided by the Wonderware Historian. The Wonderware
Historian extensions provide an easy way to query time-based data
from the history tables. They also provide additional functionality not
supported by Transact-SQL.
The time domain extensions are:
• wwCycleCount
• wwResolution
• wwRetrievalMode
• wwTimeDeadband
• wwValueDeadband
• wwEdgeDetection
• wwTimeZone
• wwVersion
• wwInterpolationType
• wwTimeStampRule
• wwQualityRule
• wwValueSelector
• wwStateCalc
• wwFilter
Note: The wwParameters and wwMaxStates parameters are reserved

for future use. The wwRowCount parameter is still supported, but is
deprecated in favor of wwCycleCount.
The extensions are implemented as "virtual" columns in the extension

tables. When you query an extension table, you can specify values for
these column parameters to manipulate the data that will be returned.
You will need to specify any real-time extension parameters each time
that you execute the query.

Wonderware Historian Time Domain Extensions155
For example, you could specify a value for the wwResolution column in
the query so that a resolution is applied to the returned data set:
SELECT DateTime, Value
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwResolution = 10
AND wwRetrievalMode = 'cyclic'
Because the extension tables provide additional functionality that is

not possible in a normal SQL Server, certain limitations apply to the
Transact-SQL supported by these tables. For more information, see
"Wonderware Historian OLE DB Provider Unsupported Syntax and
Limitations" on page 144.
Although the Microsoft SQL Server may be configured to be
case-sensitive, the values for the virtual columns in the extension
tables are always case-insensitive.
Note: You cannot use the IN clause or OR clause to specify more than
one condition for a time domain extension. For example, "wwVersion
IN ('original', 'latest')" and "wwRetrievalMode = 'Delta'
OR wwVersion = 'latest'" are not supported.


Wonderware Historian I/O Server

The Wonderware Historian I/O Server (aahIOSvrSvc.exe) is the
interface for clients to access data from the active image of a
Wonderware Historian by the SuiteLink protocol. The Wonderware
Historian I/O Server can update items with current values for given
topics, providing "real-time" I/O Server functionality. Tag values in the
Wonderware Historian I/O Server are acquired from the active image.
For more information on the active image, see "About the Active Image
for Classic Storage" on page 130.
Note: NetDDE is not supported.
The Wonderware Historian I/O Server is pre-configured with a single

topic, Tagname. The Wonderware Historian I/O Server will listen for
clients (such as WWClient or WindowViewer™) that are attempting to
establish a connection using the pre-configured topic. After a client
connects with the Wonderware Historian I/O Server, a "hot" link is
established between the client and the Wonderware Historian I/O
Server. For more information on I/O Server addressing conventions,
see "I/O Server Addressing" on page 80.
For example, the Wonderware Historian I/O Server could be used by
InTouch WindowViewer to access system tag values provided by the
Wonderware Historian to monitor system health. You could configure
WindowViewer to generate an alarm when abnormal behavior is
detected within the Wonderware Historian.
By default, the Wonderware Historian I/O Server runs as a Windows
service and can be started and stopped using the System Management
Console. You can also monitor the Wonderware Historian I/O Server
from within the System Management Console. For more information
on the System Management Console, see Chapter 1, "Getting Started
with Administrative Tools," in the Wonderware Historian
The Wonderware Historian I/O Server is a read-only server; clients
cannot update data in the active image.
The Wonderware Historian I/O Server sends the original OPC quality
as it was stored in the Wonderware Historian. The OPC quality
remains the same throughout the system, including storage, retrieval,
and the Wonderware Historian I/O Server.

157
Chapter 8
Data Retrieval Options
You can use a variety of retrieval modes and options to suit different
reporting needs and applications.
Understanding Retrieval Modes

Different retrieval modes allow you to access the data stored in a
Wonderware Historian in different ways. For example, if you retrieve
data for a long time period, you might want to retrieve only a few
hundred evenly spaced data points to minimize response time. For a
shorter time period, you might want to retrieve all values that are
stored on the server to get more accurate results.
A Wonderware Historian with a version earlier than 9.0 supports two
retrieval modes:
• Cyclic Retrieval
• Delta Retrieval
A Wonderware Historian with a version of 9.0 or higher supports
various additional modes:
• Full Retrieval
• Interpolated Retrieval
• "Best Fit" Retrieval
• Average Retrieval
• Minimum Retrieval
• Maximum Retrieval

158  Chapter 8 Data Retrieval Options
• Integral Retrieval
• Slope Retrieval
• Counter Retrieval
• ValueState Retrieval
A Wonderware Historian with a version of 10.0 or higher supports the
following additional mode:
• RoundTrip Retrieval
Cyclic Retrieval
Cyclic retrieval is the retrieval of stored data for the given time period
based on a specified cyclic retrieval resolution, regardless of whether
or not the value of the tag(s) has changed. It works with all types of
tags. Cyclic retrieval produces a virtual rowset, which may or may not
correspond to the actual data rows stored on the Wonderware
Historian.
In cyclic retrieval, one row is returned for each “cycle boundary.” You
specify the number of cycles either directly or by means of a time
resolution, that is, the spacing of cycle boundaries in time. If you
specify a number of cycles, the Wonderware Historian returns that
number of rows, evenly spaced in time over the requested period. The
cyclic resolution is calculated by dividing the requested time period by
the number of cycle boundaries. If you specify a resolution, the number
of cycles is calculated by dividing the time period by the resolution.
If no data value is actually stored at a cycle boundary, the last value
before the boundary is returned.
The default retrieval mode is cyclic for retrieval from analog tables,
including analog and state summary tables.
Cyclic retrieval is fast and therefore consumes little server resources.
However, it may not correctly reflect the stored data because
important process values (gaps, spikes, etc.) might fall between cycle
boundaries. For an alternative, see ""Best Fit" Retrieval" on page 177.

Understanding Retrieval Modes159
Cyclic Retrieval - How It Works

The following illustration shows how values are returned for cyclic
retrieval:
Data is retrieved in cyclic mode with a start time of TC0 and an end
time of TC2. The resolution has been set in such a way that the
historian returns data for three cycle boundaries at TC0, TC1, and TC2.
Each dot in the graphic represents an actual data point stored on the
historian. From these points, the following are returned:
• At TC0: P2, because it falls right on the cycle boundary
• At TC1: P7, because it is the last point before the cycle boundary
• At TC2: P11, for the same reason

Cyclic Retrieval - Supported Value Parameters

You can use various parameters to adjust which values are returned in
cyclic retrieval mode. For more information, see the following sections:
• "Cycle Count (X Values over Equal Time Intervals)
(wwCycleCount)" on page 225.
• "Resolution (Values Spaced Every X ms) (wwResolution)" on
page 228.
• "History Version (wwVersion)" on page 239.
• "Timestamp Rule (wwTimestampRule)" on page 244 (only on
Wonderware Historian 9.0 and above).
Cyclic Retrieval - Query Examples

To use the cyclic retrieval mode, set the following parameter in your
query.
wwRetrievalMode = 'Cyclic'
Query 1
The following query returns data values for the analog tag
'ReactLevel'. If you do not specify a wwCycleCount or wwResolution,
the query will return 100 rows (the default).
SELECT DateTime, Sec = DATEPART(ss, DateTime), TagName, Value
FROM History
WHERE TagName = 'ReactLevel'
AND DateTime >= '2001-03-13 1:15:00pm'
AND DateTime <= '2001-03-13 2:15:00pm'

The results are:

DateTime Sec TagName Value
2001-03-13 13:15:00.000 0 ReactLevel 1775.0
2001-03-13 13:15:00.000 36 ReactLevel 1260.0
2001-03-13 13:16:00.000 12 ReactLevel 1650.0
2001-03-13 13:16:00.000 49 ReactLevel 1280.0
2001-03-13 13:17:00.000 25 ReactLevel 1525.0
2001-03-13 13:18:00.000 1 ReactLevel 585.0
2001-03-13 13:18:00.000 38 ReactLevel 1400.0
2001-03-13 13:19:00.000 14 ReactLevel 650.0
2001-03-13 13:19:00.000 50 ReactLevel 2025.0
2001-03-13 13:20:00.000 27 ReactLevel 765.0
2001-03-13 13:21:00.000 3 ReactLevel 2000.0
2001-03-13 13:21:00.000 39 ReactLevel 830.0
2001-03-13 13:22:00.000 16 ReactLevel 1925.0
.
.
.
(100 row(s) affected)
Cyclic Retrieval - Initial Values

No special handling is done for initial values. The initial value will
behave like a normal cycle boundary at the start time. For information
on initial values, see "Delta Retrieval - Initial Values" on page 167.
Cyclic Retrieval - Handling NULL Values

No special handling is done for NULL values. They are returned just
like any other value.

Delta Retrieval
Delta retrieval, or retrieval based on exception, is the retrieval of only
the changed values for a tag(s) for the given time interval. That is,
duplicate values are not returned. It works with all types of tags.
Delta retrieval always produces a rowset comprised of only rows that
are actually stored on the historian; that is, a delta query returns all of
the physical rows in history for the specified tags, over the specified
period, minus any duplicate values. If there is no actual data point at
the start time, the last data point before the start time is returned.
Delta retrieval is the default mode for discrete and string tables and
from the History table.
Delta Retrieval - How It Works

The following illustration shows how values are returned for delta
retrieval:

Data is retrieved in delta mode with a start time of T1 and an end time
of T2. Each dot in the graphic represents an actual data point stored on
the historian. From these points, the following are returned:
• P2, because there is no actual data point at T1
• P5, P8, P9, P10, and P11, because they represent changed values
during the time period
For delta retrieval for replicated summary tags on a tier-2 historian, if
a point with doubtful quality is returned as the result of a value
selection from an input summary point with a contained gap, the same
point can be returned again with good quality if the same value is
selected again from the next input summary point that has good
quality.
Delta Retrieval - Supported Value Parameters

delta retrieval mode. For more information, see the following sections:
• "Time Deadband (wwTimeDeadband)" on page 232
• "Value Deadband (wwValueDeadband)" on page 236
• "History Version (wwVersion)" on page 239
Delta Retrieval - Query Examples

To use the delta retrieval mode, set the following parameter in your
query.
wwRetrievalMode = 'Delta'
Query 1
As an example of how delta mode works, consider the following query:
SELECT TagName, DateTime, Value, QualityDetail
FROM History
WHERE TagName = 'A001'
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'

This query can be run against the following sample data:

Tagname DateTime Value QualityDetail
A001 2009-09-12 00:11 1.0 192
A001 2009-09-12 00:13 1.6 192
A001 2009-09-12 00:16 1.3 192
A001 2009-09-12 00:21 2.0 192
A001 2009-09-12 00:24 1.2 192
A001 2009-09-12 00:27 1.2 192
A001 2009-09-12 00:28 0.0 249
A001 2009-09-12 00:29 0.0 249
A001 2009-09-12 00:32 0.6 192
A001 2009-09-12 00:35 0.0 249
A001 2009-09-12 00:37 1.5 192
A001 2009-09-12 00:43 1.3 192
A graphical representation of the data is as follows:
The results are:

A001 2009-09-12 00:20 1.3 192
A001 2009-09-12 00:21 2.0 192
A001 2009-09-12 00:24 1.2 192
A001 2009-09-12 00:28 NULL 249
A001 2009-09-12 00:32 0.6 192
A001 2009-09-12 00:35 NULL 249
A001 2009-09-12 00:37 1.5 192
The sample data points and the results are mapped on the following
chart. Only the data falling between the time start and end marks at
2009-09-12 00:20 and 2009-09-12 00:40 (shown on the chart as dark
vertical lines) are returned by the query.

Because there is no value that matches the start time, an initial value
at 2009-09-12 00:20 is returned in the results based on the value of the
preceding data point at 2009-09-12 00:16. Because there is no change
in the value at 2009-09-12 00:27 from the value at 2009-09-12 00:24,
the data point appears on the chart but does not appear in the results.
Similarly, the second 0.0 value at 2009-09-12 00:29 is also excluded
from the results.
You can further control the number of rows returned by using the
wwTimeDeadband, wwValueDeadband, and wwCycleCount
extensions. The use of a cycle count returns the first number of rows
within the time range of the query. For more information, see "Using
wwResolution, wwCycleCount, and wwRetrievalMode in the Same
Query" on page 287.
Also, the use of a time deadband and/or value deadband with delta
retrieval produces differing results. For more information, see "Time
Deadband (wwTimeDeadband)" on page 232 and "Value Deadband
(wwValueDeadband)" on page 236.
Query 1
SELECT DateTime, TagName, Value
FROM History
WHERE TagName IN ('SysTimeSec','SysTimeMin')
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:36'
The results are:

DateTime TagName Value
2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:00.000 SysTimeMin 35
2001-12-09 11:35:01.000 SysTimeSec 1
2001-12-09 11:35:02.000 SysTimeSec 2
2001-12-09 11:35:03.000 SysTimeSec 3
2001-12-09 11:35:04.000 SysTimeSec 4
.
.
.
2001-12-09 11:35:58.000 SysTimeSec 58
2001-12-09 11:35:59.000 SysTimeSec 59
2001-12-09 11:36:00.000 SysTimeSec 0
2001-12-09 11:36:00.000 SysTimeMin 36

Query 2
SELECT * FROM OpenQuery(INSQL,'SELECT DateTime, Value, Quality,
QualityDetail
FROM AnalogHistory
AND Value = 10
AND DateTime >="2001-07-27 03:00:00.000"
AND DateTime <="2001-07-27 03:05:00.000"
')
The results are:

DateTime Value Quality QualityDetail
2001-07-27 03:00:10.000 10 0 192
2001-07-27 03:01:10.000 10 0 192
2001-07-27 03:02:10.000 10 0 192
2001-07-27 03:03:10.000 10 0 192
2001-07-27 03:04:10.000 10 0 192
Query 3
For a delta query, if both a wwCycleCount and a Value comparison are
specified, the query will return the first number of rows (if available)
that meet the value indicated.
SELECT * FROM OpenQuery(INSQL,'SELECT DateTime, Value, Quality,
QualityDetail
FROM AnalogHistory
AND Value = 20
AND DateTime >="2001-07-27 03:00:00.000"
AND DateTime <="2001-07-27 03:20:00.000"
')
The results are:

2001-07-27 03:00:20.000 20 0 192
2001-07-27 03:01:20.000 20 0 192
2001-07-27 03:02:20.000 20 0 192
2001-07-27 03:03:20.000 20 0 192
2001-07-27 03:04:20.000 20 0 192
2001-07-27 03:05:20.000 20 0 192
2001-07-27 03:06:20.000 20 0 192
2001-07-27 03:07:20.000 20 0 192


2001-07-27 03:08:20.000 20 0 192
2001-07-27 03:09:20.000 20 0 192
Delta Retrieval - Initial Values

Initial values are special values that can be returned from queries that
lie exactly on the query start time, even if there is not a data point that
specifically matches the specified start time. If there is not a value
exactly on the query start time, the last point before the start time will
be returned with its DateTime set to the query start time and its
Quality set to 133. If no value exists at or prior to the query start time,
a NULL value will be returned at start time with QualityDetail set to
65536, OPCQuality set to 0, and Quality set to 1.
Querying the start time in exclusive form with the > operator indicates
that a value should not be returned for the query start time if one does
not exist. Querying the start time in inclusive form with the >=
operator indicates that an initial value should be returned.
For example, the following exclusive query statement does not return
an initial value for 2009-01-01 02:00:00.
DateTime > '2009-01-01 02:00:00'
However, the following inclusive query statement does return an

initial value for 2009-01-01 02:00:00.
DateTime >= '2009-01-01 02:00:00'
No special final value is returned.
Delta Retrieval - Handling NULL Values

The initial NULL value after a non-NULL is always returned.
Multiple NULL values are suppressed. The first non-NULL after a
NULL is always returned even if it is the same as the previous
non-NULL value.
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'


A001 2009-09-12 00:17 0.8 192
A001 2009-09-12 00:24 0.0 249
A001 2009-09-12 00:27 0.0 249
A001 2009-09-12 00:28 0.5 192
A001 2009-09-12 00:31 0.0 249
A001 2009-09-12 00:33 0.0 24
A001 2009-09-12 00:35 0.0 24
A001 2009-09-12 00:36 0.5 192
The following is a graphical representation of the data:
The results are:

A001 2009-09-12 00:20 0.8 192
A001 2009-09-12 00:24 NULL 249
A001 2009-09-12 00:28 0.5 192
A001 2009-09-12 00:31 NULL 249
A001 2009-09-12 00:36 0.5 192
00:20 and 00:40 (shown on the chart as dark vertical lines) are
returned by the query.

at 00:20 is returned in the results based on the value of the preceding
data point at 00:16. Because there is no change in the value at 00:27
from the value at 00:24, the data point appears on the chart but does
not appear in the results. Similarly, the two 0.0 values at 00:33 and
00:35 are also excluded from the results. However, the non-NULL
value at 00:36 is returned, even though it is the same as the value at
00:28, because it represents a delta from the preceding (NULL) value
at 00:35.
Full Retrieval
In full retrieval mode, all stored data points are returned, regardless of
whether a value or quality has changed since the last value. This mode
allows the same value and quality pair (or NULL value) to be returned
consecutively with their actual timestamps. It works with all types of
tags.
By using full retrieval in conjunction with storage without filtering
(that is, no delta or cyclic storage mode is applied at the historian), you
can retrieve all values that originated from the plant floor data source
or from another application.
Full retrieval best represents the process measurements recorded by
the Wonderware Historian. However, it creates a higher load for the
server, the network and the client system because a very large number
of records may be returned for longer time periods.
For full retrieval for replicated summary tags on a tier-2 historian, if a
point with doubtful quality is returned as the result of a value
selection from an input summary point with a contained gap, the same
point can be returned again with good quality if the same value is
selected again from the next input summary point that has good
quality.

Full Retrieval - How It Works

The following illustration shows how values are returned for full
retrieval:
Data is retrieved in full mode with a start time of T1 and an end time
of T2. Each dot in the graphic represents an actual data point stored on
the historian. From these points, the following are returned:
• P2, because there is no actual data point at T1
• P3 through P12, because they represent stored data points during
the time period
Full Retrieval - Supported Value Parameters

full retrieval mode. For more information, see the following sections:

Full Retrieval - Query Examples
Query 1
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:36'
AND wwRetrievalMode = 'Full'
Full Retrieval - Initial Values

Full retrieval mode handles initial values the same way as delta mode.
For more information on initial values, see "Delta Retrieval - Initial
Values" on page 167.
Interpolated Retrieval
Interpolated retrieval works like cyclic retrieval, except that
interpolated values are returned if there is no actual data point stored
at the cycle boundary.
This retrieval mode is useful if you want to retrieve cyclic data for
slow-changing tags. For a trend, interpolated retrieval results in a
smoother curve instead of a "stair-stepped" curve. This mode is also
useful if you have a slow-changing tag and a fast-changing tag and
want to retrieve data for both. Finally, some advanced applications
require more evenly spaced values than would be returned if
interpolation was not applied.
By default, interpolated retrieval uses the interpolation setting
specified for the tag in the Wonderware Historian. This means that if a
tag is set to use stair-step interpolation, interpolated retrieval gives
the same results as cyclic retrieval.
Interpolation is only applied to analog tags. If you retrieve data for
other types of tags, stair-step interpolation is used, and the results are
the same as for cyclic retrieval.
Interpolated retrieval is a bit slower than cyclic retrieval. It shares the
limitations of cyclic retrieval in that it may not accurately represent
the stored process data.

Interpolated Retrieval - How It Works

The following illustration shows how the values for an analog tag that
is configured for linear interpolation are returned when using
interpolated retrieval.
.
Data is retrieved in interpolated mode with a start time of TC0 and an

end time of TC2. The resolution has been set in such a way that the
historian returns data for three cycle boundaries at TC0, TC1, and TC2.
P1 to P12 represent actual data points stored on the historian. Of these
points, eleven represent normal analog values, and one, P7, represents
a NULL value due to an I/O Server disconnect, which causes a gap in
the data between P7 and P8.
The green points (P2, PC1, PC2) are returned. The yellow points (P7, P11,
P12) are used to interpolate the returned value for each cycle. The red
points are considered, but not used in calculating the points to return.
Because P2 is located exactly at the query start time, it is returned at
that time without the need for any interpolation. At the following cycle
boundary, point PC1 is returned, which is the NULL value represented
by P7 shifted forward to time TC1. At the last cycle boundary, point PC2
is returned, which has been interpolated using points P11 and P12.

Interpolated Retrieval - Supported Value

Parameters
interpolated retrieval mode. For more information, see the following
sections:
(wwCycleCount)" on page 225
page 228
• "Interpolation Type (wwInterpolationType)" on page 241
• "Timestamp Rule (wwTimestampRule)" on page 244
• "Quality Rule (wwQualityRule)" on page 247
Interpolated Retrieval - Query Examples

To use the interpolated mode, set the following parameter in your
query.
wwRetrievalMode = 'Interpolated'
Query 1
Two analog tags and a discrete tag are retrieved from the History
table, using linear interpolation. The start and end times are offset to
show interpolation of the SysTimeMin tag. The data points at all cycle
boundaries are interpolated for the two analog tags, while the values
returned for the discrete tag are stair-stepped.
SELECT DateTime, TagName, Value, wwInterpolationType FROM
History
WHERE TagName IN ('SysTimeMin', 'ReactTemp', 'SysPulse')
AND DateTime >= '2005-04-11 12:02:30'
AND DateTime <= '2005-04-11 12:06:30'
AND wwRetrievalMode = 'Interpolated'
AND wwInterpolationType = 'Linear'

The results are:

DateTime TagName Value wwInterpolationType
2005-04-11 12:02:30.000 SysTimeMin 2.5 LINEAR
2005-04-11 12:02:30.000 ReactTemp 23.2 LINEAR
2005-04-11 12:02:30.000 SysPulse 1.0 STAIRSTEP
Query 2
If you omit the interpolation type in the query, the historian
determines which interpolation type to use for an analog tag based on
the value of the InterpolationType column in the AnalogTag table, in
conjunction with the InterpolationTypeInteger and
InterpolationTypeReal system parameters.
In the following query both analog tags are set to use the system
default through the AnalogTag table, while the
InterpolationTypeInteger and InterpolationTypeReal system
parameters are set to 0 and 1, respectively. Because SysTimeMin is
defined as a 2-byte integer and ReactTemp is defined as a real we see
that only rows for ReactTemp are interpolated.
SELECT DateTime, TagName, Value, wwInterpolationType FROM
History
WHERE TagName IN ('SysTimeMin', 'ReactTemp', 'SysPulse')
AND DateTime >= '2005-04-11 12:02:30'
AND DateTime <= '2005-04-11 12:06:30'

The results are:

DateTime TagName Value wwInterpolationType
2005-04-11 12:02:30.000 SysTimeMin 2.0 STAIRSTEP
Query 3
SELECT TagName, DateTime, Value, QualityDetail,
wwInterpolationType
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'
AND wwResolution = '10000'


A001 2009-09-12 00:09 0.2 192
A001 2009-09-12 00:15 1.3 192
A001 2009-09-12 00:17 0.8 192
A001 2009-09-12 00:22 0.6 249
A001 2009-09-12 00:26 0.9 192
A001 2009-09-12 00:28 0.0 249
A001 2009-09-12 00:29 0.0 249
A001 2009-09-12 00:33 1.1 192
A001 2009-09-12 00:35 1.6 192
A001 2009-09-12 00:38 0.5 192
A001 2009-09-12 00:42 0.8 192
The results are:

Tagname DateTime Value QualityDetail wwInterpolationType
A001 2009-09-12 00:20 0.8 192 STAIRSTEP
A001 2009-09-12 00:30 NULL 249 STAIRSTEP
A001 2009-09-12 00:40 0.5 192 LINEAR
returned by the query.
at 00:20 is returned in the results based on the preceding data point at
00:17 because the following data point at 00:22 is NULL. Because a
NULL value precedes the 00:30 cycle boundary at 00:29, the NULL is
returned at the cycle boundary. The value at 00:40 is an interpolation
of the data points at 00:38 and 00:42.

Interpolated Retrieval - Initial and Final Values

A value is returned at the start time and end time of the query using
interpolation of the surrounding points.
Interpolated Retrieval - Handling NULL Values

When a NULL value precedes a cycle boundary, that NULL will be
returned at the cycle boundary.
If a valid value precedes a cycle boundary, but is followed by a NULL
value after the cycle boundary, no interpolation will be used and
wwInterpolationType will be set to STAIRSTEP for that value.
"Best Fit" Retrieval

For the “best fit” retrieval mode, the total time for the query is divided
into even sub-periods, and then up to five values are returned for each
sub-period:
• First value in the period
• Last value in the period
• Minimum value in the period, with its actual time
• Maximum value in the period, with its actual time
• The first “exception” in the period (non-Good quality)
“Best fit” retrieval allows for a compromise between delta retrieval
and cyclic retrieval. For example, delta retrieval can accurately
represent a process over a long period of time, as shown in the
following trend. However, to achieve this representation, a large
number of data values must be returned.

If cyclic retrieval is used to retrieve the data, the retrieval is much

more efficient, because fewer values are returned. However, the
representation is not as accurate, as the following trend shows.
“Best fit” retrieval allows for faster retrieval, as typically achieved by

using cyclic retrieval, plus the better representation typically achieved
by using delta retrieval. This is shown in the following trend.
For example, for one week of five-second data, 120,960 values would be
returned for delta retrieval, versus around 300 values for best-fit
retrieval.
Best-fit retrieval uses retrieval cycles, but it is not a true cyclic mode.
Apart from the initial value, it only returns actual delta points. For
example, if one point is both the first value and the minimum value in
a cycle, it is returned only one time. In a cycle where a tag has no
points, nothing is returned.

As in cyclic retrieval, the number of cycles is based on the specified

resolution or cycle count. However, the number of values returned is
likely to be more than one per cycle.
All points are returned in chronological order. If multiple points are to
be returned for a particular timestamp, then those points are returned
in the order in which the corresponding tags were specified in the
query.
The best-fit algorithm is only applied to analog and analog summary
tags. For all other tags, delta results are returned.
Best Fit Retrieval - How It Works

The following illustration shows how the best-fit algorithm selects
points for an analog tag.
Data is retrieved in best-fit mode with a start time of TC0 and an end
historian returns data for two complete cycles starting at TC0 and TC1
and an incomplete cycle starting at TC2. P1 to P12 represent actual data
points stored on the historian. Of these points, eleven represent
normal analog values, and one, P7, represents a NULL value due to an
I/O Server disconnect, which causes a gap in the data between P7 and
P8.
Because P2 is located exactly at the start time, no initial value needs to
be interpolated at the start time. Therefore, point P1 is not considered
at all. All other points are considered, but only the points indicated by
green markers on the graph are returned.

From the first cycle, four points are returned:

• P2 as the initial value of the query, as well as the first value in the
cycle
• P4 as the minimum value in the cycle
• P6 as both the maximum value and the last value in the cycle
• P7 as the first (and only) occurring exception in the cycle
From the second cycle, three points are returned:
• P8 as the first value in the cycle
• P9 as the maximum value in the cycle
• P11 as both the minimum value and the last value in the cycle
• As no exception occurs in the second cycle, none is returned.
Because the tag does not have a point exactly at the query end time,
where an incomplete third cycle starts, the end value PC2 is
interpolated between P11 and P12, assuming that linear interpolation is
used.
Best Fit Retrieval - Supported Value Parameters

best-fit retrieval mode. For more information, see the following
sections:
page 228
Best Fit Retrieval - Query Examples

To use the best fit retrieval mode, set the following parameter in your
query.
wwRetrievalMode = 'BestFit'

Query 1
An analog tag is retrieved over a five-minute period using the best-fit
retrieval mode. The wwResolution parameter is set to 60000, thus
specifying five 1-minute cycles. Within each cycle, the retrieval
sub-system returns the first, minimum, maximum, and last data
points. There are no exception (NULL) points in the time period.
Notice how the points at the query start time and at the query end
time are interpolated, while all other points are actual delta points.
SELECT DateTime, TagName, CONVERT(DECIMAL(10, 1), Value) AS
Value, wwInterpolationType AS IT FROM History
WHERE TagName = 'ReactTemp'
AND DateTime >= '2005-04-11 12:15:00'
AND DateTime <= '2005-04-11 12:20:00'
AND wwRetrievalMode = 'BestFit'
The results are:

DateTime TagName Value IT
(initial, 2005-04-11 12:15:00.000 ReactTemp 40.7 LINEAR
first, min)
(max in 2005-04-11 12:15:38.793 ReactTemp 196.0 STAIRSTEP
interval 1)
(last in 2005-04-11 12:15:58.810 ReactTemp 159.2 STAIRSTEP
interval 1)
(first, max in 2005-04-11 12:16:00.013 ReactTemp 156.9 STAIRSTEP
interval 2)
(last, min in 2005-04-11 12:16:58.857 ReactTemp 16.3 STAIRSTEP
interval 2)
(first, min in 2005-04-11 12:17:00.060 ReactTemp 14.0 STAIRSTEP
interval 3)
(last, max in 2005-04-11 12:17:58.793 ReactTemp 151.0 STAIRSTEP
interval 3)
(first in 2005-04-11 12:18:00.107 ReactTemp 156.0 STAIRSTEP
interval 4)
(max in 2005-04-11 12:18:10.057 ReactTemp 196.0 STAIRSTEP
interval 4)
(last, min in 2005-04-11 12:18:58.837 ReactTemp 106.3 STAIRSTEP
interval 4)
(first, max in 2005-04-11 12:19:00.040 ReactTemp 104.0 STAIRSTEP
interval 5)
(min in 2005-04-11 12:19:31.320 ReactTemp 14.0 STAIRSTEP
interval 5)
(last in 2005-04-11 12:19:58.773 ReactTemp 26.0 STAIRSTEP
interval 5)
(end bounding 2005-04-11 12:20:00.000 ReactTemp 30.7 LINEAR
value)

Best Fit Retrieval - Initial and Final Values

A point will be returned at the query start time and the query end time
for each tag queried, if a point exists for that tag at or after the end
time of the query. The values of the initial and final points will be
determined by interpolating the points preceding and following the
query start or query end time.
Standard interpolation rules will be used to return the initial and final
values. For more information, see "Interpolated Retrieval" on
page 171.
Best Fit Retrieval - Handling NULL Values

When any of the four good points are returned from a cycle that
contains gaps or from an incomplete cycle with the query end time
located inside of the calculation cycle the quality detail of each of the
non-null points returned is modified to alert the user to this fact. This
is done by performing a logical OR operation of the value 4096, which
means partial cycle, onto the existing quality detail. (This is the delta
point equivalent to the use of PercentGood for cyclic.)
Average Retrieval
For the time-weighted average (in short: “average”) retrieval mode, a
time-weighted average algorithm is used to calculate the value to be
returned for each retrieval cycle.
For a statistical average, the actual data values are used to calculate
the average. The average is the sum of the data values divided by the
number of data values. For the following data values, the statistical
average is computed as:
(P1 + P2 + P3 + P4) / 4) = Average

For a time-weighted average, values are multiplied by the time

difference between the points to determine the time-weighted value.
Therefore, the longer a tag had a particular value, the more weight
that value holds in the overall average. The overall average is
determined by adding all of the time-weighted values and then
dividing that number by the total amount of time.
Which values are weighted depends on the interpolation setting of the
tag. For a tag that uses linear interpolation, the midpoints between
values are weighted. For a tag that uses stair-step interpolation, the
earlier of two values is weighted.
For the following data values of a tag that uses linear interpolation,
the time-weighted average is computed as:
(((P1 + P2) / 2) x (T2 - T1)) + (((P2 + P3) / 2) x (T3 - T2)) + (((P3 + P4) / 2)
x (T4 - T3)) / (T4 - T1) = Average
If the same tag uses stair-step interpolation, the time-weighted

average is:
((P1 x (T2 - T1)) + (P2 x (T3 - T2)) + (P3 x (T4 - T3))) / (T4 - T1) =
Average
The SQL Server AVG aggregate is a simple statistical average. Using
the average retrieval mode with a cycle count of 1 returns data much
faster than the AVG aggregate, and usually more accurately due to the
time weighting. The event subsystem also returns a simple statistical
average.
Average retrieval returns one row for each tag in the query for each
cycle. The number of cycles is based on the specified resolution or cycle
count.

The time-weighted average algorithm is only applied to analog and

analog summary tags. If you use average retrieval with other tags, the
results are the same as when using regular cyclic retrieval.
Average Retrieval - How It Works

The following illustration shows how the time-weighted average is
calculated for an analog tag that uses linear interpolation.
Data is retrieved in average mode with a start time of TC0 and an end
historian returns data for two complete cycles starting at TC0 and TC1
and an incomplete cycle starting at TC2. P1 to P9 represent actual data
points stored on the historian. Of these points, eight represent normal
analog values, and one, P5, represents a NULL due to an I/O Server
disconnect, which causes a gap in the data between P5 and P6. Assume
that the query calls for timestamping at the end of the cycle.

Results are calculated as follows:

• The “initial value” returned at the query start time (TC0) is the
time-weighted average of the points in the last cycle preceding TC0.
• The value returned at TC1 is the time-weighted average of the

points in the cycle starting at TC0.
• The value returned at the query end time (TC2) is the

time-weighted average of the points in the cycle starting at TC1.
To understand how the time-weighted average is calculated, observe
the last cycle as an example. First, the area under the curve must be
calculated. This curve is indicated by the red line through P6, P7, P8
and PC2, where PC2 represents the interpolated value at time TC2 using
points P8 and P9. The data gap caused by the I/O Server disconnect
does not contribute anything to this area. If a quality rule of "good" has
been specified, then points with doubtful quality will not contribute
anything to the area, either.
To understand how the area is calculated, consider points P6 and P7.
The area contribution between these two points is calculated by
multiplying the arithmetic average of value P6 and value P7 by the
time difference between the two points. The formula is:
((P6 + P7) / 2) x (T7 - T6)
When the area for the whole cycle has been calculated, the
time-weighted average is calculated by dividing that area by the cycle
time, less any periods within the cycle that did not contribute anything
to the area calculation. The result is returned at the cycle end time.
If you take a closer look at points P4 and P5 in the example, you can see
that the red line through point P4 is parallel to the x-axis. This is
because P5 represents a NULL, which cannot be used to calculate an
arithmetic average. Instead, the value P4 is used for the whole time
period between points P4 and P5.
The area calculation is signed. If the arithmetic average between two
points is negative, then the contribution to the area is negative.

Average Retrieval - Supported Value Parameters

average retrieval mode. For more information, see the following
sections:
page 228
Average Retrieval - Query Examples

To use the average mode, set the following parameter in your query.
wwRetrievalMode = 'Average'
Query 1
The time-weighted average is computed for each of five 1-minute long
cycles.
Note that the wwTimeStampRule parameter is set to "Start" in the
query. This means that the value stamped at 11:18:00.000 represents
the average for the interval 11:18 to 11:19, the value stamped at
11:19:00.000 represents the average for the interval 11:19 to 11:20,
and so on. If no timestamp rule is specified in the query, then the
default setting in the TimeStampRule system parameter is used.
In the first cycle there are no points, so a NULL is returned. In the
second cycle value points are found covering 77.72 percent of the time
as returned in PercentGood. This means that the returned average is
calculated based on 77.72 percent of the cycle time. Because the same
OPCQuality is not found for all the points in the cycle, OPCQuality is
set to Doubtful. In the remaining three cycles, only good points occur,
all with an OPCQuality of 192.

Because no quality rule is specified in the query using the

wwQualityRule parameter, the query uses the default as specified by
the QualityRule system parameter. If a quality rule of Extended is
specified, any point stored with doubtful OPCQuality will be used to
calculate the average, and the point time will therefore be included in
the calculation of PercentGood.
Value, OPCQuality, PercentGood FROM History
AND DateTime >= '2005-04-11 11:18:00'
AND DateTime < '2005-04-11 11:23:00'
AND wwRetrievalMode = 'Average'
The results are:

DateTime TagName Value OPCQuality PercentGood
(cycle 1) 2005-04-11 ReactTemp NULL 0 0.0
11:18:00.000
(cycle 2) 2005-04-11 ReactTemp 70.00 64 77.72
11:19:00.000
(cycle 3) 2005-04-11 ReactTemp 153.99 192 100.0
11:20:00.000
(cycle 4) 2005-04-11 ReactTemp 34.31 192 100.0
11:21:00.000
(cycle 5) 2005-04-11 ReactTemp 134.75 192 100.0
11:22:00.000
Query 2
This query demonstrates the use of the average retrieval mode in a
wide query. Time-weighted average values are returned for the analog
tags ReactTemp and ReactLevel, while regular cyclic points are
returned for the discrete tag, WaterValve.
SELECT * FROM OpenQuery(INSQL,
'SELECT DateTime, ReactTemp, ReactLevel, WaterValve FROM
WideHistory
WHERE DateTime >= "2004-06-07 08:00"
AND DateTime < "2004-06-07 08:05"
AND wwRetrievalMode = "Average"
')

The results are:

DateTime ReactTemp ReactLevel WaterValve
2004-06-07 08:00:00.000 47.71621 1676.69716 1
2004-06-07 08:01:00.000 157.28076 1370.88097 0
2004-06-07 08:02:00.000 41.33734 797.67296 1
2004-06-07 08:03:00.000 122.99525 1921.66771 0
2004-06-07 08:04:00.000 105.28866 606.40488 1
For an additional example, see "Querying Aggregate Data in Different

Ways" on page 324.
Average Retrieval - Initial and Final Values

If wwTimeStampRule = END, the initial value is calculated by
performing an average calculation on the cycle leading up to the query
start time. No special handling is done for the final value.
If wwTimeStampRule = START, the final value is calculated by
performing an average calculation on the cycle following the query end
time. No special handling is done for the initial value.
Average Retrieval - Handling NULL Values

Gaps introduced by NULL values are not included in the average
calculations. The average only considers the time ranges with good
values. TimeGood indicates the total time per cycle that the tags value
was good.
Minimum Retrieval
The minimum value retrieval mode returns the minimum value from
the actual data values within a retrieval cycle. If there are no actual
data points stored on the historian for a given cycle, nothing is
returned. NULL is returned if the cycle contains one or more NULL
values.
resolution or cycle count. However, minimum retrieval is not a true
cyclic mode. Apart from the initial value, all points returned are delta
points.
Minimum retrieval only works with analog tags. For all other tags,
normal delta results are returned.

All returned values are in chronological order. If multiple points are to

be returned for a particular timestamp, they are returned in the order
in which the tags were specified in the query. If the minimum value
occurs several times in a cycle, the minimum value with the earliest
timestamp is returned.
The minimum retrieval mode must use the "<=" operator for the
ending date/time.
Using the minimum retrieval mode with a cycle count of 1 returns the
same results as the SQL Server MIN aggregate; however, the data is
returned much faster.
Minimum Retrieval - How It Works

The following illustration shows how the minimum value is selected
for an analog tag.
This example has a start time of TC0 and an end time of TC2. The
resolution has been set in such a way that the historian returns data
for two complete cycles starting at TC0 and TC1, a “phantom” cycle
starting at TCP, and an incomplete cycle starting at TC2. The phantom
cycle has the same duration as the first cycle in the query period,
extending back in time from the query start time.
For the queried tag, a total of 18 points are found throughout the
cycles, represented by the markers P1 through P18. Of these points, 17
represent normal analog values. The point P13 represents a NULL due
to an I/O Server disconnect, which causes a gap in the data between
P13 and P14.

The minimum value for the “phantom” cycle starting at TCP is

returned as the initial value at TC0. Point P18 is not considered at all
because it is outside of the query time frame. All other points are
considered, but only the points indicated by green markers on the
graph are returned (P10, P13, and P17).
In total, four points are returned:
• P4 as the minimum value of the “phantom” cycle and the initial
point
• P10 as the minimum value in the first cycle
• P13 as the first and only exception occurring in the first cycle
• P17 as the minimum value in the second cycle
No points are returned for the incomplete third cycle starting at the
query end time, because the tag does not have a point exactly at that
time.
If the minimum value of the first cycle is located exactly at the query
start time, both this value and the minimum value of the phantom
cycle are returned.
Minimum Retrieval - Supported Value Parameters

minimum retrieval mode. For more information, see the following
sections:
page 228
Minimum Retrieval - Query Examples

To use the minimum mode, set the following parameter in your query:
wwRetrievalMode = 'Min'
or
wwRetrievalMode = 'Minimum'

Query 1
In this example, an analog tag is retrieved over a five minute period,
using the minimum retrieval mode. Because the wwResolution
parameter is set to 60000, each cycle is exactly one minute long. The
minimum data value is returned from each of these cycles.
Value FROM History
AND DateTime >= '2005-04-11 11:21:00'
AND DateTime <= '2005-04-11 11:26:00'
AND wwRetrievalMode = 'Minimum'
The initial value at the query start time is the minimum value found
in the phantom cycle before the start time of the query.
The results are:
(phantom cycle) 2005-04-11 11:21:00.000 ReactTemp 104.00
(cycle 1) 2005-04-11 11:21:30.837 ReactTemp 14.00
(cycle 2) 2005-04-11 11:22:00.897 ReactTemp 36.00
(cycle 3) 2005-04-11 11:23:59.567 ReactTemp 18.60
(cycle 4) 2005-04-11 11:24:02.083 ReactTemp 14.00
(cycle 5) 2005-04-11 11:25:59.550 ReactTemp 108.60
Query 2
In this example, the minimum retrieval mode is used in a manner
equivalent to using the SQL Server MIN aggregate. Note that the
cycle producing the result is the five-minute phantom cycle just before
the query start time.
SELECT TOP 1 DateTime, TagName, CONVERT(DECIMAL(10, 2), Value)
AS Value FROM History
AND DateTime >= '2005-04-11 11:31:00'
AND DateTime <= '2005-04-11 11:31:00'
The results are:


Query 3
This example shows how the minimum retrieval mode marks the
QualityDetail column to indicate that a minimum value is returned
based on an incomplete cycle. In this case, an incomplete cycle is a
cycle that either contained periods with no values stored or a cycle that
was cut short because the query end time was located inside the cycle.
All values returned for the QualityDetail column are documented in
the QualityMap table in the Runtime database.
SELECT DateTime, TagName, Value, QualityDetail FROM History
AND DateTime >= '2005-04-11 11:18:50'
AND DateTime <= '2005-04-11 11:20:50'
The results are:

DateTime TagName Value QualityDetail
(phantom 2005-04-11 11:18:50.000 SysTimeSec NULL 65536
cycle)
(cycle 1) 2005-04-11 11:19:13.000 SysTimeSec 13.0 4140
Minimum Retrieval - Initial and Final Values

For analog tags, the minimum value of the tag in the cycle leading up
to the query start time is returned with its timestamp changed to the
query start time. If there is no point exactly at the “phantom” cycle
start time, the point leading up to the phantom cycle is also considered
for the minimum calculation.(No adjustments are made to the quality
of the initial point even though the timestamp may have been altered.)
Apart from the initial value, all points returned are delta points. (For
more information on initial values, see "Delta Retrieval - Initial
Values" on page 167.)
If a point occurs exactly on the query end time, that point will be
returned with the partial cycle bit, 4096, set in quality detail. If there
is more than one such point, only the first point will be returned.

Minimum Retrieval - Handling NULL Values and

Incomplete Cycles
The first NULL value in a cycle is returned.
When a minimum value is returned from a cycle that contains gaps
(including a gap extended from the previous cycle) or from an
incomplete cycle with the query end time located inside of the
calculation cycle, the point’s quality detail is modified to flag this. This
indicates a partial cycle, onto the existing quality detail.
As an example of how minimum retrieval mode handles NULLs,
consider the following query:
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'

A001 2009-09-12 00:09 0.2 192
A001 2009-09-12 00:15 1.3 192
A001 2009-09-12 00:17 0.8 192
A001 2009-09-12 00:22 0.5 192
A001 2009-09-12 00:26 0.9 192
A001 2009-09-12 00:28 0.0 249
A001 2009-09-12 00:29 0.0 249
A001 2009-09-12 00:33 1.1 192
A001 2009-09-12 00:35 1.6 192
A001 2009-09-12 00:38 0.5 192
A001 2009-09-12 00:42 0.8 192

The results are:

A001 2009-09-12 00:20 0.2 192
A001 2009-09-12 00:22 0.5 4288
A001 2009-09-12 00:28 NULL 249
A001 2009-09-12 00:38 0.5 4288
returned by the query. The resolution is set at 10,000 milliseconds.
at 00:20 is returned based on the minimum value of the preceding
cycle, which is the data point at 00:09. In the two subsequent cycles,
the minimum values are at 00:22 and 00:38. The quality for these two
values is set to 4288 (4096 + 192). The remaining data points are
excluded because they are not minimums. In addition, the first NULL
at 00:28 is included, but the second NULL (at 00:29) is not.
Maximum Retrieval
The maximum value retrieval mode returns the maximum value from
the actual data values within a retrieval cycle. If there are no actual
data points stored on the historian for a given cycle, nothing is
returned. NULL is returned if the cycle contains one or more NULL
values.
resolution or cycle count. However, maximum retrieval is not a true
cyclic mode. Apart from the initial value, all points returned are delta
points.
Maximum retrieval only works with analog tags. For all other tags,
normal delta results are returned.
in which the tags were specified in the query. If the maximum value
occurs several times in a cycle, the maximum value with the earliest
timestamp is returned.
The maximum retrieval mode must use the "<=" operator for the
ending date/time.
Using the maximum retrieval mode with a cycle count of 1 returns the
same results as the SQL Server MAX aggregate; however, the data is
returned much faster.

Maximum Retrieval - How It Works

The following illustration shows how the maximum value is selected
for an analog tag.
for two complete cycles starting at TC0 and TC1, a “phantom” cycle
starting at TCP, and an incomplete cycle starting at TC2. The phantom
cycle has the same duration as the first cycle in the query period,
extending back in time from the query start time.
For the queried tag, a total of 18 points are found throughout the
cycles, represented by the markers P1 through P18. Of these points, 17
represent normal analog values. The point P13 represents a NULL due
to an I/O Server disconnect, which causes a gap in the data between
P13 and P14.
The maximum value for the “phantom” cycle starting at TCP is
returned as the initial value at TC0. Point P18 is not considered at all
because it is outside of the query time frame. All other points are
considered, but only the points indicated by green markers on the
graph are returned (P12, P13, and P15).
In total, four points are returned:
• P6 as the maximum value of the “phantom” cycle and the initial
point
• P12 as the maximum value in the first cycle
• P13 as the first and only exception occurring in the first cycle
• P15 as the maximum value in the second cycle

No points are returned for the incomplete third cycle starting at the
query end time, because the tag does not have a point exactly at that
time.
If the maximum value of the first cycle is located exactly at the query
start time, this value and the maximum value of the phantom cycle are
returned.
Maximum Retrieval - Supported Value Parameters

maximum retrieval mode. For more information, see the following
sections:
page 228
Maximum Retrieval - Query Examples

To use the maximum mode, set the following parameter in your query:
wwRetrievalMode = 'Max'
or
wwRetrievalMode = 'Maximum'
Query 1
In this example, an analog tag is retrieved over a five-minute period,
using the maximum retrieval mode. Because the wwResolution
parameter is set to 60000, each cycle is exactly one minute long. The
maximum data value is returned from each of these cycles.
Value FROM History
AND DateTime >= '2005-04-11 11:21:00'
AND DateTime <= '2005-04-11 11:26:00'
AND wwRetrievalMode = 'Maximum'
The initial value at the query start time is the maximum value found
in the phantom cycle before the start time of the query.

The results are:

Cycle DateTime TagName Value
(cycle 1) 2005-04-11 11:21:00.853 ReactTemp 101.70
(cycle 2) 2005-04-11 11:22:40.837 ReactTemp 196.00
(cycle 3) 2005-04-11 11:23:00.833 ReactTemp 159.20
(cycle 4) 2005-04-11 11:24:59.613 ReactTemp 146.00
(cycle 5) 2005-04-11 11:25:12.083 ReactTemp 196.00
Query 2
In this example, the maximum retrieval mode is used in a manner
equivalent to using the SQL Server MIN aggregate. Note that the
cycle producing the result is the five-minute phantom cycle just before
SELECT TOP 1 DateTime, TagName, CONVERT(DECIMAL(10, 2), Value)
AS Value FROM History
AND DateTime >= '2005-04-11 11:31:00'
AND DateTime <= '2005-04-11 11:31:00'
The results are:

Query 3
This example shows how the maximum retrieval mode marks the
QualityDetail column to indicate that a maximum value is returned
based on an incomplete cycle. In this case, an incomplete cycle is a
cycle that either contained periods with no values stored or a cycle that
was cut short because the query end time was located inside the cycle.
All values returned for the QualityDetail column are documented in
the QualityMap table in the Runtime database.
SELECT DateTime, TagName, Value, QualityDetail FROM History
AND DateTime >= '2005-04-11 11:19:10'
AND DateTime <= '2005-04-11 11:21:10'

The results are:

DateTime TagName Value QualityDetail
(phantom 2005-04-11 11:19:10.000 SysTimeSec NULL 65536
cycle)
(cycle 1) 2005-04-11 11:19:59.000 SysTimeSec 59 4288
Maximum Retrieval - Initial and Final Values

For analog tags, the maximum value of the tag in the cycle leading up
to the query start time is returned with its timestamp changed to the
query start time. If there is no point exactly at the phantom cycle start
time, the point leading up to the phantom cycle is also considered for
the maximum calculation. No adjustments are made to the quality of
the initial point even though the timestamp may have been altered.
Apart from the initial value, all points returned are delta points. (For
more information on initial values, see "Determining Cycle
Boundaries" on page 288.)
If a point occurs exactly on the query end time, that point is returned
with the partial cycle bit, 4096, set in quality detail. If there is more
than one such point, only the first point is returned.
Maximum Retrieval - Handling NULL Values and

Incomplete Cycles
The first NULL value in a cycle is returned.
When a maximum value is returned from a cycle that contains gaps
(including a gap extended from the previous cycle) or from an
incomplete cycle with the query end time located inside of the
calculation cycle, the point’s quality detail is modified to flag this. This
indicates a partial cycle, onto the existing quality detail.
As an example of how maximum retrieval mode handles NULLs,
consider the following query:
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'

If you run this query against the following sample data:

A001 2009-09-12 00:09 0.2 192
A001 2009-09-12 00:15 1.3 192
A001 2009-09-12 00:17 0.8 192
A001 2009-09-12 00:22 0.5 192
A001 2009-09-12 00:26 0.9 192
A001 2009-09-12 00:28 0.0 249
A001 2009-09-12 00:29 0.0 249
A001 2009-09-12 00:33 1.1 192
A001 2009-09-12 00:35 1.6 192
A001 2009-09-12 00:38 0.5 192
A001 2009-09-12 00:42 0.8 192
The results are:

A001 2009-09-12 00:20 1.3 192
A001 2009-09-12 00:26 0.9 4288
A001 2009-09-12 00:28 NULL 249
A001 2009-09-12 00:35 1.6 4288
returned by the query. The resolution is set at 10,000 milliseconds.

at 00:20 is returned based on the maximum value of the preceding
cycle, which is the data point at 00:15. In the two subsequent cycles,
the maximum values are at 00:26 and 00:35. The quality for these two
values is set to 4288 (4096 + 192). The remaining data points are
excluded because they are not maximums. In addition, the first NULL
at 00:28 is included, but the second NULL (at 00:29) is not.
Integral Retrieval
Integral retrieval calculates the values at retrieval cycle boundaries by
integrating the graph described by the points stored for the tag.
Therefore, it works much like average retrieval, but it additionally
applies a scaling factor. This retrieval mode is useful for calculating
volume for a particular tag. For example, if one of your tags represents
product flow in gallons per second, integral retrieval allows you to
retrieve the total product flow in gallons during a certain time period.
Integral retrieval is a true cyclic mode. It returns one row for each tag
in the query for each cycle. The number of cycles is based on the
specified resolution or cycle count.
Integral retrieval only works with analog tags. For all other tags,
normal cyclic results are returned.

Integral Retrieval - How It Works

Calculating values for a cycle in integral retrieval is a two-step
process:
• First, the historian calculates the area under the graph created by
the data points. This works the same as in average retrieval. For
more information, see "Average Retrieval" on page 182.
• After this area has been found, it is scaled using the value of the
IntegralDivisor column in the EngineeringUnit table. This divisor
expresses the conversion factor from the actual rate to one of units
per second.
For example, if the time-weighted average for a tag during a 1-minute
cycle is 3.5 liters per second, integral retrieval returns a value of 210
for that cycle (3.5 liters per second multiplied by 60 seconds).
Integral Retrieval - Supported Value Parameters

integral retrieval mode. For more information, see the following
sections:
page 228
Integral Retrieval - Query Examples

To use the integral retrieval mode, set the following parameter in your
query.
wwRetrievalMode = 'Integral'

Query 1
In this example, the integral is computed for each of five 1-minute long
cycles. The wwQualityRule parameter is used to ensure that only
points with good quality are used in the computation, which means
that points with doubtful quality are discarded. The rules used to
determine the returned OPCQuality are the same as described for a
time weighted average query.
Flow, OPCQuality, PercentGood FROM History
WHERE TagName = 'FlowRate'
AND DateTime >= '2004-06-07 08:00'
AND DateTime < '2004-06-07 08:05'
AND wwRetrievalMode = 'Integral'
The results are:

DateTime TagName Flow OPCQuality PercentGood
(interval 1) 2004-06-07 FlowRate 2862.97 192 100.0
08:00:00.000
08:01:00.000
08:02:00.000
08:03:00.000
08:04:00.000
Also, the “phantom” cycle affects the integral retrieval mode just as it
does the average retrieval mode. For examples, see "Querying
Aggregate Data in Different Ways" on page 324.
Integral Retrieval - Initial and Final Values

If wwTimeStampRule = END, the initial value is calculated by
performing an integral calculation on the cycle leading up to the query
start time. No special handling is done for the final value.
If wwTimeStampRule = START, the final value is calculated by
performing an integral calculation on the cycle following the query end
time. No special handling is done for the initial value.
Integral Retrieval - Handling NULL Values

Gaps introduced by NULL values are not included in the integral
calculations. The average only considers the time ranges with good
values. TimeGood indicates the total time per cycle that the tags value
was good.

Slope Retrieval
Slope retrieval returns the slope of a line drawn through a given point
and the point immediately before it, thus expressing the rate at which
values change.
This retrieval mode is useful for detecting if a tag is changing at too
great a rate. For example, you might have a temperature that should
steadily rise and fall by a small amount, and a sharp increase or
decrease could point to a potential problem.
The slope retrieval mode can be considered a delta mode. Apart from
the initial value and a value at the query end time, all returned points
are calculated delta points returned with the timestamp of an actual
delta point.
Slope retrieval only works with analog tags. For all other tags, normal
delta results are returned.
in which the tags were specified in the query.
Slope Retrieval - How It Works

The following illustration shows how the slope is calculated for an
analog tag.
This example has a start time of TS and an end time of TE.

For the queried tag, a total of nine points are found, represented by the
markers P1 through P9. Of these points, eight represent normal analog
values. The point P5 represents a NULL due to an I/O Server
disconnect, which causes a gap in the data between P5 and P6.

For every point in the time period, slope retrieval returns the slope of
the line going through that point and the point immediately before it.
For two points P1 and P2 occurring at times T1 and T2, the slope
formula is as follows:
(P2 - P1) / (T2 - T1)
The difference between T1 and T2 is measured in seconds. Therefore,
the returned value represents the change in Engineering Units per
second.
In this example, point P2 is located at the query start time, and
because there is a prior value (P1), the slope of the line through both
points is calculated and returned at time TS. Similarly, slopes are
calculated to be returned at times T3, T4, T7, and T8. The slope is also
calculated for the line through P8 and P9, but that value is returned as
point PTE at the query end time.
For point P6, there is no prior point with which to perform a slope
calculation. Instead, the slope of the flat line going through the point
(that is, the value 0) is calculated. At the time of P5, NULL is returned.
The quality detail and OPC quality returned with a slope point is
always directly inherited from the point that also provides the time
stamp. In this example, this means that point P2 provides the qualities
for the slope point returned at the query start time, TS.
Slope Retrieval - Supported Value Parameters

slope retrieval mode. For more information, see the following sections:
Slope Retrieval - Query Example

To use the slope retrieval mode, set the following parameter in your
query.
wwRetrievalMode = 'Slope'
Query 1
The following query calculates and returns the rate of change of the
ReactTemp tag in °C/second. The initial value in the Quality column at
the query start time shows no value is located exactly at that time, so
the slope returned is the same as the one returned at the next delta
point. (For more information on initial values, see "Determining Cycle
Boundaries" on page 288.)

At 08:01:17.947 the tag has two delta points, so a slope is calculated

and returned for the first point, while a NULL is returned at the
second one with a special QualityDetail of 17, indicating that no slope
can be calculated as it is either plus or minus infinite.
Slope, Quality, QualityDetail FROM History
AND DateTime >= '2005-04-17 08:00'
AND DateTime <= '2005-04-17 08:05'
AND wwRetrievalMode = 'Slope'
The results are:
DateTime TagName Slope Quality QualityDetail

2005-04-17 08:00:00.000 ReactTemp 3.8110 133 192
2005-04-17 08:00:00.510 ReactTemp 3.8110 0 192
2005-04-17 08:00:01.713 ReactTemp 4.1563 0 192
2005-04-17 08:00:02.917 ReactTemp 4.1563 0 192
2005-04-17 08:00:04.230 ReactTemp 3.8081 0 192
2005-04-17 08:00:05.433 ReactTemp 4.1563 0 192
… … …
2005-04-17 08:01:16.743 ReactTemp -1.7517 0 192
2005-04-17 08:01:17.947 ReactTemp -27.0158 0 192
2005-04-17 08:01:17.947 ReactTemp NULL 1 17
2005-04-17 08:01:19.260 ReactTemp -1.7530 0 192
2005-04-17 08:01:20.463 ReactTemp -1.9119 0 192
2005-04-17 08:01:21.667 ReactTemp -1.9119 0 192
2005-04-17 08:01:22.977 ReactTemp -1.7517 0 192
… … …

Slope Retrieval - Initial and Final Values

An initial value is always generated. If a point is stored exactly at the
query start time, the slope is returned as the slope between that point
and the previous point. Otherwise, the slope is calculated using the
slope of the points before and after the query start time.
A final value is always generated. If a point is stored exactly at the
query end time, the slope is returned as the slope between that point
and the previous point. Otherwise, the slope is calculated using the
slope of the points before and after the query end time.
Slope Retrieval - Handling NULL Values

The first NULL following a non-NULL value is returned. Subsequent
NULL values are not. If a point is preceded by a NULL, the slope for
that point will be zero.
Counter Retrieval
Counter retrieval allows you to accurately retrieve the delta change of
a tag’s value over a period of time even for tags that are reset upon
reaching a “rollover value.” The rollover value is defined in the
Wonderware Historian for each tag.
This retrieval mode is useful for determining how much of an item was
produced during a particular time period. For example, you might
have an integer counter that keeps track of how many cartons were
produced. The counter has an indicator like this:
The next value after the highest value that can be physically shown by
the counter is called the rollover value. In this example, the rollover
value is 10,000. When the counter reaches the 9,999th value, the
counter rolls back to 0. Therefore, a counter value of 9,900 at one time
and a value of 100 at a later time means that you have produced 200
units during that period, even though the counter value has dropped
by 9,800 (9,900 minus 100). Counter retrieval allows you to handle this
situation and receive the correct value. For each cycle, the counter
retrieval mode shows the increase in that counter during the cycle,
including rollovers.
Counter retrieval also works with floating point counters, which is
useful for flow meter data. Similar to the carton counter, some flow
meters “roll over” after a certain amount of flow accumulates. For both
examples, the need is to convert the accumulating measure to a “delta
change” value over a given period.

Counter retrieval is a true cyclic mode. It returns one row for each tag
in the query for each cycle. The number of cycles is based on the
specified resolution or cycle count.
The counter algorithm is only applied to analog tags and to discrete
tags. For integer analog tags, the result will be an integer returned as
a float data type. For a real analog tag, the rollover value and the
result may be real values and can include fractional values. If a query
contains tags of other types, then no rows are returned for those tags.
For discrete tags, the rollover value is assumed to be 2.
The rules used to determine the OPCQuality returned with a counter
value are the same as for a time weighted average query. For more
information, see "Quality Rule (wwQualityRule)" on page 247. When a
rollover has occurred in the calculation cycle, a special quality detail of
212 is returned in all non-NULL cases.
CTU counters will default to "signed integer" tags when imported into
the Historian, giving a normal range of -2147483648 to 2147483647
(for a 32-bit integer). In operation, these counters will count up to the
upper limit and "rollover" to the lower limit on the next increment. If
these tags are changed to be "unsigned integers" the normal range will
be 0 to 4294967295 and values will rollover to "0", conforming to the
expected behavior of a tag used with "counter" retrieval. When used
with a 16-bit CTU counter, the same rules apply, but the range of
values is -32768 to 32767 as "signed" and 0 to 65535 for an "unsigned".
Counter Retrieval - How It Works

The following illustration shows how the counter algorithm
determines the count for an analog tag.

for three complete cycles starting at TC0, TC1, and TC2, and an
incomplete cycle starting at TC3.
For the queried tag, a total of twelve points are found throughout the
cycles represented by the markers P1 through P12. Of these points,
eleven represent normal analog values. The point P9 represents a
NULL due to an I/O Server disconnect, which causes a gap in the data
between P9 and P10. Point P12 is not considered because it is outside of
the query time frame.
All points are considered in the counter calculation, but only the
yellow ones are actually used to determine which values to return to
the client. The returned points are PC0, PC1, PC2 and PC3, shown in
green at the top to indicate that there is no simple relationship
between them and any of the actual points.
All cycle values are calculated as the delta change between the cycle
time in question and the previous cycle time, taking into account the
number of rollovers between the two points in time. The counter
algorithm assumes that a rollover occurred if the current value is
lower than the previous value. The initial value at the query start time
(PC1) is calculated the same way, only based on a phantom cycle before
For example, the formula to calculate PC1 is as follows:
PC1 = n * VR + P6 - P1
where:
n = the number of rollovers that have occurred during the cycle
VR = the rollover value for the tag
If either n or VR are equal to zero, PC1 is simply the difference between
the values P1 and P6.
In the case of cycle C2, there is no value at the cycle time, so the NULL
value represented by point P9 is returned. In the case of cycle C3, a
NULL is again returned, because there is no counter value at the
previous cycle boundary to use in the calculation. There must be a full
cycle of values in order for the counter to be calculated.
If a gap is fully contained inside a cycle, and if points occur within the
cycle on both sides of the gap, then a counter value is returned, even
though it may occasionally be erroneous. Zero or one rollovers are
assumed, even though the counter might have rolled over multiple
times.

Calculations for a Manually Reset Counter

If you have a counter that you typically reset manually before it rolls
over, you must set the rollover value for the tag to 0 so that the count
is simply how much change occurred since the manual reset.
For example, assume that you have the following data values for five
consecutive cycle boundaries, and that the value 0 occurs as the first
value within the last cycle:
100, 110, 117, 123, 3
If you set the rollover value to 0, the counter retrieval mode assumes
that the 0 following the value 123 represents a manual reset, and
returns a value of 3 for the last cycle, which is assumed to be the count
after the manual reset. The value 0 itself does not contribute 1 to the
counter value in this case.
If the rollover value is instead set to 200, then the counter retrieval
mode assumes that the value 0 represents a normal rollover, and a
count of 80 is calculated and returned (200 - 123 + 3). In this case, the
value 0 contributes 1 to the counter value, and that is the change from
the value 199 to the value 200.
Counter Retrieval - Supported Value Parameters

integral retrieval mode. For more information, see the following
sections:
page 228
Counter Retrieval - Initial and Final Values

An initial value is returned using the period leading up to the query
start time.
A data point that has a cycle time is used to generate the counter value
for its preceding cycle. A NULL point with cycle time will cause the
preceding cycle to end in a gap and the following cycle to start with a
gap.

Counter Retrieval - Handling NULL Values

If wwQualityRule is configured as OPTIMISTIC, NULL data points
will not be used in calculation. 0.0 will be used as the starting base
value for the query unless the query data starts with a NULL. If the
query starts with a NULL, the value change for the cycle is calculated
from the first actual value in the cycle, rather than 0.
Otherwise, if any points considered in a cycle have UNCERTAIN
quality, the result for that row will also have UNCERTAIN quality.
Any cycle that starts or ends in a gap will have a quality detail of
65536.
The quality detail of DOUBTFUL will be used with the counter result
for the cycles, if a NULL point is considered for the cycle and the
counter result is not NULL.
Counter Retrieval - Handling Illegal Values

If the configured rollover value is larger than 0.0, then the data points
whose values are greater than or equal to the rollover value causes the
counter value for the cycle to be set to 0.0, with
qdIO_FILTEREDPOINT applied to the quality detail.
Similarly, if any data point with a value less than 0.0 is found in a
cycle, the counter value for the cycle is set to 0.0, with
qdIO_FILTEREDPOINT applied to the quality detail.
Counter Retrieval - Query Example

To use the counter mode, set the following parameter in your query.
wwRetrievalMode = 'Counter'
In the following example, the rollover value for the SysTimeSec system
tag is set to 0. In a two-minute time span, the SysTimeSec tag
increments from 0 to 59 two times. The following query returns the
total count within the two-minute time span. The QualityDetail of 212
indicates that a counter rollover occurred during the query time range.
select DateTime, TagName, Value, Quality, QualityDetail as QD
from History
where TagName = 'systimesec'
and DateTime >= '2009-08-13 1:00'
and DateTime < '2009-08-13 1:02'
and wwRetrievalMode = 'counter'
and wwCycleCount = 1
The results are:

DateTime TagName Value Quality QD
2009-08-13 01:00:00.0000000 SysTimeSec 120 0 212

ValueState Retrieval
ValueState retrieval returns information on how long a tag has been in
a particular value state during each retrieval cycle. That is, a
time-in-state calculation is applied to the tag value.
This retrieval mode is useful for determining how long a machine has
been running or stopped, how much time a process spent in a
particular state, how long a valve has been open or closed, and so on.
For example, you might have a steam valve that releases steam into a
reactor, and you want to know the average amount of time the valve
was in the open position during the last hour. ValueState retrieval can
return the shortest, longest, average, or total time a tag spent in a
state, or the time spent in a state as a percentage of the total cycle
length.
When you use ValueState retrieval for a tag in a trend chart, you must
specify a single value state for which to retrieve information.
ValueState retrieval then returns one value for each cycle—for
example, the total amount of time that the valve was in the “open”
state during each 1-hour cycle. This information is suitable for trend
display.
If you don’t specify a state, ValueState retrieval returns one row of
information for each value that the tag was in during each cycle. For
example, this would return not only the time a valve was in the “open”
state, but also the time it was in the “closed” state. This information is
not suitable for meaningful display in a regular trend. You can,
however, retrieve this type of information in a query and view it as a
table.
ValueState retrieval works with integer, discrete, string, and state
summary tags. For other types of tags, no rows are returned. NULL
values are treated like any other distinct state.
The values returned at the query start time are the result of applying
the algorithm to a “phantom” cycle preceding the query range. It is
assumed that the tag value at the start of the cycle is located at that
point in time.
To specify the type of calculation, set the wwStateCalc parameter in
the query. For more information, see "State Calculation
(wwStateCalc)" on page 255.

ValueState Retrieval - How It Works

The following illustration shows how ValueState retrieval returns
values for a discrete tag.
Value
ValueState Retrieval
C0 C1 C2 C3
PC0 PC1 PC2 PC3
ON
OFF Gap
1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19 21 22 23 24 25 26 27 28 29
Time
TC0 TC1 TC2 TC3
incomplete cycle starting at TC3. Time is measured seconds.
A gap in the data occurs in the third cycle due to an I/O Server
disconnect.
The state calculation is based on each cycle, and the values returned at
the query start time are not regular initial values, but are the
resulting values that occur after applying the algorithm to the last
cycle preceding the query range. The returned points are PC0, PC1, PC2
and PC3, shown in green at the top to indicate that there is no simple
relationship between the calculated values and any of the actual
points.
Assume the query is set so that the total time (wwStateCalc =
‘Total’)in the two states are returned. The timestamping is set to use
the cycle end time.
• For TC0, the query returns two rows (one for the “on” state and one
for the “off” state), calculated as a result of the “phantom” cycle
that precedes the query start time. The values have a timestamp of
• For TC1, one row is returned for the “on” state. The “on” state
occurred twice during the cycle--one time for four seconds and
again for two seconds before the cycle boundary, and the total time
returned is six seconds. The state was “off” twice during the cycle
for a total time of four seconds, and one row is returned with that
value.

• For TC2, one row is returned for the “on” state, and one row is
returned for the “off” state. The “on” state occurred for a total of
nine seconds between the cycle boundaries, and the “off” state
occurred for a total of one second.
returned for the “off” state. The “on” state occurred for a total of
four seconds between the cycle boundaries, and the “off” state
occurred for a total of three seconds. An additional row is returned
for the NULL state occurring as a result of the I/O Server
disconnect.
Using the same data, if you queried the total contained time for the
states, the following is returned:
• For TC0, the query returns two values (one for the “on” state and
one for the “off” state), calculated as a result of the “phantom” cycle
the precedes the query start time. The value has a timestamp of
returned for the “off” state. The “on” state occurred one time for
four seconds within the cycle. The two seconds of “on” time that
crosses the cycle boundary does not contribute to the total time.
The state was “off” one time during the cycle for two seconds
completely within the cycle boundary.
• For TC2, the state was not “on” for any contained time between the
cycle. Both occurrences of the “on” state cross over a cycle
boundary, so no rows are returned for this state. One row is
returned for the “off” state. The state was “off” one time during the
cycle for one seconds completely within the cycle boundary.
returned for the “off” state. The state was “on” for a single
contained time of two seconds between the cycle boundaries. The
state was “off” three times during the cycle for three seconds
completely within the cycle boundary. An additional row is
returned for the NULL state occurring as a result of the I/O Server
disconnect. The state was NULL for a total of three seconds. The
I/O Server disconnect that “disrupted” the off state is treated as its
own state, thereby changing what would have been a single “off”
state instance of five seconds into two instances of the “off” state
for one second each.

ValueState Retrieval - Supported Value Parameters

ValueState retrieval mode. For more information, see the following
sections:
page 228
• "State Calculation (wwStateCalc)" on page 255
ValueState Retrieval - Query Examples

To use theValueState retrieval mode, set the following parameter in
your query.
wwRetrievalMode = 'ValueState'
To specify the type of aggregation, set the wwStateCalc parameter in

the query, such as:
wwStateCalc = 'Total'
Be sure that you use the "<=" operator for ending date/time.
Query 1: Minimum Time in State

The following query finds the minimum time-in-state for the
SteamValve discrete tag. Note that minimum times are returned for
each state for both the five-minute phantom cycle before the query
start time and for the single retrieval cycle between 10:00 and 10:05.
SELECT DateTime, TagName, vValue, StateTime, wwStateCalc FROM
History
WHERE TagName IN ('SteamValve')
AND DateTime >= '2005-04-17 10:00:00'
AND DateTime <= '2005-04-17 10:05:00'
AND wwRetrievalMode = 'ValueState'
AND wwStateCalc = 'Min'

The results are:

DateTime TagName vValue StateTime wwStateCalc
2005-04-17 10:00:00.000 SteamValve 0 35359.0 MINIMUM
Query 2: Minimum Time in State for a Single Tag

The following query finds the minimum time-in-state for the
SteamValve discrete tag for the “on” state. Note that minimum times
are returned for each state for both the five-minute phantom cycle
before the query start time and for the single retrieval cycle between
10:00 and 10:05.
History
AND DateTime >= '2005-04-17 10:00:00'
AND DateTime <= '2005-04-17 10:05:00'
AND wwStateCalc = 'Min'
AND State = '1'
The results are:


Query 2
The following query finds the maximum time-in-state for the
SteamValve discrete tag in the same time period as in Query 1. Note
how both the minimum and maximum values for the "1" state are very
similar, while they are very different for the "0" state. This is due to
the "cut-off" effect.
History
AND DateTime >= '2005-04-17 10:00:00'
AND DateTime <= '2005-04-17 10:05:00'
AND wwStateCalc = 'Max'
2005-04-17 10:00:00.000 SteamValve 0 107514.0 MAXIMUM
Query 3
The following query returns the total of time in state for a discrete tag.
In this example, the TimeStampRule system parameter is set to "End"
(the default setting), so the returned values are timestamped at the
end of the cycle. The returned rows represent the time-in-state
behavior during the period starting at 2005-04-13 00:00:00.000 and
ending at 2005-04-14 00:00:00.000.
SELECT DateTime, vValue, StateTime, wwStateCalc FROM History
WHERE DateTime > '2005-04-13 00:00:00.000'
AND DateTime <= '2005-04-14 00:00:00.000'
AND TagName IN ('PumpStatus')
AND wwStateCalc = 'Total'
The results are:

DateTime vValue StateTime wwStateCalc
2005-04-14 00:00:00 NULL 1041674.0 TOTAL
2005-04-14 00:00:00 On 56337454.0 TOTAL
2005-04-14 00:00:00 Off 29020872.0 TOTAL

Query 4
The following query returns the percentage of time in state for a
discrete tag for multiple retrieval cycles. The TimeStampRule system
parameter is set to "End" (the default setting), so the returned values
are timestamped at the end of the cycle. Note that the first row
returned represents the results for the period starting at 2003-07-03
22:00:00.000 and ending at 2003-07-04 00:00:00.000.
The "Percent" time-in-state retrieval mode is the only mode in which
the StateTime column does not return time data. Instead, it returns
percentage data (in the range of 0 to 100 percent) representing the
percentage of time in state.
SELECT DateTime, vValue, StateTime, wwStateCalc FROM History
WHERE DateTime >= '2003-07-04 00:00:00.000'
AND DateTime <= '2003-07-05 00:00:00.000'
AND TagName IN ('PumpStatus')
AND Value = 1
AND wwStateCalc = 'Percent'
The results are:

DateTime vValue StateTime wwStateCalc
2003-07-04 00:00:00 1 50.885 PERCENT
2003-07-04 02:00:00 1 82.656 PERCENT
2003-07-04 04:00:00 1 7.082 PERCENT
2003-07-04 06:00:00 1 7.157 PERCENT
2003-07-04 08:00:00 1 55.580 PERCENT
2003-07-04 10:00:00 1 28.047 PERCENT
2003-07-04 12:00:00 1 47.562 PERCENT
2003-07-04 14:00:00 1 74.477 PERCENT
2003-07-04 16:00:00 1 40.450 PERCENT
2003-07-04 18:00:00 1 78.313 PERCENT
2003-07-04 20:00:00 1 54.886 PERCENT
2003-07-04 22:00:00 1 39.569 PERCENT
2003-07-05 00:00:00 1 50.072 PERCENT

Query 5: Querying State Summary Values

If state summary values are queried and the cycle boundaries match
the summary periods, the ValueState calculations are supported and
return valid results.
If state summary points are queried and the cycle boundaries do not
match the summary periods, the ValueState calculations are
supported, but they return DOUBTFUL (QualityDetail = 64) results.
State summaries are included in the cycle where the summary end
time occurs. This causes results that do not match queries against the
source tag and may cause inaccurate results, such as a total state time
that is greater than the cycle time.
For example, this can occur if SysTimeSec is summarized with a state
summary with one minute resolution, but then queried with 10 second
intervals. In most of the retrieval cycles, there will be no values, but in
the cycle that includes the summary end time (one in six of the
retrieval cycles), all 60 states would be returned with each state
having a state time of 1 second for a total of 60 seconds of state time in
a 10 second retrieval cycle.
ValueState Retrieval - Initial and Final Values

the algorithm to the last cycle preceding the query range.
ValueState Retrieval - Handling NULL Values

NULLs are considered a state and are reported along with the other
states.
RoundTrip Retrieval
RoundTrip retrieval is very similar to ValueState retrieval in that it
performs calculations on state occurrences in the within a cycle period
you specify. However, ValueState retrieval uses the time spent in a
certain state for the calculation, and RoundTrip retrieval uses the time
between consecutive leading edges of the same state for its
calculations.
You can use the RoundTrip retrieval mode for increasing the efficiency
of a process. For example, if a process produces one item per cycle,
then you would want to minimize the time lapse between two
consecutive cycles.

The RoundTrip mode returns a rows for each state in any given cycle.
RoundTrip retrieval only works with integer analog tags, discrete tags,
and string tags. If real analog tags are specified in the query, then no
rows are returned for these tags. RoundTrip retrieval is not applied to
state summary or analog summary tags. NULL values are treated as
any other distinct value and are used to analyze the round trip for
disturbances.
RoundTrip retrieval is supported for the History and
StateWideHistory tables.
Any point on the boundary of the end cycle will be considered to the
next cycle. The point on the boundary of the end query range is not
counted in the calculation except that it is used to indicate that the
previous state is a contained state.
If no roundtrip state is found within the cycle for a supported tag, a
NULL StateTime value is returned. If there is no valid point prior to
the phantom cycle, a NULL state is returned for the phantom cycle.
RoundTrip Retrieval - How It Works

The following illustration shows how RoundTrip retrieval returns
values for a discrete tag.
Value
RoundTrip Retrieval
C0 C1 C2 C3
PC0 Round-trip PC1 PC2 PC3
ON
OFF Gap
Round-trip
1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19 21 22 23 24 25 26 27 28 29
Time
TC0 TC1 TC2 TC3
incomplete cycle starting at TC3. Time is measured seconds.
A gap in the data occurs in the third cycle due to an I/O Server
disconnect.

The state calculation is based on each cycle, and the values returned at
the query start time are not regular initial values, but are the
resulting values that occur after applying the algorithm to the last
cycle preceding the query range. The returned points are PC0, PC1, PC2
and PC3, shown in green at the top to indicate that there is no simple
relationship between the calculated values and any of the actual
points.
Assume the query is set so that the total contained time in the two
states are returned. The timestamping is set to use the cycle end time.
The RoundTrip retrieval mode returns values for states that are
completely contained within the cycle boundaries. The following is
returned:
• For TC0, the query returns two values (one for the “on” state and
one for the “off” state), calculated as a result of the “phantom” cycle
that preceeds the query start time. The value has a timestamp of
returned for the “off” state. The round-trip for the “on” state
occurred one time for four seconds completely within the cycle
boundary. The round-trip for the “off” state occurred one time
during the cycle for five seconds.
• For TC2, a round-trip did not occur for either state within the cycle
boundaries. One NULL row is returned for this cycle.
contained time of two seconds between the cycle boundaries. The
state was “off” one time during the cycle for one second completely
within the cycle boundary. An additional row is returned for the
NULL state occurring as a result of the I/O Server disconnect.
contained time of three seconds between the cycle boundaries. One
row is returned for the “off” state for a total contained time of seven
seconds. (The first round-trip for the “off” state includes the I/O
Server disconnect for a length of four seconds. The second
round-trip has a length of three seconds.) An additional row is
returned for the NULL state occurring as a result of the I/O Server
disconnect, and the returned value is NULL because there is no
round-trip during the cycle for it. The I/O Server disconnect that
“disrupted” the off state is treated as its own state, thereby
changing what would have been a single “off” state instance of five
seconds into two instances of the “off” state for one second each.

RoundTrip Retrieval - Supported Value Parameters

You can use various parameters to adjust the values that must be
returned in the RoundTrip retrieval mode. For more information, see
the following sections:
• "State Calculation (wwStateCalc)" on page 255
RoundTrip Retrieval - Query Examples

To use the RoundTrip retrieval mode, set the following parameter in
your query:
wwRetrievalMode = ‘RoundTrip’
The following queries compare the results between ValueState

retrieval and RoundTrip retrieval.
This first ValueState retrieval query returns the average amount of
time that the 'Reactor1OutletValve' tag is in “on” state and the
average amount of time it is in the “off” state for a single cycle. Any
state changes that occur across the cycle boundaries are not included.
SELECT DateTime, vValue, StateTime
FROM History
WHERE TagName IN ('Reactor1OutletValve')
AND DateTime >= '2009-09-16 12:35:00'
AND DateTime <= '2009-09-16 12:55:00'
The results are:

DateTime vValue StateTime
2009-09-16 12:35:00.0000000 0 215878
2009-09-16 12:35:00.0000000 1 61729
2009-09-16 12:55:00.0000000 1 62827.5
2009-09-16 12:55:00.0000000 0 212856
The first two rows are for the “phantom” cycle leading up to the query
start time and have a timestamp of the query start time.
The second two rows show the average amount of time for each state
and have a timestamp of the query end time (the default).

Compare these results to same basic query that instead uses

RoundTrip retrieval:
SELECT DateTime, vValue, StateTime
FROM History
WHERE TagName IN ('Reactor1OutletValve')
AND DateTime >= '2009-09-16 12:35:00'
AND DateTime <= '2009-09-16 12:55:00'
DateTime vValue StateTime

2009-09-16 12:35:00.0000000 1 277607
2009-09-16 12:35:00.0000000 0 278580
2009-09-16 12:55:00.0000000 0 275683.5
2009-09-16 12:55:00.0000000 1 273845
RoundTrip Retrieval - Initial and Final Values

the algorithm to the last cycle preceding the query range.
RoundTrip Retrieval - Handling NULL Values

Like in the ValueState retrieval mode, the NULL state is treated as a
valid distinct value. This allows you to analyze round trips for
disturbances.
Understanding Retrieval Options

In all retrieval modes, you can adjust which values the historian
returns by specifying retrieval options. The retrieval options are
implemented as special parameters that you set as part of the retrieval
query. This section explains each of these options. For an overview of
which options apply to which retrieval modes, see "Which Options
Apply to Which Retrieval Modes?" on page 223.

Understanding Retrieval Options223
Which Options Apply to Which Retrieval Modes?

The following table shows which retrieval options apply to which
modes. If you specify an option in a mode where it isn’t applicable, the
option is ignored.
(wwInterpolationType)
Cycle Count (X Values
(wwValueDeadband)
(wwTimestampRule)
(wwTimeDeadband)
Spaced Every X ms)
Resolution (Values
Interpolation Type
(wwQualityRule)
State Calculation
Value Deadband
Timestamp Rule
over Equal Time
(wwResolution)
Time Deadband
History Version
(wwStateCalc)
(wwVersion)
Quality Rule
Intervals)
Cyclic Retrieval Y Y Y Y*
Delta Retrieval Y Y Y
Full Retrieval Y
Interpolated Y Y Y Y Y Y
Retrieval
"Best Fit" Retrieval Y Y Y Y Y

Average Retrieval Y Y Y Y Y Y
Minimum Retrieval Y Y Y Y
Maximum Retrieval Y Y Y Y
Integral Retrieval Y Y Y Y Y Y
Slope Retrieval Y Y
Counter Retrieval Y Y Y Y Y
ValueState Retrieval Y Y Y Y Y Y
RoundTrip Retrieval Y Y Y Y Y Y
* (only on Wonderware Historian 9.0 and later)

Using Retrieval Options in a Transact-SQL

Statement
You can retrieve data in the Wonderware Historian extension tables
using normal Transact-SQL code, as well as the specialized SQL time
domain extensions provided by the Wonderware Historian. The
Wonderware Historian extensions provide an easy way to query
time-based data from the history tables. They also provide additional
functionality not supported by Transact-SQL.
Note: The wwParameters extension is reserved for future use. The

wwRowCount parameter is still supported, but is deprecated in favor of
wwCycleCount.
The extensions are implemented as "virtual" columns in the extension

tables. When you query an extension table, you can specify values for
these column parameters to manipulate the data that will be returned.
You will need to specify any real-time extension parameters each time
that you execute the query.
For example, you could specify a value for the wwResolution column in
the query so that a resolution is applied to the returned data set:
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwRetrievalMode = 'cyclic'
Because the extension tables provide additional functionality that is

not possible in a normal SQL Server, certain limitations apply to the
Transact-SQL supported by these tables. For more information, see
"Wonderware Historian OLE DB Provider Unsupported Syntax and
Limitations" on page 144.
Although the Microsoft SQL Server may be configured to be
case-sensitive, the values for the virtual columns in the extension
tables are always case-insensitive.
Note: You cannot use the IN clause or OR clause to specify more than
one condition for a time domain extension. For example, "wwVersion
IN ('original', 'latest')" and "wwRetrievalMode = 'Delta'
OR wwVersion = 'latest'" are not supported.


Cycle Count (X Values over Equal Time Intervals)

(wwCycleCount)
In retrieval modes that use cycles, the cycle count determines the
number of cycles for which data is retrieved. The number of actual
return values is not always identical with this cycle count. In “truly
cyclic” modes (Cyclic, Interpolated, Average, and Integral), a single
data point is returned for every cycle boundary. However, in other
cycle-based modes (Best Fit, Minimum, Maximum, Counter,
ValueState, and RoundTrip), multiple or no data points may be
returned for a cycle, depending on the nature of the data.
The length of each cycle (the “resolution” of the returned values) is
calculated as follows:
DC = DQ / (n – 1)
Where DC is the length of the cycle, DQ is the duration of the query,
and n is the cycle count.
Instead of specifying a cycle count, you can specify the resolution. In
that case, the cycle count is calculated based on the resolution and the
query duration. For more information, see Resolution (Values Spaced
Every X ms) (wwResolution).
This option is relevant in the following retrieval modes:

The application of the cycle count also depends on whether you are
querying a wide table. If you are querying the History table, the cycle
count determines the number of rows returned per tag. For example, a
query that applies a cycle count of 20 to two tags will return 40 rows of
data (20 rows for each tag). If you are querying a WideHistory table,
the cycle count specifies the total number of rows to be returned,
regardless of how many tags were queried. For example, a query that
applies a cycle count of 20 to two tags returns 20 rows of data.
Values chosen:
• If wwResolution and wwCycleCount are not specified, then a
default of 100 cycles are chosen.
• If wwResolution and wwCycleCount are set to 0, then a default of
100000 cycles are chosen.
• If wwResolution and wwCycleCount are both set, then
wwCycleCount is ignored.
• If wwCycleCount is specified and is less than 0, then a default of
100 cycles are chosen.
• For ValueState retrieval, if the start time of the cycle is excluded,
no states are returned for the first cycle.
• For ValueState retrieval, if the end time of the cycle is excluded, no
states are returned for the last cycle.
Cycle Count - Query Examples

The following queries demonstrate the cycle count behavior if applied
to a single tag or to multiple tags in the same query.
Query Using a Single Tag

FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime < '2001-12-09 11:36'

The results are:

2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:00.200 SysTimeSec 0
2001-12-09 11:35:00.400 SysTimeSec 0
2001-12-09 11:35:00.600 SysTimeSec 0
.
.
.
2001-12-09 11:35:59.200 SysTimeSec 59
2001-12-09 11:35:59.400 SysTimeSec 59
2001-12-09 11:35:59.600 SysTimeSec 59
2001-12-09 11:35:59.800 SysTimeSec 59
Query Using Multiple Tags

FROM History
WHERE TagName IN ('SysTimeMin','SysTimeSec')
AND DateTime >= '2001-12-09 11:35'
AND DateTime < '2001-12-09 11:36'
The results are:

2001-12-09 11:35:00.000 SysTimeMin 35
2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:00.200 SysTimeMin 35
2001-12-09 11:35:00.200 SysTimeSec 0
2001-12-09 11:35:00.400 SysTimeMin 35
2001-12-09 11:35:00.400 SysTimeSec 0
2001-12-09 11:35:00.600 SysTimeMin 35
2001-12-09 11:35:00.600 SysTimeSec 0
.
.
.
2001-12-09 11:35:59.200 SysTimeMin 35
2001-12-09 11:35:59.200 SysTimeSec 59
2001-12-09 11:35:59.400 SysTimeMin 35
2001-12-09 11:35:59.400 SysTimeSec 59
2001-12-09 11:35:59.600 SysTimeMin 35

2001-12-09 11:35:59.600 SysTimeSec 59

2001-12-09 11:35:59.800 SysTimeMin 35
2001-12-09 11:35:59.800 SysTimeSec 59
Notice that the values of the two tags are mixed together in the same
column.
Resolution (Values Spaced Every X ms)

(wwResolution)
In retrieval modes that use cycles, the resolution is the sampling
interval for retrieving data, that is, the length of each cycle.
The number of cycles, therefore, depends on the time period and the
resolution:
number of cycles = time period / resolution
The number of actual return values is not always identical with this
cycle count. In “truly cyclic” modes (Cyclic, Interpolated, Average, and
Integral), a single data point is returned for every cycle boundary.
However, in other cycle-based modes (Best Fit, Minimum, Maximum,
Counter, and ValueState), multiple or no data points may be returned
for a cycle, depending on the nature of the data.
Note: The rowset is guaranteed to contain one row for each tag in the
normalized query at every resolution interval, regardless of whether a
physical row exists in history at that particular instance in time. The
value contained in the row is the last known physical value in history, at
that instant, for the relevant tag.
Instead of specifying a resolution, you can specify the cycle count

directly. In that case, the resolution is calculated based on the cycle
count and the query duration. For more information, see "Cycle Count
(X Values over Equal Time Intervals) (wwCycleCount)" on page 225.

For delta retrieval, you can achieve similar results by using a time
deadband. For more information, see "Time Deadband
(wwTimeDeadband)" on page 232.
Resolution - Query Examples

The following query returns rows that are spaced at 2 sec (2000 msec)
intervals over the requested time period. Data is retrieved cyclically.
FROM History
WHERE TagName IN ('SysTimeMin','SysTimeSec')
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:36'
The results are:

2001-12-09 11:35:00.000 SysTimeMin 35
2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:02.000 SysTimeMin 35
2001-12-09 11:35:02.000 SysTimeSec 2
2001-12-09 11:35:04.000 SysTimeMin 35
2001-12-09 11:35:04.000 SysTimeSec 4
2001-12-09 11:35:06.000 SysTimeMin 35
.
.
.
2001-12-09 11:35:54.000 SysTimeMin 35
2001-12-09 11:35:54.000 SysTimeSec 54
2001-12-09 11:35:56.000 SysTimeMin 35
2001-12-09 11:35:56.000 SysTimeSec 56
2001-12-09 11:35:58.000 SysTimeMin 35
2001-12-09 11:35:58.000 SysTimeSec 58
2001-12-09 11:36:00.000 SysTimeMin 36
2001-12-09 11:36:00.000 SysTimeSec 0

About "Phantom" Cycles

The phantom cycle is the name given to the cycle that leads up to the
query start time. It is used to calculate which initial value to return at
the query start time for all retrieval modes. Some retrieval modes use
the phantom cycle to simply find the last known value prior to the
query start time, whereas other retrieval modes use the entire cycle to
calculate aggregates. The different uses of the phantom cycle can be
seen in the following table.
Cycles not defined, but

similar simple use of
Simple use of time before query start Phantom cycle used to
phantom cycle time calculate aggregates
Cyclic Delta Min

Interpolated Full Max
Best Fit Slope Average
Integral
Counter
ValueState
RoundTrip
It’s common to expect a single aggregate row returned for a particular

time interval. You can accomplish this in several ways.
The following example is querying for hourly averages. It returns a
single row time stamped at the query start time. If the query included
the query end point by including an equal sign for it, the query would
also have returned an additional row at the query end time.
SELECT DateTime, Value, Quality, QualityDetail, OPCQuality
FROM History
AND DateTime >= '2009-10-16 08:00:00'
AND DateTime < '2009-10-16 09:00:00'
AND wwRetrievalMode = 'Avg'
The results are:

DateTime Value Quality QualityDetail OPCQuality
2009-10-16 08:00:00.0000000 29.5 0 192 192
What may be confusing in this example is the calculation of the

average in the returned row for the phantom cycle leading up to the
query start time. The query specifies a positive one hour time interval
between the query start time and the query end time. You may
therefore expect that the calculated and returned average should be
for the specified interval.

However, the time difference between start and end time in the above
query is actually not required because the resolution has been
provided explicitly (wwResolution = 36000000). If the query specified
an end time equal to the specified start time and if it included the
equal sign for the end time, the query would still return the same
single row of data.
SELECT DateTime, Value, Quality, QualityDetail as QD, OPCQuality
FROM History
AND DateTime >= '2007-12-11 08:00:00'
AND DateTime <= '2007-12-11 09:00:00'
The results are:

DateTime Value Quality QD OPCQuality
2009-10-16 08:00:00.0000000 29.5 0 192 192
This second example also asks for hourly averages and it also returns
only a single row of data stamped at the query start time. This query,
however, must specify a time difference between the start and end
time, because the resolution is not explicitly defined in the query.
As in the preceding query, the specified interval and cycle count of 1
may look like the returned row has been calculated for the specified
interval, but the returned row is once again for the phantom cycle
leading up to the start time.
The StartDateTime makes it easier to see which time interval was
used to calculate the returned aggregate. This column returns the time
stamp of the beginning of the cycle used for the aggregate calculation.
The time stamp is always returned in accordance with the specified
time zone and always has the same offset as the time stamp returned
in the DateTime column, even when the two time stamps are on
different sides of a DST change.
Assuming results are timestamped at the end of the cycle (as is done
by default when wwTimeStampRule is set to END), the initial rows in
the examples above would return a DateTime equal to '2009-10-16
08:00:00', and the StartDateTime column would return '2009-10-16
07:00:00' making it easy to interpret the result.
If instead the query were to ask for results time stamped at the
beginning of the cycle with wwTimeStampRule set to START, the
initial rows in the same examples would still return a DateTime equal
to '2009-10-16 08:00:00', but the time stamp has now been shifted in
accordance with the time stamp request. The result is therefore
calculated for the specified time interval between 8 a.m. and 9 a.m. In
this example, the new StartDateTime column would return the same
time stamp as the DateTime column, '2009-10-16 08:00:00', again
making it easier to interpret the result.

For retrieval modes for which cycles are defined, the StartDateTime
column returns the cycle start time. These modes are:
• Cyclic
• Interpolated
• BestFit
• Min
• Max
• Average
• Integral
• Counter
• ValueState
• RoundTrip
In the remaining retrieval modes, the StartDateTime column returns
the same time stamp as the DateTime column.
For an additional example, see "Querying Aggregate Data in Different
Ways" on page 324.
Time Deadband (wwTimeDeadband)

A time deadband for retrieval controls the time resolution of data
returned in delta mode. Any value changes that occur within the time
deadband are not returned.
Time deadbands can be applied to analog, discrete, and string tags.
The deadband “base value” is reset each time that a value is returned,
so that the last value returned acts as the basis for the deadband.

The following illustration shows an example of applying a time

deadband:
Data is retrieved for the time period starting with TS and ending with
TE. All points in the graphic represent data values stored on the
historian. The grey areas represent the time deadband, which starts
anew with every returned value. Only the green points (P2, P4, P7, P8,
P9, P11) are returned. The other points are not returned because they
fall within a deadband.
Time Deadband - Query Examples

To apply a time deadband, set the wwTimeDeadband parameter in
your query.
The following queries return data values for the analog tag
'SysTimeSec'.
Query 1
This query specifies to only return data that changed during a 5
second time deadband.
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:37'
AND wwTimeDeadband = 5000
The results are:

2001-12-09 11:35:00.000 SysTimeSec 0

2001-12-09 11:35:06.000 SysTimeSec 6

2001-12-09 11:35:12.000 SysTimeSec 12
2001-12-09 11:35:18.000 SysTimeSec 18
2001-12-09 11:35:24.000 SysTimeSec 24
2001-12-09 11:35:30.000 SysTimeSec 30
2001-12-09 11:35:36.000 SysTimeSec 36
2001-12-09 11:35:42.000 SysTimeSec 42
2001-12-09 11:35:48.000 SysTimeSec 48
2001-12-09 11:35:54.000 SysTimeSec 54
2001-12-09 11:36:00.000 SysTimeSec 0
2001-12-09 11:36:06.000 SysTimeSec 6
2001-12-09 11:36:12.000 SysTimeSec 12
2001-12-09 11:36:18.000 SysTimeSec 18
2001-12-09 11:36:24.000 SysTimeSec 24
2001-12-09 11:36:30.000 SysTimeSec 30
2001-12-09 11:36:36.000 SysTimeSec 36
2001-12-09 11:36:42.000 SysTimeSec 42
2001-12-09 11:36:48.000 SysTimeSec 48
2001-12-09 11:36:54.000 SysTimeSec 54
2001-12-09 11:37:00.000 SysTimeSec 0
Query 2
millisecond time deadband.
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:37'
The results are:

2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:05.000 SysTimeSec 5
2001-12-09 11:35:10.000 SysTimeSec 10
2001-12-09 11:35:15.000 SysTimeSec 15
2001-12-09 11:35:20.000 SysTimeSec 20
2001-12-09 11:35:25.000 SysTimeSec 25
2001-12-09 11:35:30.000 SysTimeSec 30
2001-12-09 11:35:35.000 SysTimeSec 35

2001-12-09 11:35:40.000 SysTimeSec 40

2001-12-09 11:35:45.000 SysTimeSec 45
2001-12-09 11:35:50.000 SysTimeSec 50
2001-12-09 11:35:55.000 SysTimeSec 55
2001-12-09 11:36:00.000 SysTimeSec 0
2001-12-09 11:36:05.000 SysTimeSec 5
2001-12-09 11:36:10.000 SysTimeSec 10
2001-12-09 11:36:15.000 SysTimeSec 15
2001-12-09 11:36:20.000 SysTimeSec 20
2001-12-09 11:36:25.000 SysTimeSec 25
2001-12-09 11:36:30.000 SysTimeSec 30
2001-12-09 11:36:35.000 SysTimeSec 35
2001-12-09 11:36:40.000 SysTimeSec 40
2001-12-09 11:36:45.000 SysTimeSec 45
2001-12-09 11:36:50.000 SysTimeSec 50
2001-12-09 11:36:55.000 SysTimeSec 55
2001-12-09 11:37:00.000 SysTimeSec 0
Query 3
millisecond time deadband.
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:36'
The results are:

2001-12-09 11:35:00.000 SysTimeSec 0
2001-12-09 11:35:00.000 SysTimeMin 35
2001-12-09 11:35:03.000 SysTimeSec 3
2001-12-09 11:35:06.000 SysTimeSec 6
2001-12-09 11:35:09.000 SysTimeSec 9
2001-12-09 11:35:12.000 SysTimeSec 12
2001-12-09 11:35:15.000 SysTimeSec 15
2001-12-09 11:35:18.000 SysTimeSec 18
2001-12-09 11:35:21.000 SysTimeSec 21
2001-12-09 11:35:24.000 SysTimeSec 24
2001-12-09 11:35:27.000 SysTimeSec 27

2001-12-09 11:35:30.000 SysTimeSec 30

2001-12-09 11:35:33.000 SysTimeSec 33
2001-12-09 11:35:36.000 SysTimeSec 36
2001-12-09 11:35:39.000 SysTimeSec 39
2001-12-09 11:35:42.000 SysTimeSec 42
2001-12-09 11:35:45.000 SysTimeSec 45
2001-12-09 11:35:48.000 SysTimeSec 48
2001-12-09 11:35:51.000 SysTimeSec 51
2001-12-09 11:35:54.000 SysTimeSec 54
2001-12-09 11:35:57.000 SysTimeSec 57
2001-12-09 11:36:00.000 SysTimeSec 0
2001-12-09 11:36:00.000 SysTimeMin 36
Value Deadband (wwValueDeadband)

A value deadband for retrieval controls the value resolution of data
returned in delta mode. Any data values that change less than the
specified deadband are not returned. The deadband is a percentage of
a tag’s full scale in engineering units.
The deadband “base value” is reset each time that a value is returned,
so that the last value returned acts as the basis for the deadband.
Changes in quality will force a value to be returned even if the value
deadband has not been met.
The following illustration shows an example of applying a value
deadband:

Data is retrieved for the time period starting with TS and ending with
TE. All points in the graphic represent data values stored on the
historian. The grey areas represent the value deadband, which starts
anew with every returned value. Only the green points (P2, P5, P6, P7,
P9, P10, P11) are returned. The other points are not returned because
they fall within a deadband.
Value Deadband - Query Examples

The following queries return data values for the analog tag
'SysTimeSec'. The minimum engineering unit for 'SysTimeSec' is 0,
and the maximum engineering unit is 59.
Query 1
This query specifies to return only data that changed by more than 10
percent of the tag's full engineering unit range. Using a value
deadband of 10 percent equates to an absolute change of 5.9 for
'SysTimeSec.'
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:37'
AND wwValueDeadband = 10
The results are:

DateTime Value
2001-12-09 11:35:00.000 0
2001-12-09 11:35:06.000 6
2001-12-09 11:35:12.000 12
2001-12-09 11:35:18.000 18
2001-12-09 11:35:24.000 24
2001-12-09 11:35:30.000 30
2001-12-09 11:35:36.000 36
2001-12-09 11:35:42.000 42
2001-12-09 11:35:48.000 48
2001-12-09 11:35:54.000 54
2001-12-09 11:36:00.000 0
2001-12-09 11:36:06.000 6
2001-12-09 11:36:12.000 12
2001-12-09 11:36:18.000 18
2001-12-09 11:36:24.000 24
2001-12-09 11:36:30.000 30

2001-12-09 11:36:36.000 36
2001-12-09 11:36:42.000 42
2001-12-09 11:36:48.000 48
2001-12-09 11:36:54.000 54
2001-12-09 11:37:00.000 0
Query 2
This query specifies to only return data that changed by more than 5
percent of the tag's full engineering unit range. Using a value
deadband of 5 percent equates to an absolute change of 2.95 for
'SysTimeSec.'
FROM History
AND DateTime >= '2001-12-09 11:35'
AND DateTime <= '2001-12-09 11:37'
The results are:

DateTime Value
2001-12-09 11:35:00.000 0
2001-12-09 11:35:03.000 3
2001-12-09 11:35:06.000 6
2001-12-09 11:35:09.000 9
2001-12-09 11:35:12.000 12
2001-12-09 11:35:15.000 15
2001-12-09 11:35:18.000 18
2001-12-09 11:35:21.000 21
2001-12-09 11:35:24.000 24
2001-12-09 11:35:27.000 27
2001-12-09 11:35:30.000 30
2001-12-09 11:35:33.000 33
2001-12-09 11:35:36.000 36
2001-12-09 11:35:39.000 39
2001-12-09 11:35:42.000 42
2001-12-09 11:35:45.000 45
2001-12-09 11:35:48.000 48
2001-12-09 11:35:51.000 51
2001-12-09 11:35:54.000 54
2001-12-09 11:35:57.000 57
2001-12-09 11:36:00.000 0

2001-12-09 11:36:03.000 3
2001-12-09 11:36:06.000 6
2001-12-09 11:36:09.000 9
2001-12-09 11:36:12.000 12
2001-12-09 11:36:15.000 15
2001-12-09 11:36:18.000 18
2001-12-09 11:36:21.000 21
2001-12-09 11:36:24.000 24
2001-12-09 11:36:27.000 27
2001-12-09 11:36:30.000 30
2001-12-09 11:36:33.000 33
2001-12-09 11:36:36.000 36
2001-12-09 11:36:39.000 39
2001-12-09 11:36:42.000 42
2001-12-09 11:36:45.000 45
2001-12-09 11:36:48.000 48
2001-12-09 11:36:51.000 51
2001-12-09 11:36:54.000 54
2001-12-09 11:36:57.000 57
2001-12-09 11:37:00.000 0
History Version (wwVersion)

The Wonderware Historian allows you to overwrite a stored tag value
with later versions of the value. The original version of the value is
still maintained, so that effectively, multiple versions of the tag value
exist at the same point in time.
When retrieving data, you can specify whether to retrieve the
originally stored version or the latest version that is available. To do
this, set the history version option to “Original” for the original version
or “Latest” for the latest available version. If you do not specify the
version, the latest version is returned.
To distinguish between a latest value and an original value, the
historian returns a special QualityDetail value of 202 for a latest point
with good quality.
This option is relevant in all retrieval modes.

History Version - Query Example

For example:
SELECT TagName, DateTime, Value, wwVersion
FROM History
WHERE TagName IN ('SysTimeHour', 'SysTimeMin')
AND DateTime >= '2001-12-20 0:00'
AND DateTime <= '2001-12-20 0:05'
AND wwVersion = 'Original'
The results are:

TagName DateTime Value wwVersion
SysTimeMin 2001-12-20 00:00:00.000 0 ORIGINAL
SysTimeHour 2001-12-20 00:00:00.000 0 ORIGINAL
When retrieving the latest version, the wwVersion parameter always

returns with a value of LATEST for all values, even though many of
the values may actually be the original values that came from the I/O
Server. To distinguish between an actual latest value and an original
value, a special QualityDetail of 202 is returned for a good, latest
point.
For example:
SELECT DateTime, Value, Quality, QualityDetail, OPCQuality,
wwVersion FROM History
WHERE TagName IN ('PV')
AND DateTime >= '2005-04-17 11:35:00'
AND DateTime <= '2005-04-17 11:36:00'
AND wwVersion = 'Latest'

The results are:

DateTime Value Quality QualityDetail OPCQuality wwVersion
2005-04-17 12.5 0 192 192 LATEST
11:35:00.000
2005-04-17 17.3 0 192 192 LATEST
11:35:15.000
2005-04-17 34.0 0 202 192 LATEST
11:35:30.000
2005-04-17 43.1 0 192 192 LATEST
11:35:45.000
2005-04-17 51.2 0 192 192 LATEST
11:36:00.000
Interpolation Type (wwInterpolationType)

For various retrieval modes, you can control how analog tag values at
cycle boundaries are calculated if there is no actual value stored at
that point in time. The options are as follows:
• Stairstep: No interpolation occurs. The value at the cycle boundary
is assumed to be the same value as the last stored value before the
cycle boundary.The last known point is returned with the given
cycle time. If no valid value can be found, a NULL is returned.

• Linear: The historian calculates a new value at the cycle boundary

by interpolating between the last stored value before the boundary
and the first stored value after the boundary. If either of these
values is NULL, it returns the last stored value before the
boundary.
Expressed as a formula, Vc is calculated as:
Vc = V1 + ((V2 - V1) * ((Tc - T1) / (T2 - T1)))
The type of data that you want to retrieve usually determines the
interpolation type to use. For example, if you have a thermocouple, the
temperature change is linear, so it’s best to use linear interpolation. If
you have a tag that contains discrete measurements, such as a set
point, then you probably want to use stair-stepped values. In general,
it is recommended that you use linear interpolation as the general
setting, and use stair-stepped values for the exceptions.

The quality of an interpolated point is determined by the

wwQualityRule setting. For more information, see "Quality Rule
(wwQualityRule)" on page 247.
The interpolation type can be set on three levels:
• The Wonderware Historian system-wide setting. The system-wide
setting must be either stair-step or interpolated. For more
information, see "System Parameters" on page 36. This setting is
configured using the Wonderware Historian Configuration Editor.
• The individual analog tag setting. You can configure an individual
analog tag to use the system-wide setting or either stair-stepped
values or linear interpolation. The individual tag setting will
override the system-wide setting. This setting is configured using
the Wonderware Historian Configuration Editor.
• The setting for the wwInterpolationType parameter in the query.
This setting overrides any other setting for all tags in the query.
The wwInterpolationType parameter is dynamically used both for
input for the query, when you need to override the individual tag
settings, and for output for each individual row to show whether a
particular row value was calculated using linear interpolation
(returned as "LINEAR") or if it is a stair-stepped value (returned as
"STAIRSTEP").
To force a query to always use linear interpolation whenever
applicable, specify the following in the query:
To force a query to always return stair-stepped values, specify the

following in the query:
AND wwInterpolationType = 'StairStep'

Timestamp Rule (wwTimestampRule)

For various cycle-based retrieval modes, you can control whether the
returned values are timestamped at the beginning or at the end of
each cycle.
To force a query to timestamp results at the start of a cycle, specify the
To force a query to timestamp results at the end of a cycle, specify the

AND wwTimeStampRule = 'End'
If you include the wwTimeStampRule column in your SELECT

statement, it will show which timestamp rule has been applied for the
individual row, if applicable.
The options are as follows:
• Start: The value for a given cycle is stamped with the cycle start
time. For example, in the following illustration of a cyclic query,
the following values are returned at the cycle boundaries:
• At TC0: P7, because it falls on the cycle boundary. In cyclic
mode, if there is a value right on the cycle boundary, it is
considered to belong to the cycle before the boundary. In this
case, this is the cycle starting at TC0 and ending at TC1, and
because the Start timestamp rule is used, the value is
timestamped at TC0.
• At TC1: P11, because it is the last value in the cycle starting at
TC1 and ending at TC2
• At TC2: The last value in the “phantom” cycle starting at TC2

• End: The value for a given cycle is stamped with the cycle end
time. For example, in the following illustration of a cyclic query,
the following values are returned at the cycle boundaries:
• At TC0: P1, because it is the last value in the “phantom” cycle
ending at TC0. Because the End timestamp rule is used, the
value is timestamped at TC0.
• At TC1: P7, because it falls on the cycle boundary. In cyclic
mode, if there is a value right on the cycle boundary, it is
considered to belong to the cycle before the boundary. In this
case, this is the cycle starting at TC0 and ending at TC1, and
because the End timestamp rule is used, the value is
timestamped at TC1.
• At TC2: P11, because it is the last value in the cycle ending at
TC2
• Server default: Either Start or End is used, depending on the

system parameter setting on the Wonderware Historian.
• Cyclic Retrieval (only for Wonderware Historian 9.0 and later)

Time Zone (wwTimeZone)

For Wonderware Historian version 8.0 and later, all history data is
stored in Coordinated Universal Time (UTC). The wwTimeZone
extension allows you to specify the time zone to be used for the
timestamps of the returned data values. The retrieval subsystem will
convert the timestamps to local time in the specified time zone.
The wwTimeZone extension may be assigned any of the values stored
in the TimeZone column of the TimeZone table in the Runtime
database. In addition to specifying the name of the timezone in the
wwTimeZone parameter, you can also specify the TimeZoneID (as a
string). For example, on a typical US English system, specifying "
wwTimeZone = 'Mountain Standard Time' " and " wwTimeZone =
'64'" yields the same result.
The TimeZone table is repopulated at every system startup from
Microsoft operating system registry entries, and will therefore reflect
the time zones available from the server operating system, including
any new or custom time zones which might be added by operating
system service packs or installed software.
The retrieval subsystem will automatically correct for daylight savings
time in the requested time zone. When computing daylight savings
and time zone parameters, the settings of the server operating system
are used. The retrieval sub-system does not provide any means for
using client-side settings.
If wwTimeZone is not specified, the time zone for retrieval defaults to
the time zone of the Wonderware Historian computer.
For example:
SELECT TagName, DateTime, Value, wwTimeZone
FROM History
WHERE TagName IN ('SysTimeHour', 'SysTimeMin')
AND DateTime >= '2001-12-20 0:00'
AND DateTime <= '2001-12-20 0:05'
AND wwTimeZone = 'W. Europe Standard Time'
The results are:

TagName DateTime Value wwTimeZone
SysTimeMin 2001-12-20 00:00:00.000 0 W. Europe Standard Time
SysTimeHour 2001-12-20 00:00:00.000 15 W. Europe Standard Time

If you are using date/time functions and the wwTimeZone parameter,

you will need to use the faaTZgetdate() function.
Quality Rule (wwQualityRule)

For various retrieval modes, you can explicitly exclude values with
uncertain quality from data retrieval in modes that calculate return
values.
Where applicable, the quality rule can be used to specify whether
values with certain characteristics are explicitly excluded from
consideration by data retrieval. This parameter will override the
setting of the QualityRule system parameter. Valid values are GOOD,
EXTENDED, or OPTIMISTIC.
• A quality rule of GOOD means that data values with uncertain
(64) OPC quality are not used in the retrieval calculations and are
ignored. Values with bad QualityDetail indicate gaps in the data.
• A quality rule of EXTENDED means that data values with both
good and uncertain OPC quality are used in the retrieval
calculations. Values with bad QualityDetail indicate gaps in the
data.
• A quality rule of OPTIMISTIC means that calculations that
include some good and some NULL values do not cause the overall
calculations to return NULL.
You can apply a quality rule to all retrieval modes.
The OPTIMISTIC setting for the quality rule lets you retrieve
information that is possibly incomplete but may nevertheless provide
better results in the counter and integral retrieval modes where the
calculation cycle contains data gaps. This setting calculates using the
last known good value prior to the gap (if possible). The logic for
determining the quality of the points returned remains unchanged in
both retrieval modes. The integral retrieval mode is an exception to
this where the integral is scaled up to cover gaps. For more
information, see "Integral Retrieval" on page 200.

The following figure shows a counter retrieval situation in which three

of the four shown cycle boundaries are located in data gaps. Without
using OPTIMISTIC, counter queries would return a NULL at all cycle
boundaries because the mode needs valid good values at each end of
the cycle calculate a precise difference.
If the query were to specify OPTIMISTIC, the counter mode will

always return rows with numeric counter values and good quality.
These rows may or may not be precise. The PercentGood column of the
row returns the percentage of time in each cycle in which retrieval was
able to find values stored with good quality, so if the PercentGood is
anything less than 100, then the returned row may be incorrect.
Quality is returned as uncertain if percent good is not 100 percent.
Now look at the counter values that are returned using OPTIMISTIC
quality in the preceding illustration The query skips the value to be
returned at the first cycle boundary, because there is not enough
information about the cycle prior to that boundary. At the second cycle
boundary, the value 0 will be returned, because there was a gap in the
data for the entire first cycle. In the second cycle, there are two points,
P1 and P2. The query uses P2 as the end value of the cycle and infers a
start value of the cycle from P1. At the third cycle boundary, Tc2, the
query returns P2 – P1. Similarly, at the last cycle boundary, the query
returns P4 – P3.

For the integral retrieval mode, the query does not summarize data for
gaps because there is no way to know which value to use for the
summarization. However, if the query specifies OPTIMISTIC quality,
the query uses the last known good value for the summarization in the
gap. As described for the counter retrieval example, the PercentGood
column also expresses the quality of the calculated value in integral
retrieval, so if the PercentGood is anything less than 100, then the
returned row may be incorrect.
Quality Rule - Query Examples

To force a query to exclude points with doubtful OPC quality, specify
the following in the query:
To force a query to use points with both good and doubtful OPC
quality, specify the following in the query:
AND wwQualityRule = 'Extended'
If you include the wwQualityRule column in a SELECT statement, it

will show which quality rule was used for the individual row, if
applicable.
You can combine OPC qualities in a query. For example, if you
combine a mixture of good OPC qualities (such as 192 to 219), a good
OPC quality (192) will be returned as a combined result.
SELECT TagName, DateTime, Value, QualityDetail, OPCQuality,
wwRetrievalMode
FROM History
WHERE TagName = 'I0R5'
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'

Tagname DateTime Resolution QualityDetail
I0R5 2009-09-12 00:07 2 193
I0R5 2009-09-12 00:14 3 195
I0R5 2009-09-12 00:22 0 196
I0R5 2009-09-12 00:25 1 199
I0R5 2009-09-12 00:27 0 200
I0R5 2009-09-12 00:29 2 207
I0R5 2009-09-12 00:33 3 215
I0R5 2009-09-12 00:36 0 216
I0R5 2009-09-12 00:39 1 219

The results are:

Tagname DateTime Value QualityDetail OPCQuality wwRetrievalMode
I0R5 2009-09-12 00:20 2.6 192 192 AVERAGE
I0R5 2009-09-12 00:30 1.0 192 192 AVERAGE
I0R5 2009-09-12 00:40 1.6 192 192 AVERAGE
Similarly, if you combine a mixture of doubtful OPC qualities, a

doubtful OPC quality (64) will be returned as the combined OPC
quality.
wwRetrievalMode
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'
AND wwRetrievalMode = 'Integral'

I0R5 2009-09-12 00:07 2 65
I0R5 2009-09-12 00:14 3 68
I0R5 2009-09-12 00:22 0 71
I0R5 2009-09-12 00:25 1 74
I0R5 2009-09-12 00:27 0 79
I0R5 2009-09-12 00:29 2 80
I0R5 2009-09-12 00:33 3 88
I0R5 2009-09-12 00:36 0 92
I0R5 2009-09-12 00:39 1 64
The results are:

I0R5 00:20 26.0 64 64 INTEGRAL
I0R5 00:30 10.0 64 64 INTEGRAL
I0R5 00:40 16.0 64 64 INTEGRAL

When you combine the same OPC quality then that OPC quality will
be returned. However, when there is no good point in a cycle for cyclic
modes such as Integral, Average, Counter, or AnalogSummary, the
returned NULL value will have an OPC quality of 0 and a Quality
Detail of 65536, regardless of combined qualities.
SELECT TagName, StartDateTime, EndDateTime, OPCQuality,
PercentGood, wwRetrievalMode, first
FROM AnalogSummaryHistory
WHERE TagName = 'F0R5'
AND StartDateTime >= '2009-09-12 00:20'
AND EndDateTime <= '2009-09-12 00:40'

F0R5 2009-09-12 00:07 1.6 78
F0R5 2009-09-12 00:14 3.1 78
F0R5 2009-09-12 00:22 0.2 78
F0R5 2009-09-12 00:25 0.8 78
F0R5 2009-09-12 00:27 0.4 78
F0R5 2009-09-12 00:29 2.2 78
F0R5 2009-09-12 00:33 3.3 78
F0R5 2009-09-12 00:36 0.3 78
F0R5 2009-09-12 00:39 1.2 78
The results are:

Tagname StartDate EndDate OPCQuality PercentGood wwRetrievalMode first
Time Time
F0R5 2009-09-12 2009-09-12 78 100 CYCLIC 0.200
00:20 00:30
F0R5 2009-09-12 2009-09-12 78 100 CYCLIC 3.300
00:30 00:40

wwRetrievalMode
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'


F0R5 2009-09-12 00:07 1.6 15
F0R5 2009-09-12 00:14 3.1 15
F0R5 2009-09-12 00:22 0.2 15
F0R5 2009-09-12 00:25 0.8 15
F0R5 2009-09-12 00:27 0.4 15
F0R5 2009-09-12 00:29 2.2 15
F0R5 2009-09-12 00:33 3.3 15
F0R5 2009-09-12 00:36 0.3 15
F0R5 2009-09-12 00:39 1.2 15
The results are:

F0R5 2009-09-12 00:20 NULL 65536 0 AVERAGE
F0R5 2009-09-12 00:30 NULL 65536 0 AVERAGE
F0R5 2009-09-12 00:40 NULL 65536 0 AVERAGE
When you combine a mixture of good, bad, and uncertain OPC

qualities, a doubtful OPC quality (64) will be returned as a combined
result.
wwRetrievalMode
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'
AND wwQualityRule = 'Optimistic'

F0R5 2009-09-12 00:07 1.6 15
F0R5 2009-09-12 00:14 3.1 69
F0R5 2009-09-12 00:22 0.2 78
F0R5 2009-09-12 00:25 0.8 200
F0R5 2009-09-12 00:27 0.4 15
F0R5 2009-09-12 00:29 2.2 92
F0R5 2009-09-12 00:33 3.3 88
F0R5 2009-09-12 00:36 0.3 199
F0R5 2009-09-12 00:39 1.2 196

The results are:

F0R5 2009-09-12 00:20 2.012 64 64 AVERAGE
F0R5 2009-09-12 00:30 0.820 64 64 AVERAGE
F0R5 2009-09-12 00:40 1.751 64 64 AVERAGE
For RoundTrip, StateSummary, and ValueState modes, the OPC

qualities are only combined with the same state in a cycle. If the state
only occurs once in a cycle, then the qualities of that state will be
returned. The returned NULL state will always have an OPC quality
of 0 and Quality Detail of 65536. The same qualities are returned for a
state that has no roundtrip in RoundTrip mode.
StateTime
FROM History
WHERE TagName = 'I001'
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'
AND wwStateCalc = 'MaxContained'

I001 2009-09-12 00:12 1 90
I001 2009-09-12 00:15 2 65
I001 2009-09-12 00:22 1 85
I001 2009-09-12 00:23 2 75
I001 2009-09-12 00:26 1 75
I001 2009-09-12 00:29 2 70
The results are:

Tagname DateTime Value QualityDetail OPC-Quality StateTime
I001 2009-09-12 00:20 NULL 65536 0 NULL
I001 2009-09-12 00:20 1.0 90 90 NULL
I001 2009-09-12 00:20 2.0 65 65 NULL
I001 2009-09-12 00:20 1.0 64 64 4000
I001 2009-09-12 00:20 2.0 64 64 6000

The returned Quality Detail is the same as OPC quality unless there is
special flag for certain indication for example when there is indication
for role over in counter mode.
SELECT TagName, DateTime, Value, QualityDetail, OPCQuality
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'

I0R5 2009-09-12 00:07 2 218
I0R5 2009-09-12 00:14 3 218
I0R5 2009-09-12 00:22 0 218
I0R5 2009-09-12 00:25 1 218
I0R5 2009-09-12 00:27 0 218
I0R5 2009-09-12 00:29 2 218
I0R5 2009-09-12 00:33 3 218
I0R5 2009-09-12 00:36 0 218
I0R5 2009-09-12 00:39 1 218
The results are:

Tagname DateTime Value QualityDetail OPCQuality
I0R5 2009-09-12 00:20 2.6 218 218
I0R5 2009-09-12 00:30 1.0 218 218
I0R5 2009-09-12 00:40 1.6 218 218
For Interpolated mode only the returned row with Linear

wwInterpolationType will have combined qualities.
wwRetrievalMode, wwInterpolationType
FROM History
AND DateTime >= '2009-09-12 00:20'
AND DateTime <= '2009-09-12 00:40'


I0R5 2009-09-12 00:07 2 193
I0R5 2009-09-12 00:14 3 195
I0R5 2009-09-12 00:22 0 196
I0R5 2009-09-12 00:25 1 199
I0R5 2009-09-12 00:27 0 200
I0R5 2009-09-12 00:29 2 207
I0R5 2009-09-12 00:33 3 215
I0R5 2009-09-12 00:36 0 216
I0R5 2009-09-12 00:39 1 219
The results are:

Tagname DateTime Value QualityDetail OPCQuality
I0R5 2009-09-12 00:20 0.8 192 192
I0R5 2009-09-12 00:30 2.3 192 192
I0R5 2009-09-12 00:40 1.0 192 219
Note: Cyclic, Full, Delta, Maximum, Minimum, and BestFit do not have
combined qualities; therefore, the rules are not applied to these modes.
State Calculation (wwStateCalc)

The state calculation setting applies to ValueState and RoundTrip
retrieval.
For ValueState retrieval, you can choose the type of state calculation
(aggregation) to be performed on the data:
• Minimum: The shortest amount of time that the tag has been in
each unique state.
• Maximum: The longest amount of time that the tag has been in
each unique state.
• Average: The average amount of time that the tag has been in
each unique state.
• Total: The total amount of time that the tag has been in each
unique state.
• Percent: The total percentage of time that the tag has been in each
unique state.

• MinContained: The shortest amount of time each tag has been in

each unique state for each cycle, disregarding the occurrences that
are not fully contained with the calculation cycle.
• MaxContained: The longest amount of time that the tag has been
in each unique state for each cycle, disregarding the occurrences
that are not fully contained with the calculation cycle.
• AvgContained: The average amount of time that the tag has been
in each unique state for each cycle, disregarding the occurrences
that are not fully contained with the calculation cycle.
• TotalContained: The total amount of time that the tag has been in
• PercentContained: The percentage of time that the tag has been in
All results except Percent are in milliseconds. Percent is a percentage
typically between 0.0 and 100.0. The percentage can be higher than
100 in certain circumstances.
The nature of the data and how you set the cycle count determines
whether you should use a “contained” version of the aggregation. The
calculations apply to each unique value state that the tag was in
during each retrieval cycle for the query. The total and percent
calculations are always exact, but the minimum, maximum, and
average calculations are subject to “arbitrary” cycle boundaries that do
not coincide with the value changes. Therefore, non-intuitive results
may be returned. This is most apparent for slowly-changing tags
queried over long cycles.
For example, a string tag that assumes only two distinct values
changing every 10 minutes is queried with a cycle time of two hours.
Going into a cycle, the value (state) at the cycle boundary is recorded.
If the value then changes a short while into the cycle, the state found
at the cycle start time will most likely end up being the minimum
value. Likewise, the state at the cycle end time is cut short at the cycle
end time. The two cut-off occurrences in turn skew the average
calculation.
For RoundTrip retrieval, you can only specify the following types of
state calculations (aggregations) to be performed on the data. The
calculations are for each unique state within each retrieval cycle for
the query.
• MinContained. The shortest time span between consecutive
leading edges of any state that occurs multiple times within the
cycle, while disregarding state occurrences that are not fully
contained inside of the cycle.

• MaxContained. The longest time span between consecutive

contained inside of the cycle.
• AvgContained. The average time span between consecutive
contained inside of the cycle. (This is the default.)
• TotalContained. The total time span between consecutive leading
edges of any state that occurs multiple times within the cycle while
disregarding state occurrences that are not fully contained inside
of the cycle.
• PercentContained. The percentage of the cycle time spent in time
span between consecutive leading edges for a state that occurs
multiple times within the cycle while disregarding value
occurrences that are not fully contained inside of the cycle.
Analog Value Filtering (wwFilter)

You can use the following analog filters for all retrieval modes:
• Statistical removal of outliers

• Analog to discrete conversion
• Zero around a base value
These filters are applied in a query to retrieve data from the History
table, WideHistory table, or StateWideHistory table. These filter only
apply to analog tags. All other types of tags, including analog
summary tags, are not supported.
You need to specify a filter name in the virtual column wwFilter, with
or without an override, to the set of parameters that are defined for the
specified filter. The filters are specified as C-like functions:
parentheses are always required, even when you choose not to override
the default parameters by passing no parameters.
The default value for the wwFilter column is ‘NoFilter’. If the query
does not specify the wwFilter element at all, or if its default value is
not overridden, then no filter is applied.
When you use the analog filters in a query that uses wwQualityRule,
wwQualityRule is applied first and wwFilter is applied later. You can
only use one filter per query.

Statistically Removing Outliers (SigmaLimit)

This analog filter removes outliers from a set of analog points based on
the assumption that the distribution of point values in the set is a
normal distribution.
The following illustration shows an example of outliers.
Outliers
You can filter outliers by specifying a filter called ‘SigmaLimit()’. This

filter has one parameter defined for specifying the value of n. This
parameter is of type double. If the parameter is omitted, then a default
parameter of 2.0 is used.
When this filter is specified in any retrieval mode, a time weighted
mean, ì (mu), and time weighted standard deviation, ó (sigma), are
found for each analog tag for the entire query range including
phantom cycles if any, and points falling outside of the range [ì - nó, ì +
nó] are removed from the point set before the points are processed
further. In other words, the value will be filtered out if value > ì + nó or
value < ì – nó.
Time weighted standard deviation is calculated as:
Math.Sqrt( (integralOfSquares - 2 * timeWeightedAverage * integral +
totalTime * timeWeightedAverage *
timeWeightedAverage)/totalTime);
This is the single pass equivalent to the formula:
N
s 2 weighted = ? wi (x -µ *)
i
2
i=1

Ranges where the value is NULL are excluded from these calculations.
A cyclic query example using a ‘SigmaLimit()’ filter without specifying
the n value would look like this:
SELECT DateTime, Value, wwFilter
FROM History
WHERE TagName = ('TankLevel')
AND DateTime >= '2008-01-15 15:00:00'
AND DateTime <= '2008-01-15 17:00:00'
AND wwFilter = 'SigmaLimit()'
Not specifying the n-value as done here is the same as specifying

‘SigmaLimit(2)’. The result set might look like this:
DateTime Value wwFilter
2008-01-15 15:00:00.000 34.56 SigmaLimit()
2008-01-15 16:00:00.000 78.90 SigmaLimit()
2008-01-15 17:00:00.000 12.34 SigmaLimit()
If the first value would be filtered out by the SigmaLimit filter, the
value will be replaced with the time weighted mean.
Converting Analog Values to Discrete Values

(ToDiscrete)
The analog to discrete conversion filter allows you to convert value
streams for any analog tag in the query tag list into discrete value
streams. The filter can be used with all the retrieval modes.
To convert analog values to discrete values, specify the ToDiscrete()
filter in the wwFilter column. This filter has two parameters:
Parameter Valid Values Default Value
CutoffValue any double value 0.0

Operator >, >=, or <= >
The following are supported syntaxes.

• ToDiscrete()
• ToDiscrete(2)
• ToDiscrete(2, >=)

The following are unsupported syntaxes.

• ToDiscrete(2,)
• ToDiscrete(, >=)
• ToDiscrete(>=)
The cutoff value holds the value that signifies the boundary between
values that are to be interpreted as OFF and values that are to be
interpreted as ON.
The operator parameter controls the value range relative to the cutoff
value to convert to the ON value and vice versa.
NULLs encountered in the analog value stream are copied unchanged
to the discrete value stream. The quality of each discrete point is
copied from the analog point that causes its production. However, the
quality detail for values modified with this filter will have the
QualityDetail flag 0x2000 (value changed by filter) set. For example,
consider the following ValueState query:
SELECT DateTime, vValue, StateTime, wwFilter
FROM History
AND DateTime >= '2008-01-15 15:00:00'
AND DateTime <= '2008-01-15 17:00:00'
AND wwStateCalc = 'MinContained'
AND wwFilter = 'ToDiscrete(29, >)'
Here the operator is specified as >, so values greater than but not
including 29 are internally converted to ON, whereas values from 0 to
29 are converted to OFF. This query could return the following rows:
DateTime vValue StateTime wwFilter
2008-01-15 15:00:00.000 0 30000 ToDiscrete(29, >)
2008-01-15 15:00:00.000 1 30000 ToDiscrete(29, >)
2008-01-15 17:00:00.000 0 30000 ToDiscrete(29, >)
2008-01-15 17:00:00.000 1 30000 ToDiscrete(29, >)

The values returned in the StateTime column show that the shortest
amount of time that SysTimeSec had values equivalent to either ON or
OFF and remained in that state was 30 seconds. A similar RoundTrip
query would look like this:
SELECT DateTime, vValue, StateTime, wwFilter
FROM History
AND DateTime >= '2008-01-15 15:00:00'
AND DateTime <= '2008-01-15 17:00:00'
AND wwStateCalc = 'MaxContained'
AND wwFilter = 'ToDiscrete(29, <=)'
Here the operator is specified as <=, so the resulting conversion is

exactly opposite to that performed in the previous query. Now values
smaller than or equal to 29 are internally converted to ON, whereas
values from 30 to 59 are converted to OFF. This query could return the
following rows:
DateTime vValue StateTime wwFilter
2008-01-15 15:00:00.000 0 60000 ToDiscrete(29, <=)
2008-01-15 15:00:00.000 1 60000 ToDiscrete(29, <=)
2008-01-15 17:00:00.000 0 60000 ToDiscrete(29, <=)
2008-01-15 17:00:00.000 1 60000 ToDiscrete(29, <=)
The values returned in the StateTime column now show that the
longest amount of time found between roundtrips for both the OFF
and the ON state within the 2-hour cycles was 60 seconds.
Using the ToDiscrete() filter is similar to using edge detection for
event tags. Edge detection returns the actual value with a
timestamp in history for when a value matched a certain criteria.
The ToDiscrete() filter returns either a 1 or 0 to show that the
criteria threshold was crossed. The ToDiscrete() filter is more
flexible, however, in the following ways:
• You can use it with delta and full retrieval.
• You can combine it with “time-in-state” calculations to determine
how long a value is above a certain threshold or the duration
between threshold times.
Use the ToDiscrete() filter if you are mostly interested in when
something occurred, and not necessarily the exact value of the event.
For more information on edge detection, see "Edge Detection for
Events (wwEdgeDetection)" on page 266.

"Zeroing" around a Base Value (SnapTo)

This analog filter lets you force values in a well-defined range around
one or more base values to “snap to” that base value. For example, you
can use this filter when a tank is known to be empty, but the tag that
stores the tank level returns a “noisy” value close to zero.
The filter can be used with all retrieval modes, but its main benefits
are in the aggregate retrieval modes: average, integral, minimum, and
maximum.
To zero values around the base value, specify the SnapTo() filter in the
wwFilter column of the query.
The syntax for this filter is:
SnapTo([tolerance[,base_value_1[, base_value_2]…]])
This filter has two parameters:
Parameter Valid Values Default Value
Tolerance any double value 0.01

BaseValue zero, one, or up to 100 single base value of 0.0
comma-separated
double values
The following are supported syntaxes.

• SnapTo() – Same as SnapTo(0.01, 0.0)
• SnapTo(3.7) – Same as SnapTo(3.7, 0.0)
• SnapTo(3,) – Syntax Error
• SnapTo(,0) – Syntax error
• SnapTo(,) – Syntax error
• SnapTo(3, 4, -5) – Tolerance=3, Base Values 4 and -5.
When the Snap to filter is specified, point values falling inside any of
the ranges [Base value – Tolerance, Base value + Tolerance] will be
forced to the base value before the point goes into further retrieval
processing. The result is undefined if the base value +/- tolerance
exceeds the range of the double data type. The range is calculated
using this expression:
If (x <= Base value + Tolerance AND x >= Base value – Tolerance)
x = Base value
where x is the value of the point then

If ranges overlap, the first matching base value will be used.

A query example from the History table looks like this:

SELECT DateTime, Value, wwFilter
FROM History
WHERE TagName = ('TankLevel')
AND DateTime >= '2008-01-15 15:00:00'
AND DateTime <= '2008-01-15 17:00:00'
AND wwFilter = 'SnapTo(0.01, 0, 1000)'
The following rows might be returned:

DateTime Value wwFilter
2008-01-15 15:00:00.000 0 SnapTo(0.01, 0, 1000)
2008-01-15 16:00:00.000 875.66 SnapTo(0.01, 0, 1000)
2008-01-15 17:00:00.000 502.3 SnapTo(0.01, 0, 1000)
When a value is snapped, the QualityDetail bit flag 0x2000 is set.

If the filter syntax is not correct, a syntax error is returned and no
rows are returned.
Selecting Values for Analog Summary Tags

(wwValueSelector)
For an analog summary tag, multiple summarized values can be
stored in the historian for a single summarization period. When you
query analog summary data, a single value, time, and quality (VTQ)
must first be extrapolated from the summarized values.
You set the value selector in the query to specify which summarized
value to return. The possible values are as follows:
Value Selector Setting Value Returned Timestamp Returned
AUTO The retrieval mode determines The retrieval mode

the value. See the following table determines the timestamp.
for how AUTO applies to the See the following table for
value selection. This is the how AUTO applies to the
default value. value selection. This is the
default value.
FIRST The first value that occurs within The actual timestamp of the
the summary period. first value occurrence within
the summary period.
LAST The last value that occurs within The actual timestamp of the
the summary period. last value occurrence within
the summary period.

Value Selector Setting Value Returned Timestamp Returned
MIN or MINIMUM The first minimum value that The actual timestamp of the
occurs within the summary first minimum value
period. occurrence within the
summary period.
MAX or MAXIMUM The first maximum value that The actual timestamp of the
occurs within the summary first maximum value
period. occurrence within the
summary period.
AVG or AVERAGE A time-weighted average The summary period start
calculated from values within the time.
summary period.
INTEGRAL An integral value calculated from The summary period start
values within the summary time.
period.
STDDEV or A standard deviation calculated The summary period start
STANDARDDEVIATION from values within the summary time.
period.
The following table describes the value to be considered if the value

selector is set to AUTO:
Retrieval Mode Analog Summary Behavior
Cyclic The last value within the summary period is used. The actual timestamp
of the last value occurrence within the summary period is used.
Delta The last value within the summary period is used. The actual timestamp
Full The last value within the summary period is used. The actual timestamp
Interpolated The retrieval mode determines the appropriate value to return. See the
following table for how AUTO applies to the value selection. This is the
default value.
Best Fit The first, last, min, and max points from analog summaries are all
considered as analog input points. Best fit analysis is done with these
points. If the analog summary percentage good is not 100%, the cycle is
considered to have a NULL.

Retrieval Mode Analog Summary Behavior
Average The averages of analog summaries are calculated using the values from
the Average column of the AnalogSummaryHistory table. Interpolation
type is ignored for analog summary values, and STAIRSTEP
interpolation is always used. PercentGood is calculated by considering the
TimeGood of each analog summary.
If cycle boundaries do not exactly match the summary periods of the
stored analog summaries, the averages and time good are calculated by
prorating the average and time good values for the portion of the time the
summary period overlaps with the cycle. Quality will be set to 64
(uncertain) if cycle boundaries do not match summary periods.
If the QualityDetail of any analog summary considered for a cycle is
uncertain (64), the resulting quality is set to 64.
Minimum The first minimum value within the summary period is used. The actual
timestamp of the first minimum value occurrence within the summary
period is used.
Maximum The first maximum value within the summary period is used. The actual
timestamp of the first maximum value occurrence within the summary
period is used.
Integral The integrals of analog summaries are calculated using the Integral
column of the AnalogSummaryHistory table. Interpolation type is ignored
for analog summary values, and STAIRSTEP interpolation is always
used. PercentGood is calculated by considering the TimeGood of each
analog summary.
If cycle boundaries do not exactly match the summary periods of the
stored analog summaries, the integrals and time good are calculated by
prorating the integral and time good values for the portion of the time the
summary period overlaps with the cycle. Quality is set to 64 (uncertain) if
cycle boundaries do not match summary periods.
If the QualityDetail of any analog summary considered for a cycle is
uncertain (64), the resulting quality will be set to 64.
Slope The last value within the summary period is used. The actual timestamp
ValueState Cannot be used with analog summary data. No values are returned.
Counter Cannot be used with analog summary data. No values are returned.
RoundTrip Cannot be used with analog summary data. No values are returned.

For an analog summary tag, if any of the data within a summary

period has an OPCQuality other than Good, the OPCQuality returned
will be Uncertain. This is true even for summary values that are not
calculated, such as first, last, minimum, maximum, and so on. For
example, if the OPCQuality for a last value is actually Good, but there
was a I/O Server disconnect during the summary calculation period,
the OPCQuality for the last value is returned as Uncertain. A
QualityDetail of 202 is used to distinguish between the original point
and the latest point.
Edge Detection for Events (wwEdgeDetection)

An event is the moment at which a detection criterion is met on
historical tag values in the Wonderware Historian. At a basic level,
anything that can be determined by looking at stored data can be used
as an event.
When detecting events, it is useful to pinpoint rows in a result set
where the satisfaction of criteria in a WHERE clause changed from
true to false, or vice versa. For example, you may want to know when
the level of a tank went above 5 feet. The WHERE clause in a query for
this example might be TANKLEVEL > 5. As the tank level approaches
5 feet, the criterion does not return true. Only when the level crosses
the line from not satisfying the criterion to satisfying it, does the event
actually occur. This imaginary "line" where the change occurs is called
the edge.
Over a period of time, there may be many instances where the criteria
cross the "edge" from being satisfied to not satisfied, and vice-versa.
The values on either side of this "edge" can be detected if you configure
your event tag to include this information. There are four possible
options for edge detection: none, leading, trailing, or both. You will get
differing results based on which option you use:
Option Results
None Returns all rows that successfully meet the criteria;

no edge detection is implemented at the specified
resolution.
Leading Returns only rows that are the first to successfully
meet the criteria (return true) after a row did not
successfully meet the criteria (returned false).
Trailing Returns only rows that are the first to fail the criteria
(return false) after a row successfully met the criteria
(returned true).
Both All rows satisfying both the leading and trailing
conditions are returned.

Edge detection only applies to analog and discrete value detectors.

Also, edge detection is handled slightly differently based on whether
you are using analog tags or discrete tags.
For more information on the event system, see Chapter 11, "Event
Subsystem."
You can also use the ToDiscrete() query filter to determine when data
values cross a particular threshold. For more information, see
"Converting Analog Values to Discrete Values (ToDiscrete)" on
page 259.
Edge Detection for Analog Tags

For example, the behavior of the WHERE clause as it processes a
result set can be illustrated as:
D F
B
V
A
L
U
E
A C E G
TIME
Slopes A-B, C-D and E-F are rising edges, while slopes B-C, D-E and
F-G are falling edges. The slopes are affected by the WHERE clause,
which is a combination of the wwEdgeDetection option and the
comparison operator used against the value.

The following table describes the rows that would be returned for the
various edge detection settings:
Operator = < > <= >=

Leading Falling and Falling Rising edge Falling Rising edge
rising edge only; only; first edge only; only; first
edges; first first value value to first value value to
value that to meet the meet the to meet the meet the
meets the criteria.* criteria. criteria. * criteria.
criteria.
Trailing Falling and Rising edge Falling Rising edge Falling
rising only; equal edge only; only; first edge only;
edges; first to the first first value value to fail first value
value to fail value to fail to fail the the criteria. to fail the
the criteria the criteria. criteria.* criteria.*
after a
value
meets the
criteria.
* If the falling edge is a vertical edge with no slope, the query will
return the lowest value of that edge.
The following query selects all values of "SysTimeSec" that are greater
than or equal to 50 from the History table between 10:00 and 10:02
a.m. on December 2, 2001. No edge detection is specified.
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwEdgeDetection = 'None'
The results are:

DateTime Value
2001-12-02 10:00:50.000 50
2001-12-02 10:00:52.000 52
2001-12-02 10:00:54.000 54
2001-12-02 10:00:56.000 56
2001-12-02 10:00:58.000 58
2001-12-02 10:01:50.000 50
2001-12-02 10:01:52.000 52
2001-12-02 10:01:54.000 54

2001-12-02 10:01:56.000 56
2001-12-02 10:01:58.000 58
Leading Edge Detection for Analog Tags

If Leading is specified as the parameter in the edge detection time
domain extension, the only rows in the result set are those that are the
first to successfully meet the WHERE clause criteria (returned true)
after a row did not successfully meet the WHERE clause criteria
(returned false).
The following query selects the first values of "SysTimeSec" from the
History table to meet the Value criterion between 10:00 and 10:02 a.m.
on December 2, 2001.
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwEdgeDetection = 'Leading'
The query will return only the two values that were greater than or
equal to 50 for the time period specified:
DateTime Value
2001-12-02 10:00:50.000 50
2001-12-02 10:01:50.000 50
Compare these results with the same query using no edge detection, as
shown in "Edge Detection for Analog Tags" on page 267. Notice that
even though the original query returned ten rows, the edge detection
only returns the first row recorded after the event criteria returned
true.

Trailing Edge Detection for Analog Tags

If Trailing is specified as the parameter in the edge detection
extension, the only rows in the result set are those that are the first to
fail the criteria in the WHERE clause (returned false) after a row
successfully met the WHERE clause criteria (returned true).
The following query selects the first values of "SysTimeSec" from the
History table to fail the Value criterion between 10:00 and 10:02 a.m.
on December 2, 2001.
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwEdgeDetection = 'Trailing'
The query returns only the two values that were the first to fail the
criteria in the WHERE clause for the time period specified:
DateTime Value
2001-12-02 10:01:00.000 0
2001-12-02 10:02:00.000 0
even though the original query returned ten recorded rows for each
value, the edge detection only returns the first row recorded after the
event criteria returned false.
Both Leading and Trailing Edge Detection for

Analog Tags
If Both is specified as the parameter in the edge detection extension,
all rows satisfying both the leading and trailing conditions are
returned.

The following query selects values of "SysTimeSec" from the History

table that meet both the Leading and Trailing criteria between 10:00
and 10:02 a.m. on December 2, 2001.
FROM History
AND DateTime >= '2001-12-02 10:00:00'
AND DateTime <= '2001-12-02 10:02:00'
AND Value >= 50
AND wwEdgeDetection = 'Both'
The results are:

DateTime Value
2001-12-02 10:00:50.000 50
2001-12-02 10:01:00.000 0
2001-12-02 10:01:50.000 50
2001-12-02 10:02:00.000 0
value of the first row in the original query is returned in the result set.
Edge Detection for Discrete Tags

Edge detection for discrete tags operates differently than for analog
tags. For example, assume the following discrete tags are stored.
Tag Description
SysPulse Transitions between 1 and 0 every minute.

MyPulse Transitions between 1 and 0 every 40 seconds.
A representation of the data stored in the system is as follows:
1
MyPulse
1
SysPulse
0
00:03:20
00:03:40
00:04:00
00:01:00
00:01:20
00:01:40
00:02:00
00:02:20
00:02:40
00:03:00

The following queries select values of "SysPulse" and "MyPulse" that

have a value of 1 (On) from the History and WideHistory tables
between 12:59 and 1:04 a.m. on December 8, 2001. No edge detection is
specified.
Query 1
Query for History.
FROM History
WHERE TagName IN ('SysPulse','MyPulse')
AND DateTime > '2001-12-08 00:59:00'
AND DateTime <= '2001-12-08 01:04:00'
AND Value = 1
AND wwEdgeDetection = 'None'
The results are:

2001-12-08 00:01:00.000 SysPulse 1
2001-12-08 00:01:00.000 MyPulse 1
2001-12-08 00:02:20.000 MyPulse 1
2001-12-08 00:03:00.000 SysPulse 1
2001-12-08 00:03:40.000 MyPulse 1
Query 2
Query for WideHistory.
SELECT * FROM OpenQuery(INSQL, 'SELECT DateTime, SysPulse,
MyPulse FROM WideHistory
WHERE DateTime > "2001-12-08 00:59:00"
AND DateTime < "2001-12-08 01:05:00"
AND SysPulse = 1
AND MyPulse = 1
AND wwEdgeDetection = "None"
')
The results are:

DateTime SysPulse MyPulse
2001-12-08 00:01:00.000 1 1

Leading Edge Detection for Discrete Tags

If Leading is specified as the parameter in the edge detection time
domain extension, the only rows in the result set are those that are the
first to successfully meet the WHERE clause criteria (returned true)
after a row did not successfully meet the WHERE clause criteria
(returned false).
have a value of 1 (On) from the History and WideHistory tables
between 12:59 and 1:04 a.m. on December 8, 2001.
Query 1
For a query on the History table, if the WHERE clause criteria specify
to return only discrete values equal to 1 (On), then applying a leading
edge detection does not change the result set.
FROM History
AND DateTime > '2001-12-08 00:59:00'
AND DateTime <= '2001-12-08 01:04:00'
AND Value = 1
AND wwEdgeDetection = 'Leading'
The results are:

2001-12-08 00:01:00.000 SysPulse 1
2001-12-08 00:01:00.000 MyPulse 1
2001-12-08 00:02:20.000 MyPulse 1
2001-12-08 00:03:00.000 SysPulse 1
2001-12-08 00:03:40.000 MyPulse 1
Query 2
For a query on the WideHistory table, applying a leading edge
detection requires that the condition only evaluate to true if both
values are equal to 1 (On).
SELECT DateTime, SysPulse, MyPulse FROM OpenQuery(INSQL, 'SELECT
DateTime, SysPulse, MyPulse
FROM WideHistory
WHERE DateTime > "2001-12-08 00:59:00"
AND DateTime <= "2001-12-08 01:04:00"
AND SysPulse = 1
AND MyPulse = 1
AND wwEdgeDetection = "Leading"
')

The results are:

2001-12-08 00:01:00.000 1 1
2001-12-08 00:03:40.000 1 1
shown in "Edge Detection for Discrete Tags" on page 271. If you look at
the diagram, you might think that a row could be returned at 00:03:00,
but because both tags change at exactly this instant, their values are
not returned. In a normal process, it is unlikely that two tags would
change at exactly at the same instant.
Trailing Edge Detection for Discrete Tags

If Trailing is specified as the parameter in the edge detection
extension, the only rows in the result set are those that are the first to
fail the criteria in the WHERE clause (returned false) after a row
successfully met the WHERE clause criteria (returned true).
Query 1
For a query on the History table, if the WHERE clause criteria
specifies to return only discrete values equal to 1 (On), then applying a
trailing edge detection is the same as reversing the WHERE clause (as
if Value = 0 was applied).
FROM History
AND DateTime > '2001-12-08 00:59:00'
AND DateTime <= '2001-12-08 01:04:00'
AND Value = 1
AND wwEdgeDetection = 'Trailing'
The results are:

2001-12-08 00:01:40.000 MyPulse 1
2001-12-08 00:02:00.000 SysPulse 1
2001-12-08 00:03:00.000 MyPulse 1
2001-12-08 00:04:00.000 SysPulse 1

Query 2
For a query on the WideHistory table, applying a trailing edge
detection returns the boundaries where the condition ceases to be true
(one of the values is equal to 0).
FROM WideHistory
WHERE DateTime > "2001-12-08 00:59:00"
AND DateTime <= "2001-12-08 01:04:00"
AND SysPulse = 1
AND MyPulse = 1
AND wwEdgeDetection = "Trailing"
')
The results are:

2001-12-08 00:01:40.000 1 1
2001-12-08 00:04:00.000 1 1
shown in "Edge Detection for Discrete Tags" on page 271. If you look at
the diagram, you might think that a row could be returned at 00:03:00,
but because both tags change at exactly this instant, their values are
not returned. In a normal process, it is unlikely that two tags would
change at exactly at the same instant.
Both Leading and Trailing Edge Detection for

Discrete Tags
If Both is specified as the parameter in the edge detection extension,
all rows satisfying both the leading and trailing conditions are
returned.
meet an edge detection of Both for a value criterion of 1 (On) from the
History and WideHistory tables between 12:59 and 1:04 a.m. on
December 8, 2001.

Query 1
FROM History
AND DateTime > '2001-12-08 00:59:00'
AND DateTime <= '2001-12-08 01:04:00'
AND Value = 1
AND wwEdgeDetection = 'Both'
The results are:

2001-12-08 00:01:00.000 SysPulse 1
2001-12-08 00:01:00.000 MyPulse 1
2001-12-08 00:01:40.000 MyPulse 1
2001-12-08 00:02:00.000 SysPulse 1
2001-12-08 00:02:20.000 MyPulse 1
2001-12-08 00:03:00.000 SysPulse 1
2001-12-08 00:03:00.000 MyPulse 1
2001-12-08 00:03:40.000 MyPulse 1
2001-12-08 00:04:00.000 SysPulse 1
Query 2
FROM WideHistory
WHERE DateTime > "2001-12-08 00:59:00"
AND DateTime <= "2001-12-08 01:04:00"
AND SysPulse = 1
AND MyPulse = 1
AND wwEdgeDetection = "Both"
')
The results are:

2001-12-08 00:01:00.000 1 1
2001-12-08 00:01:40.000 1 1
2001-12-08 00:03:40.000 1 1
2001-12-08 00:04:00.000 1 1
shown in the "Edge Detection for Discrete Tags" on page 271.

277
Chapter 9
Query Examples
In addition to query examples that use the Wonderware Historian

time domain extensions, other query examples are provided to
demonstrate how to perform more complex queries or to further
explain how retrieval works.
The examples provided are not exhaustive of all possible database
queries, but they should give you an idea of the kinds of queries that
you could write.
Note: If you have configured SQL Server to be case-sensitive, be sure

that you use the correct case when writing queries.
Querying the History Table

The History table presents acquired plant data in a historical format.
For more information, see History Tables in Chapter 1, Table Categories, in
the Historian Database Reference.
The following query returns the date/time stamp and value for the tag
"ReactLevel." The query uses the remote table view (History is used in
place of INSQL.Runtime.dbo.History).

278  Chapter 9 Query Examples
If you do not specify a wwCycleCount or wwResolution, the query will

return 100 rows (the default).
SELECT DateTime, Sec = DATEPART(ss, DateTime), TagName, Value
FROM History
AND DateTime >= '2001-03-13 1:15:00pm'
AND DateTime <= '2001-03-13 2:15:00pm'
The results are:

DateTime Sec TagName Value
2001-03-13 13:15:00.000 0 ReactLevel 1775.0
2001-03-13 13:15:00.000 36 ReactLevel 1260.0
2001-03-13 13:16:00.000 12 ReactLevel 1650.0
2001-03-13 13:16:00.000 49 ReactLevel 1280.0
2001-03-13 13:17:00.000 25 ReactLevel 1525.0
2001-03-13 13:18:00.000 1 ReactLevel 585.0
2001-03-13 13:18:00.000 38 ReactLevel 1400.0
2001-03-13 13:19:00.000 14 ReactLevel 650.0
2001-03-13 13:19:00.000 50 ReactLevel 2025.0
2001-03-13 13:20:00.000 27 ReactLevel 765.0
2001-03-13 13:21:00.000 3 ReactLevel 2000.0
2001-03-13 13:21:00.000 39 ReactLevel 830.0
2001-03-13 13:22:00.000 16 ReactLevel 1925.0
.
.
.
Querying the Live Table

The Live table presents the latest available data for each tag in the
table. For more information, see History Tables in Chapter 1, Table
Categories,"Historian Database Reference" on page 1.
The following query returns the current value of the specified tag. The
query uses the remote table view (Live is used in place of
INSQL.Runtime.dbo.Live).
SELECT TagName, Value
FROM Live

Querying the WideHistory Table279
The result is:

TagName Value
ReactLevel 1145.0
(1 row(s) affected)
Querying the WideHistory Table

The wide extension table is a transposition of the History table. Use
the wide history tables any time you want to find the value of one or
more tags over time and need to specify different filter criteria for each
tag.
For more information, see History Tables in Chapter 1, Table
Categories, in the Historian Database Reference.
The following query returns the value of two tags from the
WideHistory table. The WideHistory table can only be accessed using
the OPENQUERY function. The "Runtime.dbo." qualifier is optional.
SELECT * FROM OpenQuery(INSQL,'
SELECT DateTime, ReactLevel, ReactTemp
FROM Runtime.dbo.WideHistory
WHERE Reactlevel > 1500
AND ReactTemp > 150
')
The results are:

DateTime ReactLevel ReactTemp
2001-03-02 06:20:00.000 1865.0 191.3
2001-03-02 06:21:00.000 2025.0 195.9
2001-03-02 06:22:00.000 2000.0 195.9
2001-03-02 06:23:00.000 2025.0 180.9
2001-03-02 06:27:00.000 1505.0 177.5
(5 row(s) affected).
In the WideHistory table, the column type is determined by the tag
type.
SELECT * FROM OpenQuery(INSQL, 'SELECT DateTime, SysTimeMin,
SysPulse, SysString FROM WideHistory
WHERE DateTime >= "2001-12-20 0:00"
AND DateTime <= "2001-12-20 0:05"
AND wwRetrievalMode = "delta"
')

The results are:

DateTime SysTimeMin SysPulse SysString
2001-12-20 00:00:00.000 0 0 2001/12/20 08:00:00
2001-12-20 00:01:00.000 1 1 2001/12/20 08:00:00
2001-12-20 00:02:00.000 2 0 2001/12/20 08:00:00
2001-12-20 00:03:00.000 3 1 2001/12/20 08:00:00
2001-12-20 00:04:00.000 4 0 2001/12/20 08:00:00
2001-12-20 00:05:00.000 5 1 2001/12/20 08:00:00
Querying Wide Tables in Delta Retrieval Mode

Wide tables in delta retrieval mode will behave normally if only one
tag is returned. However, for a multiple tag display, a complete row is
returned to the client for each instance in which one or more of the
tags in the query returns a different value. The row will reflect the
actual values being returned for the tags returning results, and will
reflect the previous values for the remaining tags in the result set
(similar to cyclic retrieval).
Note: The value can be "invalid" or some other quality value.
The following query returns values for three tags from the
WideHistory table. "MyTagName" is a tag that periodically is invalid.
SELECT DateTime, SysTimeSec, SysTimeMin, MyTagName
FROM WideHistory
WHERE DateTime >= "2001-05-12 13:00:00"
AND DateTime <= "2001-05-12 13:02:00"
')
The results are:

DateTime SysTimeSec SysTimeMin MyTagName
: : : :
: : : :
2001-05-12 13:00:55.000 55 00 1
2001-05-12 13:00:56.000 56 00 1
2001-05-12 13:00:57.000 57 00 1
2001-05-12 13:00:57.500 57 00 null
2001-05-12 13:00:58.000 58 00 null
2001-05-12 13:00:59.000 59 00 null
2001-05-12 13:01:00.000 00 01 null
2001-05-12 13:01:00.500 00 01 2
2001-05-12 13:01:01.000 01 01 2

Querying the AnalogSummaryHistory View281
2001-05-12 13:01:02.000 02 01 2
2001-05-12 13:01:03.000 03 01 2
: : : :
: : : :
: : : :
Notice that 57 appears twice since the occurrence of 1 changing to

NULL for tag "MyTagName" occurs sometime between the 57th and
58th second. The same applies for NULL changing to 2. The same
behavior applies to discrete values.
Querying the AnalogSummaryHistory View

The AnalogSummaryHistory view is a “wide” view that allows you to
return multiple statistics for a single tag within a single query.
The following query returns the minimum, maximum, and average
values for the SysTimeSec tag for the last minute.
declare @End datetime
set @End = left(convert(varchar(30),getdate(),120),14)+'00:00'
SELECT Tagname, OPCQuality, Minimum as MIN, Maximum as MAX,

Average as AVG
AND StartDateTime >= dateadd(minute,-60,@End)
AND EndDateTime < @End
The results are:

Tagname OPCQuality MIN MAX AVG
SysTimeSec 192 0 59 29.5

Querying the StateSummaryHistory View

The StateSummaryHistory view is a “wide” view that allows you to
return multiple statistics for a single tag within a single query.
The following query returns the state count, total time in state, and
the percentage of time in state for the SysPulse system tag for the last
minute. One row is returned for each state.
declare @End datetime
set @End = left(convert(varchar(30),getdate(),120),14)+'00:00'
SELECT TagName, Value, OPCQuality, StateCount, StateTimeTotal,

StateTimePercent
FROM StateSummaryHistory
WHERE TagName = 'SysPulse'
AND StartDateTime >= dateadd(minute,-60,@End)
AND EndDateTime < @End
The results are:

TagName Value OPCQuality StateCount StateTimeTotal StateTimePercent
SysPulse 0 192 15 900000 50
SysPulse 1 192 15 900000 50
The following query returns the minimum time in state, the minimum
contained time in state, and value for the SysTimeSec system tag.
SELECT TagName, StartDateTime, EndDateTime, StateTimeMin as STM,
StateTimeMinContained as STMC, Value
FROM StateSummaryHistory
WHERE TagName='SysTimeSec'
AND wwRetrievalMode='Cyclic'
AND wwResolution=5000
AND StartDateTime>='2009-10-21 17:40:00.123'
AND StartDateTime<='2009-10-21 17:40:05.000'
The results are:

TagName StartDateTime EndDateTime STM STMC Value
SysTimeSec 2009-10-21 2009-10-21 877 0 0
17:40:00.123 17:40:05.123
SysTimeSec 2009-10-21 2009-10-21 1000 1000 1
17:40:00.123 17:40:05.123
SysTimeSec 2009-10-21 2009-10-21 1000 1000 2
17:40:00.123 17:40:05.123
SysTimeSec 2009-10-21 2009-10-21 1000 1000 3
17:40:00.123 17:40:05.123

Using an Unconventional Tagname in a Wide Table Query283
TagName StartDateTime EndDateTime STM STMC Value

SysTimeSec 2009-10-21 2009-10-21 1000 1000 4
17:40:00.123 17:40:05.123
SysTimeSec 2009-10-21 2009-10-21 123 0 5
17:40:00.123 17:40:05.123
Using an Unconventional Tagname in a Wide

Table Query
In a SQL query against a wide table, unconventional tag names must
be delimited with brackets ( [ ] ), because the tagname is used as a
column name. For example, tagnames containing a minus ( - ) or a
forward slash ( / ) must be delimited, otherwise the parser will attempt
to perform the corresponding arithmetic operation. No error will result
from using brackets where not strictly necessary. For more
information on unconventional tagnames, see "Naming Conventions
for Tagnames" on page 22.
The following is an example of how to delimit a tagname in a query on
a wide table. "ReactTemp-2" and "ReactTemp+2" are tagnames.
Without the delimiters, the parser would attempt to include the "-2"
and "+2" suffixes on the tagnames as part of the arithmetic operation.
For clarity and maintainability of your queries, however, it is
recommended that you do not use special characters in tagnames
unless strictly necessary.
'SELECT ReactTemp, [ReactTemp-2]-2, [ReactTemp+2]+2 FROM
WideHistory WHERE ... ')
Using an INNER REMOTE JOIN

Instead of using " … WHERE TagName IN (SELECT TagName … ) ", it
is more efficient to use INNER REMOTE JOIN syntax.
In general, use the following pattern for INNER REMOTE JOIN
queries against the historian:
<SQLServerTable> INNER REMOTE JOIN
<HistorianExtensionTable>

Query 1
This query returns data from the history table, based on a string tag
that you filter for from the StringTag table:
SELECT DateTime, T.TagName, vValue, Quality, QualityDetail
FROM StringTag T inner remote join History H
ON T.TagName = H.TagName
WHERE T.MaxLength = 64
AND DateTime >='2002-03-10 12:00:00.000'
AND DateTime <='2002-03-10 16:40:00.000'
Query 2
This query returns data from the history table, based on a discrete tag
that you filter for from the Tag table:
SELECT DateTime, T.TagName, vValue, Quality, QualityDetail
FROM Tag T inner remote join History H
ON T.TagName = H.TagName
WHERE T.TagType = 2
AND T.Description like 'Discrete%'
AND DateTime >='2002-03-10 12:00:00.000'
AND DateTime <='2002-03-10 16:40:00.000'
Setting Both a Time and Value Deadband for

Retrieval
If both time and value deadbands are specified, then every sample is
checked for both deadbands, against the current basis value (the last
sample returned).
If it passes both tests, then it is returned and acts as the basis for
checking the next sample.
For example:
FROM History
AND DateTime >= '2002-03-13 10:08'
AND DateTime <= '2002-03-13 10:28'

Setting Both a Time and Value Deadband for Retrieval285
The tag selected, 'ReactTemp,' has a MinEU value of 0 and a MaxEU

value of 220. Thus, the value deadband will be 5 percent of (220 - 0),
which equals 11. ReactTemp changes rapidly between its extreme
values, but the value remains constant for short periods near the high
and low temperature limits. Therefore, when changes are rapid, the
value deadband condition is satisfied first, then the time deadband is
satisfied. In this region, the behavior is dominated by the time
deadband, and the returned rows are spaced at 5 second intervals.
Where the temperature is more constant (particularly at the low
temperature end), the time deadband is satisfied first, followed by the
value deadband. Both deadbands are only satisfied when the value of a
row is more than 11 degrees different from the previous row. Thus, the
effect of value deadband can be seen to dominate near the low and
high temperature extremes of the tag.
The results are:
2002-03-13 10:08:00.000 ReactTemp 121.0
2002-03-13 10:08:10.000 ReactTemp 189.10000610351562
2002-03-13 10:08:20.000 ReactTemp 147.69999694824219
2002-03-13 10:08:30.000 ReactTemp 106.30000305175781
2002-03-13 10:08:40.000 ReactTemp 30.100000381469727
2002-03-13 10:08:50.000 ReactTemp 16.399999618530273
2002-03-13 10:09:00.000 ReactTemp 61.0
2002-03-13 10:09:10.000 ReactTemp 151.0
2002-03-13 10:09:20.000 ReactTemp 173.0
2002-03-13 10:09:30.000 ReactTemp 131.60000610351562
2002-03-13 10:09:40.000 ReactTemp 57.700000762939453
2002-03-13 10:09:50.000 ReactTemp 16.299999237060547
2002-03-13 10:10:10.000 ReactTemp 96.0
2002-03-13 10:10:20.000 ReactTemp 186.0
2002-03-13 10:10:30.000 ReactTemp 156.89999389648437
2002-03-13 10:10:40.000 ReactTemp 115.5
2002-03-13 10:10:50.000 ReactTemp 41.599998474121094
2002-03-13 10:11:00.000 ReactTemp 21.0
2002-03-13 10:11:10.000 ReactTemp 41.0
2002-03-13 10:11:20.000 ReactTemp 131.0
2002-03-13 10:11:30.000 ReactTemp 184.5
2002-03-13 10:11:40.000 ReactTemp 140.80000305175781
2002-03-13 10:11:50.000 ReactTemp 99.400001525878906
2002-03-13 10:12:00.000 ReactTemp 25.5
2002-03-13 10:12:20.000 ReactTemp 76.0
2002-03-13 10:12:30.000 ReactTemp 166.0


2002-03-13 10:12:50.000 ReactTemp 124.69999694824219
2002-03-13 10:13:00.000 ReactTemp 50.799999237060547
2002-03-13 10:13:10.000 ReactTemp 16.399999618530273
2002-03-13 10:13:30.000 ReactTemp 111.0
2002-03-13 10:13:40.000 ReactTemp 193.69999694824219
2002-03-13 10:13:50.000 ReactTemp 152.30000305175781
2002-03-13 10:14:00.000 ReactTemp 108.59999847412109
2002-03-13 10:14:10.000 ReactTemp 34.700000762939453
2002-03-13 10:14:20.000 ReactTemp 21.0
2002-03-13 10:14:30.000 ReactTemp 51.0
2002-03-13 10:14:40.000 ReactTemp 146.0
2002-03-13 10:14:50.000 ReactTemp 177.60000610351562
2002-03-13 10:15:00.000 ReactTemp 136.19999694824219
2002-03-13 10:15:10.000 ReactTemp 92.5
2002-03-13 10:15:20.000 ReactTemp 18.600000381469727
2002-03-13 10:15:40.000 ReactTemp 86.0
2002-03-13 10:15:50.000 ReactTemp 181.0
2002-03-13 10:16:00.000 ReactTemp 161.5
2002-03-13 10:16:10.000 ReactTemp 120.09999847412109
2002-03-13 10:16:20.000 ReactTemp 76.400001525878906
2002-03-13 10:16:30.000 ReactTemp 20.899999618530273
2002-03-13 10:16:50.000 ReactTemp 81.0
2002-03-13 10:17:00.000 ReactTemp 176.0
2002-03-13 10:17:10.000 ReactTemp 163.80000305175781
2002-03-13 10:17:20.000 ReactTemp 122.40000152587891
2002-03-13 10:17:30.000 ReactTemp 46.200000762939453
2002-03-13 10:17:40.000 ReactTemp 18.700000762939453
2002-03-13 10:18:00.000 ReactTemp 116.0
2002-03-13 10:18:10.000 ReactTemp 189.10000610351562
2002-03-13 10:18:20.000 ReactTemp 147.69999694824219
.
.
.

Using wwResolution, wwCycleCount, and wwRetrievalMode in the Same Query287
Using wwResolution, wwCycleCount, and

wwRetrievalMode in the Same Query
The results of a database query will vary depending on the
combination of resolution, cycle count, and retrieval mode that you use
in the query. These results are summarized in the following table. N =
A numeric value.
Retrieval Cycle
Mode Resolution Count Results
CYCLIC N 0 (or no All stored data for tags during the specified time
value) interval are queried, and then a resolution of N
ms applied.
CYCLIC 0 (or no 0 The server will return 100,000 rows per tag
value) specified.
CYCLIC 0 (or no N All stored data for tags during the specified time
value) interval are queried, and then a cycle count of N
evenly spaced rows is applied.
CYCLIC N (any value All stored data for tags during the specified time
is ignored) interval are queried, and then a resolution of N
ms applied.
CYCLIC (no value) (no value The server will return 100 rows per tag specified.
or a value
less than 0)
DELTA (any value is 0 All values that changed during the specified time
ignored) interval are returned (up to 100,000 rows total).
DELTA (any value is N Values that changed during the specified time
ignored) interval are queried, and then a cycle count (first
N rows) is applied. The cycle count limits the
maximum number of rows returned, regardless of
how many tags were queried. For example, a
query that applies a cycle count of 20 to four tags
will return a maximum of 20 rows of data. An
initial row will be returned for each tag, and the
remaining 16 rows will be based on subsequent
value changes for any tag.
DELTA (any value is (no value) All values that changed during the specified time
ignored) interval are returned (no row limit).
In general, if there is an error in the virtual columns, or an

unresolvable conflict, then zero rows are returned.

Determining Cycle Boundaries

Cycle boundaries are calculated based on the query start and end
times, wwCycleCount, and wwResolution.
If you only specify wwCycleCount, evenly spaced cycles are returned
based on the value of wwCycleCount.
If you only specify wwResolution, cycles are spaced wwResolution
milliseconds apart starting at the query start time until query end
time is reached. The last cycle will have whatever duration is required
to end exactly at the query end time. If this last duration is shortened
by this rule, it is known as a partial cycle. Because of this, the final
cycle duration may not match wwResolution.
If both wwCycleCount and wwResolution are specified, no result rows
will be returned. If you specify neither wwCycleCount nor
wwResolution in the query, the query will return 100 rows.
Unless otherwise specified, a value is considered in a given full or
partial cycle if its timestamp occurs at or after the cycle start
(timestamp >= cycle start) and before the cycle end (timestamp < cycle
end).
Mixing Tag Types in the Same Query

The History and Live tables use the sql_variant data type for the
vValue column, allowing the return of various data types in a single
column. In other words, these tables allow values for tags of different
types to be retrieved with a simple query, without the need for a JOIN
operation.
For example:
SELECT TagName, DateTime, vValue
FROM History
WHERE TagName IN ('SysTimeMin', 'SysPulse', 'SysString')
AND DateTime >= '2001-12-20 0:00'
AND DateTime <= '2001-12-20 0:05'
The results are:

TagName DateTime vValue
SysTimeMin 2001-12-20 00:00:00.000 0
SysPulse 2001-12-20 00:00:00.000 0
SysString 2001-12-20 00:00:00.000 2001/12/20 08:00:00
SysTimeMin 2001-12-20 00:01:00.000 1
SysPulse 2001-12-20 00:01:00.000 1
SysTimeMin 2001-12-20 00:02:00.000 2
SysPulse 2001-12-20 00:02:00.000 0

Using a Criteria Condition on a Column of Variant Data289
SysTimeMin 2001-12-20 00:03:00.000 3

SysPulse 2001-12-20 00:03:00.000 1
SysTimeMin 2001-12-20 00:04:00.000 4
SysPulse 2001-12-20 00:04:00.000 0
SysTimeMin 2001-12-20 00:05:00.000 5
SysPulse 2001-12-20 00:05:00.000 1
Using a Criteria Condition on a Column of

Variant Data
The Wonderware Historian OLE DB provider sends variant data to
the SQL Server as a string. If the query contains a criteria condition
on a column containing variant type data, the filtering is handled by
SQL Server. An example of a criteria condition is:
WHERE ... vValue = 2
To perform the filtering, the SQL Server must determine the data type
of the constant (in this example, 2), and attempt to convert the variant
(string) to this destination type. The SQL Server assumes that a
constant without a decimal is an integer, and attempts to convert the
string to an integer type. This conversion will fail in SQL Server if the
string actually represents a float (for example, 2.00123).
You should explicitly state the destination type by means of a
CONVERT function. This is the only reliable way of filtering on the
vValue column, which contains variant data.
For example:
SELECT DateTime, Quality, OPCQuality, QualityDetail, Value,
vValue, TagName
FROM History
WHERE TagName IN ('ADxxxF36', 'SysTimeMin', 'SysPulse')
AND DateTime >= '12-04-2001 04:00:00.000'
AND DateTime <= '12-04-2001 04:03:00.000'
AND convert(float, vValue) = 2

The following is another example:

SELECT DateTime, Quality, OPCQuality, QualityDetail, Value,
vValue, TagName
FROM History
WHERE TagName IN ('VectorX', 'SysTimeMin', 'SysPulse')
AND DateTime >= '20020313 04:00:07.000'
AND DateTime <= '20020313 04:01:00.000'
AND convert(float, vValue) > 1
AND convert(float, vValue) < 2
Using DateTime Functions

Date functions perform an operation on a date and time input value
and return either a string, numeric, or date and time value. The
following query returns the date/time stamp and value for the tag
'SysTimeSec' for the last 10 minutes.
SELECT DateTime, TagName, Value, Quality
FROM History
AND DateTime >= dateadd(Minute, -10, GetDate())
AND DateTime <= GetDate()
The results are:

DateTime TagName Value Quality
2001-12-15 13:00:00.000 SysTimeSec 0.0 0
2001-12-15 13:00:06.060 SysTimeSec 6.0 0
2001-12-15 13:00:12.120 SysTimeSec 12.0 0
2001-12-15 13:00:18.180 SysTimeSec 18.0 0
2001-12-15 13:00:24.240 SysTimeSec 24.0 0
2001-12-15 13:00:30.300 SysTimeSec 30.0 0
2001-12-15 13:00:36.360 SysTimeSec 36.0 0
2001-12-15 13:00:42.420 SysTimeSec 42.0 0
.
.
.
If you want to use date/time functions and the wwTimeZone

parameter in the same query, you will need to use the faaTZgetdate()
function. This is because of differences in how the SQL Server and the
Wonderware Historian OLE DB provider determine the end date for a
query.

Using DateTime Functions291
For any query, the SQL Server performs all date/time computations in
local server time, reformulates the query with specific dates, and sends
it on to the Wonderware Historian OLE DB provider. The Wonderware
Historian OLE DB provider then applies the wwTimeZone parameter
in determining the result set.
For example, the following query requests the last 30 minutes of data,
expressed in Eastern Daylight Time (EDT). The server is located in
the Pacific Daylight Time (PDT) zone.
SELECT DateTime, TagName, Value FROM History
WHERE TagName IN ('SysTimeHour', 'SysTimeMin',
'SysTimeSec')
AND DateTime > DateAdd(mi, -30, GetDate())
AND wwTimeZone = 'eastern daylight time'
If it is currently 14:00:00 in the Pacific Daylight Time zone, then it is

17:00:00 in the Eastern Daylight Time zone. You would expect the
query to return data from 16:30:00 to 17:00:00 EDT, representing the
last 30 minutes in the Eastern Daylight Time zone.
However, the data that is returned is from 13:30:00 to 17:00:00 EDT.
This is because the SQL Server computes the "DateAdd(mi, -30,
GetDate())" part of the query assuming the local server time zone (in
this example, PDT). It then passes the Wonderware Historian OLE DB
provider a query similar to the following:
WHERE TagName IN ('SysTimeHour', 'SysTimeMin', 'SysTimeSec')
AND DateTime > 'YYYY-MM-DD 13:30:00.000'
Because the OLE DB provider is not provided an end date, it assumes

the end date to be the current time in the specified time zone, which is
17:00:00 EDT.
To work around this problem, use the faaTZgetdate() function with
intermediate variables. For example:
DECLARE @starttime datetime
SET @starttime = dbo.faaTZgetdate('eastern daylight time')

WHERE TagName IN ('SysTimeHour', 'SysTimeMin', 'SysTimeSec')
AND DateTime > DateAdd(mi, -30, @starttime)
AND DateTime < DateAdd(mi, -5, @starttime)

The following example uses a wide table:

SELECT * FROM OpenQuery(INSQL, '
SELECT DateTime, SysTimeHour, SysTimeMin, SysTimeSec FROM
WideHistory
WHERE DateTime > DateAdd(mi, -30, faaTZgetdate("eastern
daylight time"))
AND DateTime < DateAdd(mi, -5, faaTZgetdate("eastern
daylight time"))
AND wwTimeZone = "eastern daylight time"
')
Using the GROUP BY Clause

The GROUP BY clause works if the query uses the four-part naming
convention or one of the associated views.
The following example will find the highest value of a specified set of
tags over a time period.
SELECT TagName, Max(Value)
FROM INSQL.Runtime.dbo.History
WHERE TagName IN
('ReactTemp','ReactLevel','SysTimeSec')
AND DateTime > '2001-12-20 0:00'
AND DateTime < '2001-12-20 0:05'
GROUP BY TagName
The results are:

SysTimeSec 59.0
Using the COUNT() Function

The COUNT(*) function works directly in a four-part query, but is not
supported inside of the OPENQUERY function.
For example:
SELECT count(*)
FROM History
AND DateTime >= '2001-12-20 0:00'
AND DateTime <= '2001-12-20 0:05'
AND Value >= 30
The result is:

150

Using an Arithmetic Function293
If you use the OPENQUERY function, you cannot perform arithmetic

functions on the COUNT(*) column. However, you can perform the
count outside of the OPENQUERY, as follows:
SELECT count(*), count(*)/2 FROM OPENQUERY(INSQL, 'SELECT
DateTime, vValue, Quality, QualityDetail
FROM History
WHERE TagName IN ("SysTimeSec")
AND DateTime >= "2002-04-16 03:00:00.000"
AND DateTime <= "2002-04-16 06:00:00.000"
')
The result is:

10801 5400
(1 row(s) affected)
Using an Arithmetic Function

The following query adds the values of two tags from the WideHistory
table.
SELECT DateTime, ReactLevel, ProdLevel, "Sum" =
ReactLevel+Prodlevel
FROM WideHistory
WHERE DateTime > "2001-02-28 18:56"
AND DateTime < "2001-02-28 19:00"
')
The results are:

DateTime ReactLevel Prodlevel Sum
2001-02-28 18:56:00.000 1525.0 2343.0 3868.0
2001-02-28 18:56:00.000 1525.0 2343.0 3868.0
2001-02-28 18:56:00.000 1525.0 2343.0 3868.0
2001-02-28 18:56:00.000 1525.0 2343.0 3868.0
2001-02-28 18:56:00.000 2025.0 2343.0 4368.0
2001-02-28 18:56:00.000 2025.0 2343.0 4368.0
.
.
.
If you use a math operator, such as plus (+), minus (-), multiply (*), or
divide (/), you will need to add a blank space in front of and after the
operator. For example, "Value - 2" instead of "Value-2".

Using an Aggregate Function

The following query returns the minimum, maximum, average, and
sum of the tag 'ReactLevel' from the WideHistory table.
SELECT "Minimum" = min(ReactLevel),
"Maximum" = max(ReactLevel),
"Average" = avg(ReactLevel),
"Sum" = sum(ReactLevel)
FROM WideHistory
WHERE DateTime > "2001-02-28 18:55:00 "
AND DateTime < "2001-02-28 19:00:00"
')
The results are:

Minimum Maximum Average Sum
-25.0 2025.0 1181.2 118120.0
(1 row(s) affected)
If you perform a SUM or AVG in delta retrieval mode against the Wide
table, the aggregation will only be performed when the value has
changed. The aggregation will not apply to all of the rows returned for
each column.
For example, the following query has no aggregation applied:
SELECT * FROM OpenQuery(INSQL, 'SELECT DateTime, SysTimeHour,
SysTimeMin, SysTimeSec, SysDateDay
FROM AnalogWideHistory
WHERE DateTime >= "2001-08-15 13:20:57.345"
AND DateTime < "2001-08-15 13:21:03.345"
')
GO
The results are:

DateTime SysTimeHour SysTimeMin SysTimeSec SysDateDay
2001-08-15 13:20:57.343 13 20 57 15
2001-08-15 13:20:58.000 13 20 58 15
2001-08-15 13:20:59.000 13 20 59 15
2001-08-15 13:21:00.000 13 21 0 15
2001-08-15 13:21:01.000 13 21 1 15
2001-08-15 13:21:02.000 13 21 2 15
2001-08-15 13:21:03.000 13 21 3 15
(7 row(s) affected)

Using an Aggregate Function295
Then, a SUM is applied to all of the returned column values:

SELECT * FROM OpenQuery(INSQL,'SELECT Sum(SysTimeHour),
Sum(SysTimeMin), Sum(SysTimeSec), Sum(SysDateDay)
FROM WideHistory
WHERE DateTime >= "2001-08-15 13:20:57.345"
AND DateTime < "2001-08-15 13:21:03.345"
')
GO
The results are:

SysTimeHour SysTimeMin SysTimeSec SysDateDay
13 41 180 15
Thus, for delta retrieval mode, a SUM or AVG is applied only if the
value has changed from the previous row.
If you perform an AVG in delta retrieval mode, AVG will be computed
as:
SUM of delta values/number of delta values
For example, an AVG is applied to all of the returned column values:
SELECT * FROM OpenQuery(INSQL,'SELECT Avg(SysTimeHour),
Avg(SysTimeMin), Avg(SysTimeSec), Avg(SysDateDay)
FROM WideHistory
WHERE DateTime >= "2001-08-15 13:20:57.345"
AND DateTime < "2001-08-15 13:21:03.345"
')
GO
The results are:

SysTimeMin SysTimeSec
20.5 25.714285714285715

Making and Querying Annotations

The following query creates an annotation for the specified tag. The
annotation is made in response to a pump turning off. Then, the
annotations for a particular tag are returned.
DECLARE @@UserKey INT
SELECT @@UserKey = UserKey

FROM UserDetail
WHERE UserName = 'wwAdmin'
INSERT INTO Annotation (TagName, UserKey, DateTime, Content)

VALUES ('ReactLevel', @@UserKey, GetDate(), 'The Pump is
off')
SELECT DateTime, TagName, Content

FROM Annotation
WHERE Annotation.TagName = 'ReactLevel'
AND DateTime > '27 Feb 01'
AND DateTime <= GetDate()
The results are:

DateTime TagName Content
2001-02-28 19:18:00.000 ReactLevel The Pump is off
(1 row(s) affected)
Using Comparison Operators with Delta

Retrieval
The system behaves differently when doing typical delta-based queries
where a start date and end date are specified using the comparison
operators >=, >, <= and <. The comparison operators can be used on
the History and WideHistory tables. The comparison operators also
apply regardless of how the query is executed (for example, four-part
naming, OLE DB provider views, and so on).
Delta queries that use the comparison operators return all the valid
changes to a set of tags over the specified time span. Using deadbands
and other filters may modify the set of valid changes.
Specifying the Start Date with ">="

If the start date is specified using >= (greater than or equal to), then a
row is always returned for the specified start date. If the start
date/time coincides exactly with a valid value change, then the Quality
is normal (0). Otherwise, the value at the start date is returned, and
the Quality value is 133 (because the length of time that the tag's
value was at X is unknown).

Using Comparison Operators with Delta Retrieval297
Query 1
For this query, the start date will not correspond to a data change:
SELECT DateTime, Value, Quality
FROM History
WHERE TagName = 'SysTimeMin'
AND DateTime >= '2001-01-13 12:00:30'
AND DateTime < '2001-01-13 12:10:00'
The start time (12:00:30) does not correspond with an actual change in
value, and is therefore marked with the initial quality of 133:
DateTime Value Quality
2001-01-13 12:00:30.000 0 133
2001-01-13 12:01:00.000 1 0
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0
2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
Query 2
For this query, the start date will correspond to a data change:
FROM History
AND DateTime >= '2001-01-13 12:01:00'
AND DateTime < '2001-01-13 12:10:00'
The start time (12:01:00) does correspond exactly with an actual

change in value, and is therefore marked with the normal quality of 0.
2001-01-13 12:01:00.000 1 0
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0

2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
(9 row(s) affected)
Query 3
For this query, the start date will return at least one row, even though
the query captures no data changes:
FROM History
AND DateTime >= '2001-01-13 12:00:30'
AND DateTime < '2001-01-13 12:01:00'
The query does not capture an actual change in value, and is therefore
marked with the initial value quality of 133 for the start time of the
query:
2001-01-13 12:00:30.000 0 133
(1 row(s) affected)
Specifying the Start Date with ">"

If the start date is specified using > (greater than), then the first row
returned is the first valid change after (but not including) the start
date. No initial value row is returned. A query that uses > to specify its
start date may return zero rows.
Query 1
For this query, the first row that will be returned will be the first valid
change after (but not including) the start time (12:00:30):
FROM History
AND DateTime > '2001-01-13 12:00:30'
AND DateTime < '2001-01-13 12:10:00'
The first row returned is the first valid change after (but not including)
the start time (12:00:30):
2001-01-13 12:01:00.000 1 0
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0

Using Comparison Operators with Delta Retrieval299
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0
2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
(9 row(s) affected)
Query 2
For this query, the start date will correspond to a data change, but it
will be excluded from the result set because the operator used is
greater than, not greater than or equal to.
FROM History
AND DateTime > '2001-01-13 12:01:00'
AND DateTime < '2001-01-13 12:10:00'
The start time (12:01:00) corresponds exactly with an actual change in

value, but it is excluded from the result set because the operator used
is greater than, not greater than or equal to.
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0
2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
(8 row(s) affected)
Query 3
This query will return no rows, because no data changes are captured:
FROM History
AND DateTime > '2001-01-13 12:00:30'
AND DateTime < '2001-01-13 12:01:00'

The query does not capture an actual change in value; therefore, no

rows are returned.
(0 row(s) affected)
Specifying the End Date with "<="

If the end date is specified using <= (less than or equal to) then the last
row returned is the last valid change up to, and including, the end
date. If the end date uses "<=" then the last change returned may have
a date/time exactly at the end date. If there is a value exactly at the
end date, it will be returned.
This query uses the remote table view.
FROM History
AND DateTime > '2001-01-13 12:00:30'
AND DateTime <= '2001-01-13 12:10:00'
Note that there is a valid change at exactly the end time of the query
(12:10:00):
2001-01-13 12:01:00.000 1 0
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0
2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
2001-01-13 12:10:00.000 10 0
Specifying the End Date with "<"

If the end date is specified using < (less than), then the last row
returned is the last valid change up to (but not including) the end date.
If the end date uses "<" then the last event returned will have a
date/time less than the end date. If there is an event exactly at the end
date, it will not be returned.

Using Comparison Operators with Cyclic Retrieval and Cycle Count301

FROM History
AND DateTime > '2001-01-13 12:00:30'
AND DateTime < '2001-01-13 12:10:00'
Note that there is a valid change at exactly the end time of the query
(12:10:00), but it is excluded from the result set.
2001-01-13 12:01:00.000 1 0
2001-01-13 12:02:00.000 2 0
2001-01-13 12:03:00.000 3 0
2001-01-13 12:04:00.000 4 0
2001-01-13 12:05:00.000 5 0
2001-01-13 12:06:00.000 6 0
2001-01-13 12:07:00.000 7 0
2001-01-13 12:08:00.000 8 0
2001-01-13 12:09:00.000 9 0
(9 row(s) affected)
Using Comparison Operators with Cyclic

Retrieval and Cycle Count
Cyclic queries with the wwCycleCount time domain extension return a
set of evenly spaced rows over the specified time span. The result set
will always return the number of rows specified by the cycle count
extension for each tag in the query. The resolution for these rows is
calculated by dividing the time span by the cycle count.
Using Two Equality Operators

If the time range is specified using >= and <=, then the first row falls
exactly on the start time, and the last row falls exactly on the end
time. In this case, the resolution used is (end date – start date) /
(cyclecount – 1).

This query uses a cycle count of 60, resulting in a 1 second resolution

for the data. The query uses the remote table view.
FROM History
AND DateTime >= '2001-01-13 12:00:00'
AND DateTime <= '2001-01-13 12:00:59'
The results are:

DateTime Value
2001-01-13 12:00:00.000 0
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
Using One Equality Operator

If one end of the time range is excluded (by using > instead of >= or <
instead of <=). then a gap of "resolution" is left at the beginning (or
end) of the result set.
The resolution is calculated as (end date – start date) / (cyclecount).
The row that equates to the time which is designated using the < (or >)
operator is not returned.
These queries use the remote table view.

Using Comparison Operators with Cyclic Retrieval and Cycle Count303
Query 1
This query uses a cycle count of 60, resulting in a 1 second resolution
for the data. The starting time is set to >=.
FROM History
AND DateTime >= '2001-01-13 12:00:00'
AND DateTime < '2001-01-13 12:01:00'
The results are:

DateTime Value
2001-01-13 12:00:00.000 0
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
Query 2
This query also uses a cycle count of 60, resulting in a 1 second
resolution for the data. The ending time is set to <=.
FROM History
AND DateTime > '2001-01-13 12:00:00'
AND DateTime <= '2001-01-13 12:01:00'

The results are:

DateTime Value
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
2001-01-13 12:01:00.000 0
Using No Equality Operators

If both ends of the time range are excluded (by using > and <) then a
gap of "resolution" is left at the beginning and end of the result set.
The resolution is calculated as (end date – start date) / (cyclecount + 1).
The row(s) that equate to the start and end times are not returned.
FROM History
AND DateTime > '2001-01-13 12:00:00'
AND DateTime < '2001-01-13 12:01:01'

Using Comparison Operators with Cyclic Retrieval and Resolution305
The results are:

DateTime Value
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
2001-01-13 12:01:00.000 0
Using Comparison Operators with Cyclic

Retrieval and Resolution
Cyclic queries that use comparison operators and the resolution time
domain extension return a set of evenly spaced rows over the specified
time span. The resolution for these rows is specified in the query.
Using Two Equality Operators

If the time range is specified using >= and <=, then the first row falls
exactly on the start time. The last row will fall exactly on the end time,
if the resolution divides exactly into the specified time duration. If the
resolution does not divide exactly into the specified time duration,
then the last row returned will be the last row satisfying (start date +
N*resolution) which has a timestamp less than the end date.
In short:
• <= endtime MAY return a last row containing the exact endtime
(but it is not guaranteed to do so)
• < endtime is guaranteed NOT to return a last row containing the
exact endtime

This query sets the resolution to 1 second.

FROM History
AND DateTime >= '2001-01-13 12:00:00'
AND DateTime <= '2001-01-13 12:01:00'
The results are:

DateTime Value
2001-01-13 12:00:00.000 0
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
2001-01-13 12:01:00.000 0
Using One Equality Operator

If the start time is excluded (by using > instead of >=), then a gap of
"resolution" is left at the beginning of the result set. In this case, the
first row returned will have the timestamp of the (start date +
resolution). If the end date uses "<" then the last row returned will be
the last row defined by (start date + N*resolution) which has a
timestamp less than the end date.
The row that equates to the time that is designated using the < (or >)
operator is not returned.
Query 1
This query uses a resolution of 1000, resulting in a 1 second resolution
for the data.. The starting time is set to >=.
FROM History
AND DateTime >= '2001-01-13 12:00:00'
AND DateTime < '2001-01-13 12:01:00'

Using Comparison Operators with Cyclic Retrieval and Resolution307
The results are:

DateTime Value
2001-01-13 12:00:00.000 0
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:55.000 55
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
Query 2
This query also uses a row resolution of 1000, resulting in a 1 second
resolution for the data. The starting time is set to <=.
FROM History
AND DateTime > '2001-01-13 12:00:00'
AND DateTime <= '2001-01-13 12:01:00'
The results are:

DateTime Value
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
2001-01-13 12:01:00.000 0

Using No Equality Operators

If both ends of the time range are excluded (by using > and <), then a
gap of resolution is left at the beginning and end of the result set.
The row(s) that equate to the start and end times are not returned.
This query uses a resolution of 1000, resulting in a 1 second resolution
for the data.
FROM v_AnalogHistory
AND DateTime > '2001-01-13 12:00:00'
AND DateTime < '2001-01-13 12:01:01'
The results are:

DateTime Value
2001-01-13 12:00:01.000 1
2001-01-13 12:00:02.000 2
2001-01-13 12:00:03.000 3
2001-01-13 12:00:04.000 4
.
.
.
2001-01-13 12:00:56.000 56
2001-01-13 12:00:57.000 57
2001-01-13 12:00:58.000 58
2001-01-13 12:00:59.000 59
2001-01-13 12:01:00.000 0
Returning Time Between Value Changes

You can return the amount of time before a tag's value changed to a
subsequent value. This time is returned using the wwResolution
column.
This functionality works with the cyclic, delta, and full retrieval
modes. The delta and full mode behavior of wwResolution does not
apply to the AnalogSummaryHistory and StateSummaryHistory
tables.
If the time change value is greater than 2,147,000,000 milliseconds
(~25 days), then the value of wwResolution column is -1.

Returning Time Between Value Changes309
Example 1: Cyclic Retrieval

For this example, the following data is stored in the Historian:
DateTime Value
2012-01-01 07:59:53 34.42384
2012-01-01 08:00:13 15.02637
2012-01-01 08:00:33 20.29732
2012-01-01 08:00:53 37.40273
2012-01-01 08:01:13 24.31662
For cyclic retrieval, cycles start at the start DateTime and occur at
intervals specified by wwResolution in the query. If you query for the
data 2012-01-01 08:00:00 to 2012-01-01 08:01:00 with four cycles, the
wwResolution column in the results show the time in milliseconds
until the next point.
DateTime Value wwResolution
2012-01-01 08:00:00 34.42384 20000
2012-01-01 08:00:20 15.02637 20000
2012-01-01 08:00:40 20.29732 20000
2012-01-01 08:01:00 37.40273 20000
Example 2: Delta and Full Retrieval

In delta and full mode, the wwResolution column is used in the query
results to show the time in milliseconds until the next point. When the
first point in the result occurs before the query start time, the
wwResolution column shows the time from the start of the query to the
next point. If there are no more points after the first point, the
wwResolution column is NULL.
For example, the following data is stored:
DateTime Value
2012-01-01 07:59:53 34.42384
If you query for the data between 2012-01-01 08:00:00 to 2012-01-01

08:00:10, the results are:
2012-01-01 08:00:00 34.42384 NULL
However, if the data stored is:

DateTime Value
2012-01-01 07:59:53 34.42384
2012-01-01 08:00:13 15.02637
Then the results are:

2012-01-01 08:00:00 34.42384 10000

When the last point in the result occurs before the query end time, the
wwResolution column shows the time until the end of the query when
there is a next available point. If there are no more points, then the
wwResolution column shows NULL.
For example, the following is stored:
DateTime Value
2012-01-01 07:59:53 34.42384
2012-01-01 08:00:13 15.02637
2012-01-01 08:00:33 20.29732
2012-01-01 08:00:53 37.40273
2012-01-01 08:01:13 24.31662
If you query for data from 2012-01-01 08:00:00 to 2012-01-01 08:05:00,

the results are:
2012-01-01 08:00:00 34.42384 13000
2012-01-01 08:00:13 15.02637 20000
2012-01-01 08:00:33 20.29732 20000
2012-01-01 08:00:53 37.40273 20000
2012-01-01 08:01:13 24.31662 NULL
If you query for data from 2012-01-01 08:00:00 to 2012-01-01 08:01:00,

the results are:
2012-01-01 08:00:00 34.42384 13000
2012-01-01 08:00:13 15.02637 20000
2012-01-01 08:00:33 20.29732 20000
2012-01-01 08:00:53 37.40273 7000
If the last point happens to be end of the query, then the wwResolution
value is zero, even when there are no more points after the last point.
For example, if you query for data from 2012-01-01 08:00:00 to
2012-01-01 08:01:13, the results are:
2012-01-01 08:00:00 34.42384 13000
2012-01-01 08:00:13 15.02637 20000
2012-01-01 08:00:33 20.29732 20000
2012-01-01 08:00:53 37.40273 20000
2012-01-01 08:01:13 24.31662 0
Example 3: Querying the WideHistory Table

If you execute a query on the WideHistory table for analog tags,
wwResolution shows the time between the first value change for ANY
of the tags.


SELECT DateTime, SysTimeSec, SysTimeMin, wwResolution
FROM WideHistory
WHERE DateTime >= "20120119 12:44:00.000"
AND DateTime <= "20120119 12:45:00.000"
')
The results are:

DateTime SysTimeSec SysTimeMin wwResolution
2012-01-19 12:44:00.0000000 0 44 1000
2012-01-19 12:44:01.0000000 1 44 1000
2012-01-19 12:44:02.0000000 2 44 1000
2012-01-19 12:44:03.0000000 3 44 1000
2012-01-19 12:44:04.0000000 4 44 1000
2012-01-19 12:44:05.0000000 5 44 1000
. . . .
. . . .
. . . .
The wwResolution column shows 1000 milliseconds because the

smallest time change is for the SysTimeSec tag, which is changing
every second.
If you run the same query using the SysTimeHour tag instead of the
SysTimeSec tag, the results are:
DateTime SysTimeHour SysTimeMin wwResolution
2012-01-19 12:44:00.0000000 12 44 60000
2012-01-19 12:45:00.0000000 12 45 0
The wwResolution column shows 60000 milliseconds because the

smallest time change is for the SysTimeMin tag, which is changing
every minute (every 60 seconds). Because the query ended at the time
of the last value, a 0 is shown for wwResolution for the ending value.
Example 4: Querying the History Table with the

wwValueSelector parameter
You can query the History table with the wwValueSelector parameter.
SELECT DateTime, TagName, Value, wwResolution
FROM History
WHERE Tagname like 'MyTag'
AND DateTime >= '2012-01-19 10:00:00'
AND DateTime <= '2012-01-19 11:00:00'
AND wwValueSelector = 'STDDEV'

The results are:

DateTime TagName Value wwResolution
2012-01-19 10:00:00.0000000 MyTag 977.157928752564 60000
2012-01-19 10:01:00.0000000 MyTag 16.5619987924163 60000
2012-01-19 10:02:00.0000000 MyTag 16.5619987924163 60000
2012-01-19 10:03:00.0000000 MyTag 16.5619987924165 60000
2012-01-19 10:04:00.0000000 MyTag 16.5619987924163 180000
2012-01-19 10:07:00.0000000 MyTag 16.5619987924171 60000
2012-01-19 10:08:00.0000000 MyTag 16.5619987924163 60000
2012-01-19 10:09:00.0000000 MyTag 16.5619987924179 60000
2012-01-19 10:10:00.0000000 MyTag 16.5619987924163 180000
2012-01-19 10:13:00.0000000 MyTag 16.5619987924195 60000
2012-01-19 10:14:00.0000000 MyTag 16.5619987924163 240000
2012-01-19 10:18:00.0000000 MyTag 16.5619987924226 180000
2012-01-19 10:21:00.0000000 MyTag 16.5619987924163 60000
2012-01-19 10:22:00.0000000 MyTag 16.5619987924226 60000
2012-01-19 10:23:00.0000000 MyTag 16.5619987924038 60000
. . . .
. . . .
. . . .
Example 5: Calculating Total Time between Value

Changes
You can calculate the total time the multiple discrete tags are in a
certain state. For example, you want to know the total time that two
pumps were on during a 24-hour period.
The following query returns a dataset that shows the time when both
discrete tags had a value of 0:
SELECT * FROM OPENQUERY(INSQL,'
SELECT DateTime, Pump1, Pump2, wwResolution
FROM WideHistory
WHERE DateTime >= "2012-03-08 16:00"
AND DateTime < "2012-03-08 17:00"
AND wwRetrievalMode="DELTA"
')
WHERE Pump1+Pump2=0
The results are:

DateTime Pump1 Pump2 wwResolution
2012-03-08 16:00:00.0000000 0 0 67
2012-03-08 16:00:01.5980000 0 0 2521
2012-03-08 16:00:04.4470000 0 0 18500
2012-03-08 16:00:23.6000000 0 0 13995

DateTime Pump1 Pump2 wwResolution

2012-03-08 16:00:37.9140000 0 0 2625
. . . .
. . . .
. . . .
The following query shows how to return the total time when both tags
had a value of 0:
SELECT SUM(wwResolution) FROM OPENQUERY(INSQL,'
SELECT DateTime, Total=Pump1+Pump2, wwResolution
FROM WideHistory
WHERE DateTime >= "2012-03-08 16:00"
AND DateTime < "2012-03-08 17:00"
AND wwRetrievalMode="DELTA"
')
WHERE Total=0
The results are:

(No column name)
2551289
If you changed the ending WHERE clause to Total>0, the returned

time would be for when more than one discrete tag was true.

SELECT INTO from a History Table

The following query inserts the specified data from the WideHistory
table into another table called MyTable. Then, the data in the
MyTable table is queried. This query uses the OPENQUERY function.
DROP TABLE MyTable
SELECT DateTime,
"Sec" = datepart(ss, DateTime),
"mS" = datepart(ms, DateTime),
ReactTemp, ReactLevel
INTO MyTable
FROM OpenQuery(INSQL, 'SELECT DateTime, ReactTemp,
ReactLevel FROM WideHistory
WHERE wwResolution = 5000
AND DateTime >= "2001-03-13 1:58pm"
AND DateTime <= "2001-03-13 2:00pm" ')
SELECT * FROM MyTable
The results are:

DateTime Sec mS ReactTemp ReactLevel
2001-03-13 13:58:00.000 0 0 190.9 2025.0
2001-03-13 13:58:00.000 5 0 190.9 2025.0
2001-03-13 13:58:00.000 10 0 168.3 1215.0
2001-03-13 13:58:00.000 15 0 168.3 1215.0
2001-03-13 13:58:00.000 20 0 133.8 315.0
2001-03-13 13:58:00.000 25 0 133.8 315.0
2001-03-13 13:58:00.000 30 0 101.6 0.0
2001-03-13 13:58:00.000 35 0 101.6 0.0
2001-03-13 13:58:00.000 40 0 32.4 750.0
2001-03-13 13:58:00.000 45 0 32.4 750.0
2001-03-13 13:58:00.000 50 0 20.9 1700.0
2001-03-13 13:58:00.000 55 0 20.9 1700.0
2001-03-13 13:59:00.000 0 0 85.9 2000.0
2001-03-13 13:59:00.000 5 0 85.9 2000.0
2001-03-13 13:59:00.000 10 0 185.9 2000.0
2001-03-13 13:59:00.000 15 0 185.9 2000.0
2001-03-13 13:59:00.000 20 0 168.3 1235.0
2001-03-13 13:59:00.000 25 0 168.3 1235.0
2001-03-13 13:59:00.000 30 0 136.1 335.0
2001-03-13 13:59:00.000 35 0 136.1 335.0
2001-03-13 13:59:00.000 40 0 103.9 -25.0
2001-03-13 13:59:00.000 45 0 103.9 -25.0
2001-03-13 13:59:00.000 50 0 34.7 625.0

Moving Data from a SQL Server Table to an Extension Table315
2001-03-13 13:59:00.000 55 0 34.7 625.0

2001-03-13 14:00:00.000 0 0 20.9 1575.0
Moving Data from a SQL Server Table to an

Extension Table
The following queries show how to insert manual data into a normal
SQL Server table and then move it into the History extension table.
First, insert the data into the SQL Server table. The following query
inserts two minutes of existing data for the SysTimeSec tag into the
ManualAnalogHistory table:
INSERT INTO ManualAnalogHistory (DateTime, TagName, Value,
Quality, QualityDetail, wwTagKey)
SELECT DateTime, TagName, Value, Quality, QualityDetail,
wwTagKey
FROM History WHERE TagName = 'SysTimeSec'
AND DateTime >= '20050329 12:00:00'
AND DateTime <= '20050329 12:02:00'
Then, create a manual tag using the System Management Console.

For a manual tag, "MDAS/Manual Acquisition" is specified as the
acquisition type. Be sure to commit the changes to the system. In this
example, a manual analog tag named MDAS1 was created.
Finally, insert the data from the ManualAnalogHistory table into
History:
INSERT INTO History (TagName, DateTime, Value, QualityDetail)
SELECT 'MDAS1', DateTime, Value, QualityDetail FROM
ManualAnalogHistory
AND DateTime >= '20050329 12:00:00'
AND DateTime <= '20050329 12:02:00'

Using Server-Side Cursors

Cursors are a very powerful feature of SQL Server. They permit
controlled movement through a record set that results from a query.
For in-depth information on cursors, see your Microsoft SQL Server
documentation.
The Wonderware Historian OLE DB Provider provides server-side
cursors. Cursors can be used to do joins that are not possible in any
other way. They can be used to join date/times from any source with
date/times in the history tables.
The following query provides an example of using a server-side cursor.
This query:
• Fetches all of the events in the EventHistory table.
• Shows a "snapshot" of three tags at the time of each event.
• Shows the event tag and its associated key value.
This query could easily be encapsulated into a stored procedure. The
query uses the four-part naming convention.
SET QUOTED_IDENTIFIER OFF
DECLARE @DateValue DateTime

DECLARE @EventTag nvarchar(256)
DECLARE @EventKey int
DECLARE @Qry1 nvarchar(500)
SELECT @Qry1 = N'SELECT EventTag = @EventTag, EventKey =

@EventKey, DateTime, TagName, Value, Quality
FROM History
WHERE TagName IN (N''SysTimeSec'', N''SysTimeMin'',
N''SysTimeHour'')
AND DateTime = '''
SELECT @Qry2 = N''''

SELECT @Qry3 = N''
DECLARE Hist_Cursor CURSOR FOR

SELECT DateTime, TagName, EventLogKey
FROM Runtime.dbo.EventHistory
OPEN Hist_Cursor
FETCH NEXT FROM Hist_Cursor INTO @DateValue, @EventTag,
@EventKey
WHILE @@FETCH_STATUS = 0

Using Server-Side Cursors317
BEGIN
SELECT @Qry3 = @Qry1 + convert(nvarchar, @DateValue, 121) +

@Qry2
--PRINT @Qry3
EXEC sp_executesql @Qry3, N'@EventTag nvarchar(256),
@EventKey int', @EventTag, @EventKey
FETCH NEXT FROM Hist_Cursor INTO @DateValue, @EventTag,
@EventKey
END
CLOSE Hist_Cursor
DEALLOCATE Hist_Cursor
The results are:

EventTag EventKey DateTime TagName Value Quality
SysStatusEvent 3 2001-01-12 SysTimeSec 27.0 0
13:00:27.000
SysStatusEvent 3 2001-01-12 SysTimeMin 0.0 0
13:00:27.000
SysStatusEvent 3 2001-01-12 SysTimeHour 13.0 0
13:00:27.000
(3 row(s) affected)
EventTag EventKey DateTime TagName Value Quality

SysStatusEvent 4 2001-01-12 SysTimeSec 28.0 0
14:00:28.000
SysStatusEvent 4 2001-01-12 SysTimeMin 0.0 0
14:00:28.000
SysStatusEvent 4 2001-01-12 SysTimeHour 14.0 0
14:00:28.000
(3 row(s) affected)

Using Stored Procedures in OLE DB Queries

Any normal SQL Server stored procedure can make use of the tables
exposed by the Wonderware Historian OLE DB Provider. Stored
procedures can use any valid Transact-SQL syntax to access
Wonderware Historian historical data.
In other words, stored procedures can make use of four-part-queries,
OPENQUERY and OPENROWSET functions, cursors, parameterized
queries and views. Stored procedures can be used to encapsulate
complex joins and other operations for easy re-use by applications and
end users.
Querying Data to a Millisecond Resolution

using SQL Server 2005
Internally, the Wonderware Historian uses a Win32 FILETIME data
type for representing date/time. This 64-bit integer yields a time
resolution of 100 nano-seconds. From a retrieval point of view,
however, the limiting factor is the resolution of the SQL Server 2005
datetime type. For SQL Server 2005, datetime represents time with a
resolution of 1/300 second, or 3.33 milliseconds. SQL Server 2005
rounds the time values in increments of .000, .003, or .007 seconds.
Note: When you are using Wonderware Historian with SQL Server
2008, millisecond resolution is supported for retrieval by using the
DateTime2 data type.
For consistency with SQL Server 2005, it is important that any OLE
DB time functions present time data according to the same convention.
A rounding algorithm is incorporated to achieve the desired results.
This is necessary when a user calls a datetime function that
manipulates milliseconds within an OPENQUERY statement. For
example:
'SELECT DATEPART(millisecond, getdate()), Value
FROM Live WHERE TagName = "ReactTemp"')
returns results in the same format as:

SELECT DATEPART(millisecond, getdate())

Querying Data to a Millisecond Resolution using SQL Server 2005319
If you require millisecond time resolution, the Wonderware Historian

provides a means of bypassing the SQL Server 2005 datetime
restriction by retrieving the datetime as a string. This is implemented
by the use of specific style specifiers for the CONVERT function. These
specifiers (numbered 909, 913, 914, 921, and 926) return a string in
the same format as the corresponding 100-series style specifiers, but
with millisecond time resolution. Style specifier 909 formats the string
in the same way as standard specifier 109; 913 formats as for 113; and
so on.
For details of the string format corresponding to each 100-series
specifier, see the description of the CONVERT(data_type, expression,
style) function in your Microsoft SQL Server 2005 documentation.
An OPENQUERY statement must be used to ensure that the
CONVERT statement is passed to the Wonderware Historian parser,
because this is a specific extension to CONVERT functionality
provided by the Wonderware Historian.
For example:
SELECT CONVERT(varchar(30), DateTime, 113), Value
FROM Live
might return '2001-08-10 14:30:45:337' (3.33 millisecond resolution),

while
'SELECT CONVERT(varchar(30), DateTime, 913), Value FROM Live
WHERE TagName = "ReactTemp" ')
would return '2001-08-10 14:30:45:335' (1 millisecond resolution).

In all, a total of ten CONVERT styles are available within the
Wonderware Historian. These are 20, 120, 21, 121, 126, 909, 913, 914,
921, and 926.

Getting Data from the OPCQualityMap Table

In general, an OPC quality has 16 significant bits. The lower 8 bits
contain the quality as described in the table, while the upper 8 bits
hold server-specific information. To ensure correct results, it is
important to consider only the lower 8 bits in a query or join involving
the OPCQualityMap table.
For example:
SELECT h.DateTime, h.TagName, h.Value, o.Description FROM
History h
INNER JOIN OPCQualityMap o
ON (h.OPCQuality & 255) = o.OPCQuality
WHERE TagName in (…)
AND …
Using Variables with the Wide Table

You cannot use variables in an OPENQUERY statement. Therefore, if
you want to use variables in a query on the wide table, you must first
build up the OPENQUERY statement "on the fly" as a string, and then
execute it.
DECLARE @sql nvarchar(1000)
DECLARE @DateStart datetime
DECLARE @DateEnd datetime
SET @DateStart = '2001-8-29 11:00:00'

SET @DateEnd = '2001-8-29 11:11:00'
SET @sql = N'select *
FROM OPENQUERY(INSQL, ''SELECT DateTime, ReactLevel, ReactTemp,

ProdLevel, BatchNumber, ConcPump, Mixer, TransferValve,
TransferPump, WaterValve, ConcValve, OutputValve, SteamValve
FROM WideHistory
WHERE DateTime >= "' + CONVERT(varchar(26), @DateStart,
113) + '"
and DateTime <= "' + CONVERT(varchar(26), @DateEnd,
113) + '"
AND wwRetrievalMode = "cyclic"'') '
EXEC sp_executesql @sql

Retrieving Data Across a Data "Hole"321
Retrieving Data Across a Data "Hole"

If the data to be retrieved spans more than one history block, and the
start time of the later block is equal (within one tick) to the end time of
the first block, you will not notice any difference than when querying
within a single block.
However, if the system has been stopped between history blocks, there
will be a "hole" in the data, as shown in the following diagram:
Upon retrieval, additional data points (labeled A and B) will be added

to mark the end of the first block's data and the beginning of the
second block's data. Point C is a stored point generated by the storage
subsystem. (Upon a restart, the first value from each IDAS will be
offset from the start time by 2 seconds and have a quality detail of
252.)
The following paragraphs explain this in more detail.
For delta retrieval, the data values in the first block are returned as
stored. After the end of the block is reached and all of the points have
been retrieved, an additional data point (A) will be inserted by
retrieval to mark the end of the data. The value for point A will be
Point A attribute Value(Hex) Value(Dec)
Value 0 0
Quality 100 256
Quality Detail 0 0

If there is no value stored at the beginning of the next block, an initial

data point (B) will be inserted by retrieval and will have the snapshot
initial value as stored. The quality and quality detail values are as
follows:
Point B attribute Value(Hex) Value(Dec)
Value Snapshot Snapshot
Quality 0 0
Quality Detail 96 150
In the case of cyclic retrieval, a point is required for each specified

time. If the time coincides with the data hole, a NULL point for that
time will be generated. The inserted points will have the values
defined in the following table.
Cyclic NULL point Value(Hex) Value(Dec)
Value 0 0
Quality 100 256
Quality Detail 0 0
If you are using time or value deadbands for delta retrieval across a
data gap, the behavior is as follows:
• For a value deadband, all NULLs will be returned and all values
immediately after a NULL will be returned. That is, the deadband
is not applied to values separated by a NULL.
• For a time deadband, null values are treated like any other value.
Time deadbands are not affected by NULLs.

Returned Values for Non-Valid Start Times323
Returned Values for Non-Valid Start Times

One example of a non-valid query start time is a start time that is
earlier than the start time of the first history block. For delta retrieval,
the first row returned will be NULL. The timestamp will be that of the
query start time. The next row returned will be timestamped at the
start of the history block and have the following attributes:
Point attribute Value(Hex) Value(Dec)
Value Snapshot Snapshot
Quality 0 0
Quality Detail 96 150
For cyclic retrieval, NULL will be returned for data values that occur
before the start of the history block.
Another non-valid start time is a start time that is later than the
current time of the Wonderware Historian computer. For delta
retrieval, a single NULL value will be returned. For cyclic retrieval, a
NULL will be returned for each data value requested.
Retrieving Data from History Blocks and the

Active Image
During initialization, the retrieval subsystem will note the oldest
timestamp for each tag in the active image. If the query start time is
more recent than the oldest timestamp, all of the requested tag values
will be obtained directly from the active image. If the query start time
is older than the oldest timestamp, data will be retrieved from the
history blocks.
If the query spans up to the oldest timestamp, the retrieval subsystem
will check the active image to re-determine oldest timestamp (because
older values in the active image are overwritten to make room for new
values). If the requested data is still in the active image, it will be
used. Otherwise, data is retrieved from the history blocks. This process
will continue until the active image can be used or the query end time
is surpassed. It is important to note that this process is applied on a
per-tag basis.
For more information, see "How the Active Image Storage Option
Affects Data Retrieval" on page 132.

Querying Aggregate Data in Different Ways

There are four different ways you can retrieve summary data, such as
an average, using the Historian.
• Using the SQL Server average function. This is appropriate for
discrete samples. For example, a check weigher, where you are
measuring individual units against a target weight.
• Using the average retrieval mode. This is appropriate for most
situations where you want to find an average, as it is weighted
according to time. For example, if you want to find the average for
a flow rate or a temperature.
• Setting up summary replication and then querying the
AnalogSummaryHistory table. Replication uses the average
retrieval mode to do the calculations.
• Setting up a summary event and then querying the SummaryData
table. The event subsystem uses the SQL Server average function.
The following examples show how you can get the same data using
these different methods. All examples use the SysTimeSec system tag,
which has a range of 0 to 59.
Query 1
The following query uses the SQL Server average function to return
the average value of the SysTimeSec tag over the span of one minute.
SELECT AVG(Value) as 'SysTimeSec AVG'
FROM History
AND DateTime > '2009-11-15 6:30:00'
AND DateTime < '2009-11-15 6:31:00'
AND wwRetrievalMode = 'Full'
The results are:

SysTimeSec AVG
29.5

Querying Aggregate Data in Different Ways325
Query 2
The following query uses the historian time-weighted average
retrieval mode to return the average for the same time period. Because
the cycle count is set to 2, a first row is returned for the
“phantom”cycle leading up to the query start time. The StartDateTime
column shows the time stamp at the start of the data sampling, which
is the start time of the phantom cycle. The second row returned
reflects is the actual data that you expect. The time stamp for the data
value is 2009-11-15 06:31:00 because the default time stamping rule is
set so that the ending time stamp for the cycle is returned. For more
information about the phantom cycle, see "About "Phantom" Cycles"
on page 230.
SELECT StartDateTime, DateTime, TagName, Value
FROM History
AND DateTime >= '2009-11-15 6:30:00'
AND DateTime <= '2009-11-15 6:31:00'
AND wwTimeStampRule = 'end'
The results are:

StartDateTime DateTime TagName Value
2009-11-15 06:29:00 2009-11-15 06:30:00 SysTimeSec 29.5
2009-11-15 06:30:00 2009-11-15 06:31:00 SysTimeSec 29.5
Query 3
For the following query, local replication has been set up so that the
average of the SysTimeSec tag is calculated every minute and stored
to the SysTimeSec.1M analog summary tag. The query returns the
value of the SysTimeSec.1M tag for the time period specified.
SELECT TagName, StartDateTime, EndDateTime, Average as AVG
WHERE TagName = 'SysTimeSec.1M'
AND StartDateTime >= '2009-11-15 6:30:00'
AND EndDateTime <= '2009-11-15 6:31:00'
The results are:

TagName StartDateTime EndDateTime AVG
SysTimeSec.1M 2009-11-15 06:30:00 2009-11-15 06:31:00 29.5

Query 4
The following query, the History table is used instead of the
AnalogSummaryHistory table. Because the cycle count is set to 2, this
query returns a row for the phantom cycle. The time stamp for the
data value is 2009-11-15 06:31:00 because the default time stamping
rule is set so that the ending time stamp for the cycle is returned.
SELECT TagName, DateTime, Value
FROM History
WHERE TagName = 'SysTimeSec.1M'
AND DateTime >= '2009-11-15 6:30:00'
AND DateTime <= '2009-11-15 6:31:00'
AND wwRetrievalMode = 'avg'
The results:
TagName DateTime Value
SysTimeSec.1M 2009-11-15 06:30:00 29.5
SysTimeSec.1M 2009-11-15 06:31:00 29.5
Query 5
The following query returns five minutes of summary data for an
event tag that has been configured to store the average value of the
SysTimeSec tag every minute.
SELECT TagName, CalcType, SummaryDate, Value
FROM v_SummaryData
AND SummaryDate >= '2009-11-15 18:30:00'
AND SummaryDate <= '2009-11-15 18:31:00'
The results are:

TagName CalcType SummaryDate Value
SysTimeSec AVG 2009-11-15 18:30:00.000 29.5
SysTimeSec AVG 2009-11-15 18:31:00.000 29.5

327
Chapter 10
Replication Subsystem
Data from one Wonderware Historian can be replicated to one or more

other Wonderware Historians, creating a “tiered” relationship between
the historians.
About Tiered Historians

You can set up Wonderware Historians in a variety of tiered
configurations. In a common configuration, data from multiple
individual historians (called tier-1 historians) is fed into a single
centralized historian (called a tier-2 historian).

328  Chapter 10 Replication Subsystem
The tier-2 historian stores data replicated from the tier-1 historians.
Tier-2: Centralized
reporting and system
of record
Tier-1 Local
troubleshooting and
buffering
I/O
InTouch Application Server
Another configuration is to have multiple tier-1 historians that feed

information to multiple tier-2 historians in a many-to-many
relationship.
New Orleans Houston
Tier-2
Tier-1

How Tags are Used During Replication329
How Tags are Used During Replication

Data from a tier-1 historian is replicated to a tier-2 historian using
tags in the same way that information is collected by an individual
Wonderware historian.
The tier for a tag is determined by where it comes from:
• Values for tier-1 tags are gathered directly from an IDAS, MDAS,
or HCAL.
• Values for tier-2 tags come from another Wonderware Historian
server.
A historian can act as a tier-1 and a tier-2 historian simultaneously.
A typical scenario for a tiered historian appears in the following
example. Tag1 is collected on historian A and all its values are
replicated to historian B, where they are stored as values of tag2. At
the same time historian B collects data for its tag3 and all its values
are replicated to historian C, where they are stored as values of tag4.
Historian A Historian B Historian C

(tier-1 historian) (tier-1 and tier-2 (tier-2 historian)
historian)
tag1 tag2
replicated to tag2 is a
tag2 on tier-2 tag of
Historian B Historian B
tag1 is a
tier-1 tag of
Historian A
tag3 is a tag4 is a
tier-1 tag of tier-2 tag of
Historian B Historian C
tag3 tag4
replicated to
tag4 on
Historian C

In this example, the tags are identified as follows:

• tag1 is a tier-1 tag of historian A
• tag2 is a tier-2 tag of historian B
• tag3 is a tier-1 tag of historian B
• tag4 is a tier-2 tag of historian C
• historian A is a tier-1 historian
• historian B is both tier-1 and tier-2 historian
• historian C is a tier-2 historian
Important: Be careful not to create or modify a replicated tag on a

tier-1 historian to have the same tagname that already exists on a
tier-2 historian. The system does not prevent you from having a
replicated tag on a tier-2 historian receiving data from two or more
different tier-1 historians. However, when you retrieve data for that
replicated tag on the tier-2 historian using the tagname, an incorrect
blend of data from the two (or more) data sources is returned.
There are two types of replication: simple replication and summary

replication. Summary replication provides periodic summaries of high
resolution data, while simple replication retains the original data
resolution.

Simple Replication331
Simple Replication
When a tag is configured for simple replication, all values stored in the
tier 1 historian are replicated to the tier 2 server. Analog, discrete, and
string tags can be configured for simple replication. Replicated tags of
a tier-2 historian cannot be configured for further replication.
Tier-2 example: 1-
second data
Replicate all
data for tags
Tier-1 example: 1-second data
The results of replication are stored on the tier-2 historian using the
same tag type as was used for the tag on the tier-1 historian.
Simple replication of tag values occurs:
• Every time a new stored data value arrives into the active image of
the tier-1 historian.
• Every time there is a change made in the history of the tier-1 tag to
keep tier-1 and tier-2 tag values synchronized in history. For more
information, see "How Replication is Handled for Different Types
of Data" on page 339.

Summary Replication
Summary replication involves analyzing and storing statistical
information about the tag value at the tier-1 historian. This occurs at
an interval you specify, called the calculation cycle. The algorithm
used for the summaries is the same as for the summary retrieval
modes. However, when a summary computation is performed on a
tier-1 tag with multiple values at the same time, the order of these
values cannot be guaranteed and the summary may differ from the
summary computed by the replication service at run time.
The result of the calculation is sent to the tier-2 historian to be stored
with the timestamp of the cycle. Tier-2 tags are not dependent on the
"real-time window" that applies to tags that use classic storage.
Tier-2 example:
5-minute, hourly, daily rate
Many aggregate
values for each
“summary” tag
Tier-1 example: 1-second data
There are two types of summary replication:

• Analog summary replication
• State summary replication
The results of summaries are stored on the tier-2 historian using
analog summary or state summary tags.
If a tier-1 historian is unable to perform a scheduled summary
calculation for any reason, it adds a record about the event into a
replication queue. When there are enough system resources available,
or there is a specific event from another subsystem, the tier-1 historian
can perform the summary calculations and clear the queue.

Summary Replication333
Analog Summary Replication

Analog summary replication produces summary statistics for analog
tags. The statistics relate only to the recorded interval. Statistics
available are:
• Time-weighted average
• Standard deviation
• Integral
• First value in a period with timestamp
• Last value in a period with timestamp
• Minimum value in a period with timestamp
• Maximum value in a period with timestamp
• Start time of summary period
• End time of summary period
• OPC Quality
• Percentage of values with Good quality
• Value
When you retrieve the data, you specify which calculation you want to
return. For more information, see "Querying the
AnalogSummaryHistory View" on page 281.
The functionality provided by analog summary replication is similar to
using the minimum, maximum, average, and integral retrieval modes.
For a comparison example, see "Querying Aggregate Data in Different
Ways" on page 324.
When you use the Wonderware Historian SDK to retrieve analog
summary tag data, the values returned through the SDK for analog
summary tags from history correspond to the "Last" values in the
AnalogSummaryHistory table when using defaults. Use the
corresponding retrieval mode to get the minimum, maximum, average,
slope, and integral values.

State Summary Replication

State summary replication summarizes the states of a tag value. State
summary replication can be applied to analog (integer only), discrete,
and string tags.
You use this for analyzing process variables with a limited number of
states, such as a machine’s state of running/starting/stopping/off.
State summary replication provides the following, for each distinct
state:
• Total time
• Percent of the cycle
• Shortest time
• Longest time
• Average time
• OPC Quality
• Value
A state summary results in a series of values, each representing a
different state, for the same tag and time period.
When you retrieve the data, you specify which calculation you want to
return. For more information, see "Querying the
StateSummaryHistory View" on page 282.
The functionality provided by analog summary replication is similar to
using the ValueState and RoundTrip retrieval modes.
You can define state summary replication for a large number of states,
but state data is dropped if the number of states occurring in the same
reporting period exceeds the maximum number of states allowed. You
configure the maximum states when you create the state summary
tag. The default number of maximum states is 10. The replication
subsystem will calculate summaries for the first 10 distinct states, in
the order in which they occur in the data (not in numeric or alphabetic
order). Be aware that the higher the number of maximum states, the
more system resources are used to keep track of the time-in-state for
each distinct state.

Replication Schedules335
Replication Schedules
Each real-time summary has a specified schedule, by which the
summary is calculated and sent to the tier-2 historian to be stored
with the timestamp of the cycle.
There are two types of replication schedules:
• Periodic replication schedules
You can configure a summary to replicate based on an cycle such
as 1 minute, 5 minutes, 1 hour, 1 day, and so on. The cycle
boundaries are calculated starting at midnight, tier-1 server local
time, and continue in fixed time increments. The time between
cycles is constant within a day even through a daylight savings
change. Note that the last cycle in a day may be shorter to force
replication at midnight. The calculation cycle starts at midnight.
For example, a 13-minute cycle is stored at 12:00 a.m., 12:13,
12:26, ... 11:27 p.m., 11:40, 11:53 and then again at 12:00 a.m.
• Custom replication schedules
Custom schedules force replication cycles to occur at fixed times of
the day in tier-1 server local time. You choose the fixed times of
day.

Replication Schedules and Daylight Savings Time

Daylight Savings Time affects replication schedules that are triggered
according to a time period, such as every hour, every thirty minutes,
and so on. Replication schedules that are triggered at a fixed time that
you specify are not affected.
In the following examples, the time change occurs at 2:00 a.m.
In this example, the summary period is configured to be every 30
minutes. On the “fall back” day, will be two extra summaries
performed during the repeated hour for that day. For the “spring
forward” day, there will be two summaries missing because of the
skipped hour. The next replication occurs at the next scheduled time.
In this case, it would be 3:00 a.m.
Summary Period = 30 Minutes
“Fall Back” Day Regular Day “Spring Forward” Day
1:00 a.m. Daylight 1:00 a.m. Standard 1:00 a.m. Standard

1:30 a.m. Daylight 1:30 a.m. Standard 1:30 a.m. Standard
1:00 a.m. Standard 2:00 a.m. Standard
1:30 a.m. Standard 2:30 a.m. Standard
2:00 a.m. Standard 3:00 a.m. Standard 3:00 a.m. Daylight time gap
extra summaries 2:30 a.m. Standard 3:30 a.m. Standard 3:30 a.m. Daylight
3:00 a.m. Standard
3:30 a.m. Standard
In the next example, the summary period is configured for every four
hours. The scheduled summaries do not fall exactly on or within the
boundaries of the time change hour. In this case, on the “fall back” day,
the summary subsequent to the time change hour includes four hours
of data for the “fall back” day. An extra summary for an hour’s worth of
data is performed at the end of the “fall back” day. On the “spring
forward” day, the summary period that contains the skipped hour
includes one less hour of data.
Summary Period = 4 Hours
0:00 Daylight 0:00 Standard 0:00 Standard

3:00 Standard 4:00 Standard 5:00 Daylight
11:00 Standard 12:00 Standard 4 hours of data
4 hours of data 14:00 Daylight
15:00 Standard 16:00 Standard summarized
summarized 17:00 Daylight
0:00 Standard
missing an hour
“extra” hour here

Replication Groups337
For a custom summary period, the summaries always occur at the

fixed times of day that you specify in local time. However, the
summary includes and extra hour of data for the “fall back” day
(because of the overlap hour) and for the “spring forward” day (because
of the skipped hour).
Summary Period = Custom: 0, 4, 8, 12, 16, 20
0:00 Daylight 0:00 Standard 0:00 Standard

extra hour - 12:00 Standard 12:00 Standard 12:00 Daylight only 3 hours of
5 hours of data 16:00 Standard 16:00 Standard 16:00 Daylight data summarized
summarized 20:00 Standard 20:00 Standard 20:00 Daylight
If a Daylight Savings Time change causes a scheduled time to be

ambiguous, such as 1:30 a.m. on a "fall back" day when the clock jumps
from 1:59 a.m. Daylight Savings Time to 1:00 a.m. standard time and
the time could be interpreted as 1:30 a.m. Daylight Savings Time or
1:30 a.m. standard time, the replication will occur at the latter of the
two occurrences. In this case it would be 1:30 a.m. standard time.
Replication Groups
A replication group abstracts a tag from a schedule. You can assign
multiple summary tags to a single replication group.
Replication Schedule
1 Day
Summary Tags
Summary Tag A
Summary Tag B
Replication 1 Day Summary Tag C
Server A
.
Replication Group Summary Tag N

Multiple groups can be assigned to a single schedule. This simplifies

maintenance. If you need to edit the schedule (for example, change the
time of day for a shift end), you only need to edit the replication
schedule, not the individual groups or summary tag configurations.
8-hour shift
Summary Tags
Summary Tag A
My Group A Summary Tag B
Replication Summary Tag C
Server A
Summary Tag D
My Group B Summary Tag E
Summary Tag F
Replication Groups
A replication group must be unique for a type of summary tag, either

analog or state. You can, however, have the same group name for
analog summary tags as you do for state summary tags. You can also
have the same replication group defined in multiple servers.
8-hour shift
Summary Tags
Replication Summary Tag A

Server A My Group A Summary Tag B
Summary Tag C
Summary Tag D
Replication My Group A Summary Tag E
Server B Summary Tag F
Replication Groups

How Replication is Handled for Different Types of Data339
How Replication is Handled for Different Types

of Data
An accurate mapping between the data on the tier-1 and tier-2
historians is maintained over time. This mapping includes both tag
configuration and data synchronization.
Replication is unidirectional, from tier-1 to tier-2 historians. If the
data on the tier-2 historian is changed in any way, the system does not
try to map the change back to the tier-1 historian.
Assume that some tag1 of historian A is configured to be replicated in
real-time to tag2 of historian B. The tag2 of historian B will have
exactly the same data and OPC quality values as tag1 of historian A.
The replication system performs the following actions:
• When a new original value fitting the real-time window gets stored
on historian A, it gets transmitted and stored on historian B, as
well as the original value.
• If you perform an insert or update operation for some old values of
the historian A, the same change is reflected on historian B.
• If some store-and-forward data gets merged into history on
historian A, the same data gets transmitted to historian B and gets
merged into history of historian B.
Replication is implemented in two ways: streaming replication and
queued replication. The replication system uses a combination of
streamed replication and queued replication as required.
Streaming Replication
When values of tier-1 tags are received from an IDAS or HCAL
(Wonderware Application Server) and arrive at the tier-1 historian as
a time-ordered data stream directly, the historian not only stores the
data, but also forwards it to the replication subsystem if replication is
configured for those tags.
Then the replication subsystem immediately streams that data to the
tier-2 historian for simple replication, or performs summary
calculations and streams the resulting summaries.
This happens equally efficiently for tag values of timestamps close to
the current system time and for late data tags.

If the tier-2 historian becomes unavailable, the replication subsystem

continues to stream replicated data into the local store-and-forward
path. When the connection is restored, all replicated data is sent as
compressed snapshots to the tier-2 historian and incorporated into
history.
Streaming replication is the fastest and most efficient way of data
replication, but there are some scenarios where it cannot be used and
another method called queued replication is applied.
Queued Replication
Queued replication is used for scenarios in which streaming
replication is not appropriate, such as the following:
• An interruption occurs in the tier-1 data stream, such as when a
remote IDAS configured for store-and-forward cannot
communicate with the tier-1 historian for a long period of time.
When the connection is restored and the store-and-forward data
finally arrives at the tier-1 historian, it may be already streaming
newer data.
• An insert, update, or CSV file import operation could be performed
for tier-1 tag values, so the summaries should be recalculated for
that time period and re-replicated to the tier-2 historian.
• If the tier-1 historian is started or stopped, and there are some
summaries spanning across the startup/shutdown time that must
be recalculated and re-replicated to the tier-2 historian.
In such cases, the replication subsystem receives notifications from the
manual data storage service that contain information about what kind
of tier-1 tag data (original or latest) has changed for a particular time
interval. Then the replication subsystem places that notification
record into the replication queue stored in the Runtime database of the
tier-1 historian. Later, when the connection to the tier-2 historian is
restored, the replication subsystem processes that queue by querying
the tier-1 data and replicating it to the tier-2 historian. As soon as a
queue item is successfully processed, it is removed from the replication
queue.
Although the replication subsystem optimizes the queue by combining
adjacent records, queued replication is slower and requires more
system resources as compared to streamed replication, because it
involves querying tier-1 data already stored on disk.
Queued replication does not support data values with timestamps
before the year 1970.

Tag Configuration Synchronization between Tiered Historians341
Tag Configuration Synchronization between

Tiered Historians
If a summary tag gets deleted on the tier-1 historian, its corresponding
tag on the tier-2 historian remains intact to allow for retrieval of data
already collected. A tier-1 tag cannot be deleted from the tier-1
historian if it is being replicated--first delete the tag replication and
then delete the tier-1 tag. A tier-2 tag can be deleted from the tier-2
historian, but it should only be deleted after the corresponding
replication has been deleted from the tier-1 source historian.
Otherwise, it will be recreated.
Replication Run-time Operations

When a tier-1 historian cannot communicate with a tier-2 historian
because of a network outage or other reason, the replicated tags can
still be configured and the data collected. When the tier-2 historian
becomes available, the replication configuration and data are sent to
the tier-2 historian.
System and data integrity is not guaranteed if a disorderly shutdown
occurs, such as a power outage.
Replication for tags will stop if:
• You delete the source tag configuration on the tier-1 Historian.
• You configure the tier-1 tag so that its data values are not stored.
• An integer analog tag is being replicated as a state summary, and
you change the source tag to be a real analog tag.
Replication Latency
Replication latency is the time it takes for the system to make a value
available for retrieval on tier-2 from the moment it was stored or
calculated on tier-1.
Replication latency depends primarily on whether the streaming or
queued replication method is being applied in each particular case and
the available system resources to execute that method in each
particular case.
Streaming replication tends to have a shorter latency period than
queued replication as it deals with smaller amounts of data and is
processed at a higher priority.

Replication Delay for "Old" Data

The replication delay identifies how frequently "old" data, which
includes inserts, updates, and store-and-forward data, is sent from the
tier1 historian to the tier-2 historian. The replication delay applies
only to queued replication.
You specify the delay using the OldDataSynchronizationDelay system
parameter. For more information, see "System Parameters" on
page 36.
This delay represents your intent, while the replication latency
identifies the real time difference. If the latency period becomes longer
than the replication delay, the system will not be able to maintain the
expected replication.
If you set the OldDataSynchronizationDelay system parameter to
zero, all changes detected in the tier-1 are immediately sent to the
tier-2, which may be very inefficient for certain applications.
Continuous Operation
If a tier-2 historian becomes unavailable and is configured for
store-and-forward operations, you can still add, modify and delete
replication and summary objects in the local configuration of tier-1,
and store data locally for tier-2 tags created before tier-2 became
unavailable or while it is still unavailable.
After the tier-2 historian is available, the Replication Service
compares the latest replication and summary objects with the tier-2
tags currently existing on the tier-2 historian and performs a dynamic
reconfiguration to ensure all data is synchronized. The reconfiguration
history that was stored locally is also sent to the tier-2 historian and
merged into its history. It will appear as though the disconnection
between the tiers never took place.
Overflow Protection
If the tier-2 historian is unable to handle the incoming data from the
tier-1 historian, the Replication Service detects the situation and
switches into store-and-forward mode, where the data gets
accumulated locally until the storage limit is reached. If the limit is
reached, all data to be sent to the tier-2 historian gets discarded and
an error message is logged.

Security for Data Replication343
Security for Data Replication

Connections from a tier-1 historian to tier-2 historian have to be
authenticated before any replication task can be performed on the
tier-2 historian.
A local Windows user group called aaReplicationUsers is created on
the tier-2 historian during the tier-2 historian installation. The
ArchestrA user account is automatically added to this group. Only
members of the aaReplicationUsers group are allowed to perform
replication tasks including adding, modifying, and sending values for
replication tags. This group is not allowed to perform other
non-replication tasks, such as adding or modifying a tier-1 tag.
When you configure a replication server on the tier-1 historian, you
must specify a valid Windows user account on the tier-2 historian for
the replication service to use. This security account does not have to be
a valid account on the tier-1 historian or even be in the same security
domain as the tier-1 historian. If no replication user credentials are
configured in at the tier-1 historian, the ArchestrA user account
credential is passed to the tier-2 historian for authentication.
Using Summary Replication instead of

Event-Based Summaries
You can use summary replication instead of event-based summaries.
Compared to event-based summaries, replicated summaries:
• Support more types of calculations.
• Are stored to history blocks, instead of in SQL Server tables, thus
taking advantage of all that the Wonderware Historian storage
subsystem has to offer in terms of data compression and
throughput as compared to regular SQL Server database storage.
• Can be retrieved using the full array of real-time query extensions
provided by the Wonderware Historian.
For more information on event-based summaries, see Chapter 11,
"Event Subsystem."


345
Chapter 11
Event Subsystem
Plant events range from startups and shutdowns, through trips and
shift changes, to batch events and operator actions.
You can use the Wonderware Historian event subsystem to detect
events and associate actions when they are detected. At a basic level,
anything that can be determined by examining stored data can be used
as an event. The event subsystem can be configured to periodically
check to see if an event occurred. This is called event detection. A
subsequent action can then be triggered after an event is detected.
However, there is no guarantee of immediacy for actions; in fact, other
mechanisms can preempt actions under certain circumstances.
For the historian, event storage encapsulates more than just the fact
that something happened. An event is the set of attributes describing
the moment a detection criterion is met on historical tag values in the
historian. Attributes of an event include the date and time that the
event occurred in history and the date and time that it was detected.
Records of detected events can be logged to the database regardless of
whether or not any configured actions are subsequently initiated. In
other words, sometimes it may be desirable to simply log the fact that
an event occurred without initiating an action. The opposite may be
true, as well.

346  Chapter 11 Event Subsystem
In short, the event subsystem performs the following basic functions:

• Detects when events occur by comparing sets of criteria against
historical data in the database.
• Optionally logs event records to a dedicated SQL server table
(EventHistory).
• Optionally triggers a configured action each time an event is
occurs.
For information about configuring events, see Chapter 11,
"Configuring Events," in the Wonderware Historian Administration
Guide.
The event subsystem does not support Daylight Savings Time
changes. The replication subsystem, however, does handle Daylight
Savings Time changes, and you can use replication to generate data
summaries according to a schedule. For more information, see Chapter
10, "Replication Subsystem."
Event Subsystem Components

The following table describes the components of the event subsystem.
Configuration Editor Part of the System Management Console. Used to specify event
definitions and possible actions.
Runtime database Stores event definition information and all data generated by the
event subsystem, such as records of event detections, data
summaries, and data snapshots.
Event System Service Internal process that coordinates event detection and action
(aahEventSvc.exe) functions. This process runs as a Windows service. Using the System
Management Console, you can configure the event service to
automatically start and stop at the same time as the Wonderware
Historian. The event service is responsible for:
• Reading event definition information from the Runtime database.
• Creating event detectors and actions, including allocating the
necessary processing threads and establishing database
connections.
• Initiating the event detection cycle.
SQL variables Available for use in event queries.
You use the System Management Console to configure the event

subsystem.

Uses for the Event Subsystem347
Uses for the Event Subsystem

Generally, you should use the Wonderware Historian event subsystem
to monitor non-critical system conditions that occur only occasionally.
For example, possible event detections that you can set up include:
• Detect all occurrences in history when the value of a discrete tag is
set to 0
• Detect if the system clock is set to a specified date and/or time
• Determine the state of information in the database by a SQL
statement
You can use event actions to perform tasks such as the following:
• Send e-mail messages to remind managers about weekly
maintenance checks
• Summarize plant data to create a statistical analysis over defined
periods of time
• Take "snapshots" of system data
• Modify storage conditions (such as time and value deadbands)
• Generally perform any database-related task
The event subsystem is not designed to transfer data to and from the
database continually and should not be used in this manner. The only
exception is for summary actions; the event subsystem can continually
process data aggregates so that they are available for reporting
purposes.
The event subsystem should not be used as an alarm system. An alarm
system such as provided with InTouch HMI software can be used to
alert operators to specific satisfied conditions. The InTouch alarm
system is intended as a notification system to inform operators of
process and system conditions promptly upon their occurrence. The
InTouch alarm system supports displaying, logging, and printing
capabilities for process alarms and system events. (Alarms represent
warnings of process conditions, while events represent normal system
status messages.) For more information on the InTouch alarm system,
see your InTouch documentation.
In contrast, the event subsystem is intended to initiate actions based
upon historical event detection. An alarm system presupposes an
immediate message response is propagated for all configured alarms
at the time the respective conditions are met. In this sense, the
historian event subsystem is not an alarm system. The event
subsystem queues up detected events and processes them accordingly
based upon pre-configured priorities.

Event Subsystem Features and Benefits

You can obtain a number of distinct operational benefits from properly
using the features of the event subsystem. A list of key benefits is as
follows:
• Unlike real-time alarming, the event subsystem determines events
from stored historical data and is not dependent on real-time
detection. No events are missed unless the machine is severely
overloaded for a long period of time.
• The event subsystem is SQL-based, thus providing a means of
managing database-related tasks within the system. You can use
custom SQL queries as detectors, as well as create custom
SQL-based actions.
• A number of pre-configured detectors and actions are available.
• Detections may be made by external sources. (A COM mechanism
is available for invoking the detector in the event subsystem.)
• Time-based detection (based on the system clock time) allows you
to schedule certain tasks, such as data aggregations (summaries).
• The event subsystem is designed to manage overload situations. If
the system is currently busy due to some other processing for a
period of time, the event subsystem will "catch up" at a later time
during off-peak periods. If the overall Wonderware Historian is
continuously overloaded, the event subsystem degrades in
functionality gracefully.
• You can select which actions have priority and can assign certain
actions (preferably only a few) never to be compromised, even
under overload conditions.
• System tags are available to monitor event subsystem conditions.

Event Subsystem Performance Factors349
Event Subsystem Performance Factors

The overall performance of the Wonderware Historian event
subsystem is subject to factors related to data storage and query
processing time. Too often, systems are commissioned with
specifications that estimate average or "typical" expected loading.
Instead, you should size the system so that it can accommodate the
peak load that you expect during the projected system life cycle. Some
performance factors you should consider are:
• Sufficient hardware. Your selection of hardware is important to
guarantee peak performance for the range of behaviors required
for a given operating environment. For example, you should make
sure that you have enough disk space to store the records of
detected events and the results of any actions (summaries, value
snapshots, and so on).
• Processor availability. The event subsystem is subject to processor
availability as much as any other software sharing a common
platform. At any given moment, multiple processes contend for
processor time.
• Nature of the database queries executed by the event subsystem.
For example, because event subsystem actions typically operate on
normal SQL Server tables, they are subject to performance
limitations of the Microsoft SQL Server. Also, query activity tends
to be very CPU-intensive and is extremely sensitive to other
concurrent activities being performed on the same server.
• Time intervals for SQL-based detectors. For more information, see
"Time Intervals for SQL-Based Detectors" on page 352.
Performance can vary greatly for the same event task, depending upon
the computer configuration, user interaction, and other unpredictable
activity common in a plant situation with shared database and server
resources. It is often very difficult to determine precisely what
combinations of hardware and software parameters will work
optimally for your required operating environment. Therefore, you
should test your event subsystem configuration before running it in a
production environment to make sure that the system will not become
overloaded during peak use.

Event Tags
An event tag is a name for an event definition in the system. For
example, if you want to detect an event when a tank temperature
reaches 100 degrees, you can define an event tag and name it
"TankAt100." Event tags differ from the other tag types in the
Wonderware Historian (analog, discrete, and string). Analog, discrete,
and string tag types are the definitions of variables to be stored. In
contrast, an event tag is a named reference for the definition of the
specific event you want to detect, including an optional action to
perform when the event is detected. An event tag provides a way to
reference all event definition information in the system.
Event tags are created and maintained using the System Management
Console. When you define an event tag, you must specify:
• A name, description, and other general configuration information.
• The event criteria, which describes the conditions that must exist
for the event and how often the event subsystem checks to see if an
event occurred.
• Whether or not to log the event detection.
• Whether or not to enable or disable event detection.
• An optional action that is triggered when an event is detected.
Event Detectors
Each event tag must have an associated event detector. An event
detector is a mechanism for determining when the set of event criteria
for an event tag has been satisfied. When you configure an event
detector, you must first configure its type and then configure the
parameters associated with that detector type. You can choose from
the following types of event detectors:
• SQL-Based Detectors
• Schedule Detectors
• External Detectors
The generic SQL, analog specific value, and discrete specific value
detectors are SQL-based detectors. The schedule detector is a
time-based detector. The external detector is used when triggering an
event by the ActiveEvent ActiveX control.
For all detectors, the event subsystem will initially base the query for
data in history at the time the event subsystem starts. Subsequently,
the event subsystem will base the query on the last successful
detection; that is, the time of the most recent detection becomes the
starting time for the next detection.

Event Detectors351
SQL-Based Detectors
Analog specific value, discrete specific value, and generic SQL
detectors operate on data stored in the database. The detection criteria
for each of these detectors is a SQL statement that is executed against
the Wonderware Historian. Generic SQL detectors can query against
both the historian and Microsoft SQL Server.
Generic SQL Detectors

A generic SQL detector detects an event based on criteria that are
specified in a SQL statement. You can use pre-configured SQL
templates that are stored in the database as the basis for your script,
or you can create your own script from scratch.
To use a pre-configured SQL template, simply select it from a list of
available templates when defining the event tag.
If you create a new script, you will need to add it to the SQLTemplates
table in the Runtime database in order for it to appear in the list of
pre-configured templates. You should test your SQL queries in SQL
Server Query Analyzer before using them in a generic SQL event
detector.
Specific Value Detectors

Two specific value detectors are available:
• Analog specific value detector
• Discrete specific value detector
These detectors can be used to detect if a historical tag value matches
the state defined by the detector criteria. For the criteria, historical
values are compared to a target value that you specify. If a value
matches the criteria, then an event is logged into the EventHistory
table, and any associated actions will be triggered. For example, an
analog specific value detector could be configured to detect if the value
of 'MyAnalogTag' was ever greater than 1500. Likewise, a discrete
value detector could be configured to detect if the value of
'MyDiscreteTag' was ever equal to 0.
For a specific value detectors, you can apply either edge detection or a
resolution to the returned data. The resolution is used only when the
edge detection is set to NONE (in which case the retrieval mode is
cyclic). For more information, see "Resolution (Values Spaced Every X
ms) (wwResolution)" on page 228 and "Edge Detection for Events
(wwEdgeDetection)" on page 266.

Time Intervals for SQL-Based Detectors

For SQL-based detectors, you must specify a time interval that
indicates how often the detector will execute. The time interval is very
important in that it affects both the response rate of any event actions
and the overall performance of the system.
The detection of an event may occur significantly later than the actual
time that the event occurred, depending on the value you specify for
the time interval. The time between when an event actually occurred
in history and when it was detected is called latency.
For example, you configure a detector to detect a particular event
based on a time interval of 10,000 ms (10 seconds). This means that
every 10 seconds, the event detector will check to see if the event
occurred. If the event occurs 2,000 ms (2 sec) after the last check, the
event detector will not detect that the event occurred until the full 10
seconds has elapsed. Thus, if you want a greater possibility of
detecting an event sooner, you should set the time interval to a lower
value.
Also, the time interval affects when an associated action will occur,
because there could be some actions that are queued to a time equal to
or greater than the interval.
The following are recommendations for assigning time intervals:
• When configuring multiple event detectors, distribute them evenly
across multiple time intervals; don't assign them all to the same
interval.
All configured detectors are first divided into groups, based on
their assigned time interval. The detectors are then sequentially
ordered for processing in the time interval group. The more
detectors assigned to a particular time interval, the longer it will
take the system to finally process the last one in the group. While
this should not have a negative impact on actual detection of
events, it may add to increased latency.
• Avoid assigning a faster time interval than is really necessary.
The time interval for detectors should not be confused with a rate
required by a real-time system that needs to sample and catch the
changes. For the event subsystem, a slower time interval simply
means that more rows are returned for each scan of the history
data; no events are lost unless then detection window is exceeded
(for more information, see "Detector overloads" on page 362). For
example, you create an event tag with a detector time interval of 1
minute, and you expect an event to occur every 5 seconds. This
means that the system would detect 12 events at each time
interval. In most cases, this is an acceptable rate of detection. Also,
assigning short time intervals will result in higher CPU loading
and may lead to degraded performance.

Event Detectors353
For detailed information on how detectors are executed, see "Event

Subsystem Resource Management" on page 358.
The EventHistory table can be used to determine if too many event
tags have the same time interval. If the latency between when the
event actually occurs (stored in the DateTime column) and when it
was detected (stored in the DetectDateTime column) is constantly
growing and/or multiple event occurrences are being detected during
the same detector time interval, you need to move some of the event
detectors to a different time interval.
Schedule Detectors
The schedule detector is a time-based detector. A schedule detector
detects whether the system clock is equal to or greater than a specific
date and/or time. For example, you could log an event every week on
Monday at 2:00 p.m.
Schedule detectors are different from other detectors in that they are
real-time detectors. The value of the system clock is checked every
second. Schedule detectors are very fast and can be used without great
concern about efficiency. Thus, a schedule detector provides the only
real-time event processing. However, there is no guarantee of when
the action will occur.
All of the schedule detectors that you set up are handled by a
dedicated scheduling thread. This allows for a separation between the
processing load needed to execute schedule detectors and the
processing load needed to perform all of the other event work. The
scheduling thread will maintain a list of detection times in a time
queue. If you add a schedule detector, the thread will register the
detection time in the queue and then re-sort the list of all detection
times from the earliest to the latest.
The time of the system clock is then compared with the time of the first
item in the schedule queue. If the system clock time is equal to or
greater than the time of the first item, the detection algorithm for the
first item will be invoked and the detection will be performed.
The event subsystem does not account for Daylight Savings Time
changes. If you set up a schedule detector that runs periodically with a
specified start time, you will need to change the start time to reflect
the time change. Another solution would be to use the time-weighted
average retrieval mode instead of the event subsystem to generate
averages, because the retrieval mode handles the Daylight Savings
Time changes. However, if the period for the average is hourly, then it
is recommended that you use the event subsystem, as the amount of
data will not generally not be a factor in the speed of calculating the
average.

External Detectors
For an external detector, event detection is triggered from an external
source by the ActiveEvent ActiveX control that is provided as part of
the Wonderware Historian. For example, an InTouch or Visual Basic
script can invoke the necessary ActiveEvent methods to trigger an
event. This ActiveX control must be installed on the computer from
which you want to trigger the external event.
For more information, see Configuring an External Detector in
Chapter 11, Configuring Events, in the Historian Database Reference.
Event Actions
An event may or may not be associated with an event action. An event
action is triggered after the event detector determines that the event
has occurred. The event subsystem is not intended to run external
processes. There is only a very limited ability to run external program
files or to call methods from COM interfaces within the given system
or network.
Actions are not required; there are times when you may want to
simply store when events happened. In this case, you would select
"None" for the action type when defining the event tag.
Generic SQL Actions

A generic SQL action executes an action that is outlined in a SQL
statement. For example, a SQL action can update the database (for
example, turning off storage for tags) or copy data to a separate table
or database.
You can use pre-configured SQL templates that are stored in the
database as the basis for your script, or you can create your own script
entirely from scratch. You cannot submit multiple queries against the
Wonderware Historian in a single event action and you cannot use GO
statements. Also, if you are querying against history data, the SQL
statement is subject to the syntax supported by the Wonderware
Historian OLE DB provider. You should test your SQL queries in SQL
Server Query Analyzer before using them in a generic SQL event
action.

Event Actions355
Snapshot Actions
A snapshot action logs into dedicated SQL Server tables the data
values for selected analog, discrete, or string tags that have the same
timestamp as the detected event. Quality is also logged. Value
snapshots are stored in tables according to the tag type, either
AnalogSnapshot, DiscreteSnapshot, or StringSnapshot.
A snapshot action requires an expensive SQL join between the
extension tables and the snapshot tag table. The process of performing
the join and logging the retrieved results to the snapshot tables can be
very slow. This is because most of the tables used for event snapshots
are normal SQL Server tables, subject to the data processing
limitations of Microsoft SQL Server. Thus, the higher the number of
snapshots that are being taken by the event system, the higher the
transaction load on the Microsoft SQL Server.
Important: The event subsystem is not a data acquisition system.

DO NOT attempt to use snapshot actions to move data stored in the
extension tables to normal SQL Server tables. This type of
misapplication is guaranteed to result in exceedingly poor throughput
and storage rates.
When trying to determine how many snapshots can be made by the

system, you should execute the intended snapshot queries to the
server using a batch file, leaving the event subsystem out of the
exercise. By executing repeated snapshot queries at the server as fast
as the computer will allow, you can better determine how many
snapshots can be performed on a system over a given time period.
Using this result and applying a safety factor may provide a good
guideline for assessing how much your system can safely handle. Keep
in mind that discrete snapshots are many times slower than analog
snapshots.
E-mail Actions
An e-mail action sends a pre-configured Microsoft Exchange e-mail
message. Although e-mail actions are useful for sending non-critical
messages triggered by an event detection, these types of actions are
not to be used for alarm-type functionality. For e-mail notifications of
alarm situations, use an alarm system such as the SCADAlarm alarm
notification software.

Deadband Actions
A deadband action changes the time and/or value storage deadband for
one or more tags that are using delta storage. (Value deadbands only
apply to analog tags.) Deadband change actions are useful for
increasing data storage based on an event occurring. For example, an
event detector has detected that a boiler has tripped, you might want
to start saving the values of certain tags at a higher rate to help you
determine the cause of the trip.
Event deadbands only apply to tags that use classic storage.
Summary Actions
A summary action is a set of aggregation calculations to be performed
on a set of tags between a start time and an end time with a defined
resolution. When you configure a summary action, you must define the
type of aggregation you want to perform (called a summary operation)
and the analog tags that you want to be summarized. The event
subsystem performs average, minimum, maximum and sum
calculations on the basis of a specific event being detected.
Note: Summary actions using the event subsystem are retained for
backward compatibility. We recommend that you use the more robust
and flexible replication subsystem to perform data summaries. For
more information, see Chapter 10, "Replication Subsystem."
Data summaries are useful for:

• Extremely long-term data storage. Because summarized data
takes up less space than full resolution data, even a moderately
sized system can store daily summary information for many years.
• Production reporting. For many reporting purposes, aggregate
data is more important than raw data. For example, the total mass
produced in a day is often more relevant than the actual rate of
production during the day.
• Integration with business systems. The full resolution,
high-performance Wonderware Historian history and real-time
data tables are best accessed with tools that can take advantage of
the Wonderware Historian time domain extensions. However, not
all client tools support these SQL extensions. The summary tables
reduce the volumes of data to manageable quantities that can be
used by any normal SQL client application.

Event Actions357
A summary action is usually triggered by a schedule detector.

However, you can perform a summary as a result of any event
detection.
Tag values with bad quality are not filtered out before the aggregation
is performed. To perform an aggregation with only good quality, for
example, use a generic SQL action that executes an aggregation
calculation query on the History table where the value of the Quality
column equals 0.
The results of all summaries are stored in the SummaryData table in
the Runtime database.
Important: Use caution when setting up summary actions. Using a

high resolution for your summary queries can have a negative impact
on the overall performance of the system.
Average, minimum, and maximum values can also be determined by

using the time-weighted average, minimum, and maximum retrieval
modes, respectively. For more information on these retrieval modes,
see "Understanding Retrieval Modes" on page 157. Keep the following
in mind when deciding to use either the event summaries or the
retrieval modes:
• For the time-weighted average retrieval mode, the amount of time
between the data values is a factor in the average, whereas the
event summary action is a straight statistical average. For more
information, see "Average Retrieval" on page 182.
• Performing an average at retrieval eliminates problems that occur
during Daylight Savings Time adjustments for schedule-based
summaries. For more information, see "Schedule Detectors" on
page 353.
For a comparison of all the different types of summaries that the
Wonderware Historian supports, see "Querying Aggregate Data in
Different Ways" on page 324.

Event Action Priorities

The event subsystem contains three different queues for event actions:
• A "critical" queue, which contains any actions for event tags that
have been assigned a critical priority. Actions for events that are
given a critical priority will be processed first. It is extremely
important that the critical queue is used with caution. Only
singularly important actions with short processing times should be
assigned as critical. You should never assign snapshot or summary
actions as critical. There is no overload protection for processing
critical actions; if the system becomes overloaded, actions may not
execute in a timely fashion or may not execute at all.
• A "normal" queue, which contains any actions for event tags that
have been assigned a normal priority. All non-critical events are
labeled with a "normal" priority and will be processed after the
critical events.
• A “delayed action” queue, which contains any actions for event tags
that have been assigned a post-detector delay. The post detector
delay is the minimum amount of time that must elapse after an
event was detected before the associated action can be executed.
Event Subsystem Resource Management

The Event System Service (aahEventSvc.exe) manages all of the
system resources required to detect events and process actions.
System resources are allocated for detectors and actions by means of
threads. A thread is an operating system component that
independently performs a particular function within a larger process.
Within the overall process of the event subsystem, event detectors and
actions are assigned different threads, so that they can execute
independently of each other and thus perform more efficiently.
The event subsystem uses two thread groups, or "pools." One thread
pool is for detectors and the other one is for actions. The Event Service
automatically creates both of these thread pools if there is at least one
event tag defined.
Other aspects of resource management include the number of
database connections required by event system components, and how
the system handles event overloads and query failures.

Event Subsystem Resource Management359
Detector Thread Pooling

The detector thread pool is made up of one or more threads allocated
for SQL-based detectors and a single thread for schedule detectors.
Each thread maintains a connection to the database. The detector
thread pool is illustrated in the following diagram:
'HWHFWRU7KUHDG3RRO
7LPH,QWHUYDO
'HWHFWRUV
VLQJOHWKUHDG
7LPH,QWHUYDO
'HWHFWRUV
VLQJOHWKUHDG
7LPH,QWHUYDO1
'HWHFWRUV
VLQJOHWKUHDG
6FKHGXOH
VLQJOHWKUHDG 'HWHFWRUV
A SQL-based detector is assigned to a thread based on the time

interval that you specify when you define the event tag. Each time
interval requires its own thread. For example, you define three event
detectors and assign them time intervals of 10, 15, and 20 seconds,
respectively. Each event detector will be running in its own thread, for
a total of three threads.
As another example, you define three event detectors, assigning the
first two a 10 second interval, and the third a 15 second interval. The
first two will be running under the same thread, while the third will be
running under its own thread, for a total of two threads.
For multiple detectors that are assigned to the same time interval, the
SQL detection statement for each event tag will be executed in
sequential order. That is, the first SQL statement must return results
before the next statement can be executed. After each detection has
taken place (results are returned), the detection is logged into the
EventHistory table and any associated action is queued into the action
thread pool.

All schedule detectors are assigned to a single thread.

The efficiency of the detector thread pool depends on how you have
spread the load when assigning time intervals to different event tags.
Detections generally do not cause overloading on the system: the
actions (especially snapshots and summaries) are where most
processing and resource loading occurs.
Action Thread Pooling

The action thread pool is essentially a pool of four threads that execute
actions from three different action queues. Each thread in the pool
maintains a database connection.
$FWLRQ7KUHDG3RRO
$FWLRQ3URFHVVRU
6LQJOH7KUHDG
&ULWLFDO4XHXH
$FWLRQ3URFHVVRU
6LQJOH7KUHDG
3RVW'HWHFWRU'HOD\4XHXH
$FWLRQ3URFHVVRU
6LQJOH7KUHDG
1RUPDO4XHXH
$FWLRQ3URFHVVRU
6LQJOH7KUHDG
The three action queues are:

• Critical queue
• Normal queue
• Post-detector delay queue
For detailed information about each of these queues, see "Event Action
Priorities" on page 358.
As a processor thread completes its previous task, a new action will be
fetched from one of the queues. If there are any actions in the critical
queue, these will be processed first. Actions in the critical queue are
executed in the order in which they were added to the queue; that is,
the oldest action sitting in the queue will be processed first.

Event Subsystem Resource Management361
If the critical queue is empty, actions will be fetched from the

post-detector delay queue. Actions in the post-detector delay queue are
ordered by time. Actions assigned the shortest post-detector delay will
be executed first.
If both the critical and post-detector delay queues are empty, actions
will be fetched from the normal queue. Like critical actions, normal
actions are processed in the order in which they were added to the
queue.
Event Subsystem Database Connections

The following table contains the number of SQL Server database
connections required by the different components of the event
subsystem.
Component Number of Connections Used
Event Service 1
SQL-based detectors 1 per each time interval used
Schedule detectors 1
Action threads 4
Handling of Event Overloads and Failed Queries

The event subsystem handles SQL-based detector and action queries
that fail, as well as to degrade gracefully if detector and action
overload conditions occur.
• Event query failures
If the query for a SQL-based detector fails, the query will
automatically be executed again. The detection window start time
will remain the same until the next detection is made.
For a failed SQL-based action query, the query will be submitted
three times. The system will establish a new connection to the
database each time the query executes. If the action query is a
snapshot query, the snapshot tables will first be "cleaned up" as
part of the re-query process.

• Detector overloads
A detector overload occurs when the system cannot process all of
the detectors in a timely manner. Detector overload is handled by
means of the detection window. This window is defined by the
difference between the current system time and the time of the last
detection. If the window grows larger than one hour, some
detections will be missed. This condition will be reported in the
error log.
• Action overloads
An action overload occurs when the system cannot process all of
the actions in a timely manner. Only actions assigned a normal
priority have overload protection. An action will not be loaded into
the normal queue by a detector if the earliest action currently
sitting in the queue has been there for an hour. (Basically, it is
assumed that the system has become so overloaded that it has not
had the resources to process a single action in the past hour.) This
prevents an accumulation of actions in the normal queue when the
system is unable to process them. The system will be allowed time
to recover, and actions will not start to be queued again until the
time difference between earliest and latest action in the queue is
less than 45 minutes (75 percent of the time limit). In short, when
the system becomes too overloaded, actions are not queued. This
condition is reported in the error log, but not for every single action
missed. The first one missed is reported, and thereafter, every
hundredth missed action will be logged.
There is no overload protection for critical actions, because these
types of actions should only be configured for a very small number
of critical events. There is also no overload protection for actions
that have been assigned a post-detector delay.
For more information on action priorities, see "Event Action Priorities"
on page 358. For more information on how actions are queued, see
"Action Thread Pooling" on page 360.

Event Subsystem Variables363
Event Subsystem Variables

The event subsystem uses a set of variables to facilitate event
detections and actions. The purpose of these variables is to provide
ease of query creation by a user (or a configuration editor). These
variables are replaced with the associated values by the event
components immediately before actual query execution. The query
actually being received by the Wonderware Historian never contains
the variables.
The variables and their associated values are as follows:
Variable Description/Associated Value
@EventTime Date/time of the detected event of the current

detector.
@EventTagName Tagname associated with the detected event.
@StartTime Start date/time for the detector query.
@EndTime End date/time for the detector query.
The @StartTime and @EndTime variables can be used only in detector

strings. The @EventTime and @EventTagName variables can be used
only in action strings.
All of the variables are case-sensitive.
Typically, a detection query executed by a detector component is
similar to the following example:
SELECT DateTime
FROM History
WHERE Tagname = 'BoilerPressure' AND Value > 75
AND DateTime > '@StartTime'
AND DateTime < '@EndTime'
@StartTime and @EndTime are simply placeholders for the detector

component to coordinate event detection over a moving time range.
The following action query show how event variables can be used:
SELECT * INTO TEMPTABLE
FROM History
WHERE DateTime = '@EventTime'
AND TagName IN
(SELECT TagName FROM SnapshotTag
WHERE EventTagName = '@EventTagName'
AND TagType = 1)
Note: These variables only function in the internal context of the

event subsystem and do not apply to queries from client tools such as
SQL Server Query Analyzer.


365
Index
Symbols service 43
< (less than) 296, 300, 302, 304, 308 sources of data 77
<= (less than or equal to) 296, 300, 301, 302, Transact-SQL 94
305 actions
> (greater than) 296, 298, 302, 304, 306, 308 See also event actions
>= (greater than or equal to) 296, 301, 302, active image 131
305, 306 about 130
automatic resizing 130
Numerics resizing 36
32-bit 114 retrieval 323
64-bit 114 storage 132
ActiveEvent 354
A ActiveX control 354
aaAdmin 30 aggregate functions 294
aadbo 30 aggregations 356
aaPower 30 AIAutoResize system parameter 36
aaUser 30 AIResizeInterval system parameter 36
acquisition alarm system 347
about 17 AllowOriginals system parameter 36
components 78 alternate storage location 116, 117
CSV data 99 analog specific value detector 351
I/O Servers 79 analog summary replication
IDASs 81 about 333
MDAS clients 98 analog tags
quality 65 about 22

366Index
edge detection 267, 269, 270 components 70

AnalogSummaryHistory table 139 configuration tables 69
AnalogSummaryTypeAbbreviation system CONVERT function 149, 289
parameter 36 COUNT(*) function 292
annotations counter retrieval mode 206
making 296 counter, retrieval modes 206
application name 80 criteria condition 289
ArchestrA CSV data
user account 27 acquisition 99
architecture 18 import path 37
Wonderware Historian 19 manual data 23
arithmetic functions 293 quality 66
authentication CTU counters 207
databases 28 cursors 316
SQL Server 28 cycle count 225, 301
AutoStart system parameter 36 cyclic retrieval
average (time-weighted), retrieval modes 182 using comparison operators 301, 305
averages 182 cyclic retrieval, about 158
cyclic storage 102
B about 113
best fit retrieval mode 177 cyclic, retrieval modes 158
best fit, retrieval modes 177
blocks D
See history blocks data acquisition
buffer storage location 116, 118 See acquisition
data blocks
C See history blocks
cache data files
IDAS 84 See history blocks, Runtime database
catalog 141 data quality
circular storage location 116 See also quality
about 117 data retrieval
client See retrieval
quality 66 data sources 136
client applications 16 data storage
support 18 See storage
client context 145 data stores 136
client/server 18 database
comparison operators 296, 301, 305 authorization 28
ConfigEditorVersion parameter 36 database authorization 31
configuration database connections
modification tracking 54 for events 361
service 42 DatabaseVersion system parameter 36
See also dynamic configuration DataImportPath system parameter 37
configuration data 16, 69 DataIndexPath parameter 37
Configuration Service 70 date functions 290
about 72 Daylight Savings Time 336
configuration subsystem DCOM 98

Index367
DDE 53, 79 deadband actions

about 53 about 356
deadband 104, 232, 236, 284 e-mail actions
swinging door 104 about 355
time 104 generic SQL actions
value 104 about 354
deadband actions 356 overloads 362
deadband override period 108 priorities 358
delayed action queue 358 queues 360
delta retrieval snapshot actions
comparison operators 296 about 355
querying wide tables 280 summary actions
delta retrieval, about 162 about 356
delta storage 102 thread pooling 360
about 103 event data 16
delta, retrieval modes 162 event detectors
detectors about 350
See event detectors
external detectors
discrete specific value detector 351 about 354
generic SQL detectors
discrete tags
about 351
about 22
overloads 362
edge detection 271, 273, 274, 275
schedule detectors
disk space
about 353
plant data 17
specific value detectors
document conventions 12
about 351
documentation set 11 SQL-based detectors
dynamic configuration about 351
about 72 thread pooling 359
committing changes 75 event history
effects 74 storage duration 37
storage 132 event subsystem
about 345
E Event System Service 43, 346
edge detection 266 about 358
analog tags 267, 269, 270 event tags
discrete tags 271, 273, 274, 275 about 350
e-mail actions 355 events
EndTime variable 363 about 345
engineering units 236 components 346
error count database connections 361
system tags 44 edge detection 266
error messages 41 features and benefits 348
categories 41 overloading 361
See also system messages performance factors 349
errors uses 347
IDAS 84 variables 363
event actions See also event actions, event detectors
about 354

368Index
EventStorageDuration system parameter 37 See also Management Console

EventTagName variable 363 Historian Data Acquistion service 43
EventTime variable 363 Historian DataAcquistion service 42, 43
extension tables 154, 224 Historian Event System service 43
about 139 Historian EventSystem service 42, 43
external detectors 354 Historian I/O Server
about 156
F service 43
failover Historian IOServer service 42, 43
I/O Servers 90 Historian Manual Storage service 42
IDAS 86 Historian OLE DB Provider
FastDDE 79 See OLE DB provider
FILETIME 136, 318 Historian Replication service 43
firewall Historian Retrieval service 43
IDAS 83 Historian SCM service 43
forced storage 102 Historian Storage service 42
about 103 Historian SystemDriver service 43
four-part naming convention 140 HistorianPartner parameter 37
full, retrieval modes 169 HistorianVersion parameter 37
historical data 16
G history blocks
generic SQL actions about 114
about 354 creating 116
generic SQL detectors 351 deleting 118
about 351 duration 37
GROUP BY clause 144, 292 notation 115
groups querying data 321
security 29 remote data source 136
retrieval 323
H storage locations 116
handle 52 history data
HCAL acquisition 78
about 94 modification tracking 55
acquisition 78 History table 139
clients inserting manual data 315
retrieval 137 querying 277
data adjustments 96 history version 239
retrieval 97 HistoryCacheSize parameter 37
security 97 HistoryDaysAlwaysCached parameter 37
store-and-forward 95 Holding database
HCAP about 72
about 94 HoursPerBlock system parameter 37
acquisition 78
Headroom system parameter 37 I
heterogeneous query 138 I/O Driver 81
Historian Client Access Point service 42 I/O Servers
Historian Configuration service 42 acquisition 17, 78, 79
Historian Console addressing 80

Index369
data quality 65 L
failover 90 late data 87
inserting original data 36 LateDataPathThreshold system
modification effects 75 parameter 38
redirecting to InTouch 91 latency 341
time synchronization 39, 91 event detections 352
IDASs leading edge detection 269, 270, 273, 275
about 81 LicenseRemoteIDASCount parameter 38
acquisition 78, 81 LicenseTagCount parameter 38
configuration information 81 licensing
data processing 82 remote IDASs 38
data transmission 83 tag count 38
error logging 84 LIKE clause 145
failover 86 linear interpolation 242
late data handling 87 linear scaling 137
licensing 38 linked server 140, 153
modification effects 75 Live table 139
performance 51 querying 278
security 83 log file
slow and intermittent networks 89 See system message logs
store-and-forward 84 logins
system tags 45 Windows 27
time synchronization 39, 40, 91 Wonderware Historian 30
importing Wonderware Historian services 33
Holding database 72
IN clause 145 M
indexing 43 Management Console
indexing service 133 security 33
INNER REMOTE JOIN 147, 283 ManualDataPathThreshold system
INSERT statements 94 parameter 38
InSQLIOS maximum retrieval mode 194
See Historian I/O Server maximum, retrieval modes 194
integral retrieval mode 200 MDAS
integral, retrieval modes 200 acquisition 78
intermittent network. 89 clients
interpolated, retrieval modes 171 acquisition 98
interpolation 38, 171 retrieval 137
interpolation type 241 time synchronization 93
InterpolationTypeInteger parameter 38 data quality 66
InterpolationTypeReal parameter 38 storage 126
InTouch memory
redirecting I/O Server data 91 management 133
InTouch History Importer 78 tag information 37
item name 80 Microsoft Query 152
Microsoft SQL Server
J integration 18
JOIN clause 146, 147, 283 security 28
minimum, retrieval modes 188

370Index
mixed mode 29 performance

modification tracking system tags 40, 51
about 53 Performance Logs and Alerts 51
configuration changes 54 permanent storage location 116, 118
historical data changes 55 permissions
turning on or off 38 security 31
ModLogTrackingStatus system parameter 38 post detector delay 358
monitoring priority
performance tags 40 event actions 358
system tags 45, 46 process data 16
processes
N Wonderware Historian 42
net time command 93 production data 16
network protocols 53, 79
IDAS 89
Q
O quality
Object Linking and Embedding for Databases about 61, 66
(OLE) 137 client-side 66
object permissions 31 data acquisition 65
ODBC 19 HCAL 96
old data retrieval rule 247
replication delay 342 viewing values for 62
OldDataSynchronizationDelay system quality detail codes 62
parameter 38 quality flags 64
OLE DB 137 quality rule 247
OLE DB provider parameter 38
about 137 QualityDetail
four-part query 140 about 61
limitations 144 QualityDetail bit layout 65
linking 153 QualityRule parameter 38
stored procedures 318 query
syntax supported 144 examples 277
OPC quality 247, 320 OLE DB provider syntax 140, 144
about 61 queued replication 340
OPENQUERY function 142 queues
OPENQUERY statement 146, 152, 320 event actions 358
OPENROWSET function 143
operations R
See summary operations rate of change 104
operators 296, 301, 305 real-time
OR clause 146 data storage 126
ORDER BY clause 144 real-time data 16
original data 57, 239 real-time data window
overloading 361, 362 about 127
late data 88
P parameter 38
page faults 52 swinging door 108, 128

Index371
RealTimeWindow parameter 38 value deadband 236

reconfiguration version of data 239
See dynamic configuration See also cyclic retrieval, delta retrieval
redirect I/O Servers 91 retrieval modes
redundancy average (time-weighted) 182
See failover best fit 177
relational databases 70 counter 206
remote data source 136 cyclic 158
remote table delta 162
See also extension tables full 169
replication integral 200
about 327 interpolated 171
analog summary 333 maximum 194
comparison to event system 343 minimum 188
continuous operation 342 slope 203
data 339 time-in-state 211
Daylight Savings Time 336 ValueState 211
delay for old data 342 RevisionLogPath system parameter 39
latency 341 roles
overflow protection 342 permissions 31
queued 340 Wonderware Historian 32
run-time operations 341 rollover value 206
security 343 RTU 87
state summary 334 Runtime database
streaming 339 about 71
tags 329 configuration data 69
replication groups
about 337 S
replication schedules samples 130
about 335 sampling interval 228
ReplicationConcurrentOperations system schedule detectors
parameter 38 about 353
ReplicationDefaultPrefix system schema 141
parameter 39 security 18
ReplicationTCPPort system parameter 39 about 26
resolution 16, 228, 318 database authorization 31
using comparison operators 305 HCAL 97
retrieval IDAS 83
about 135 Management Console 33
active image 132 permissions 31
deadbands 284 replication 343
edge detection 266 roles
features 136 Wonderware Historian defaults 32
HCAL 97 SQL Server authentication 28
quality rule 247 SQL Server security 28
service 43, 137 users
time deadband 232 Wonderware Historian default 32
time zone 246 Windows groups 29

372Index
Windows operating system 27 service 42

Wonderware Historian logins 30 system tags 45
Wonderware Historian services 27 Wonderware Historian 17
SELECT INTO statement 314 See also delta storage, cyclic storage
SELECT statements 147 See also history blocks, storage locations,
syntax 140 delta storage, cyclic storage
services storage by exception
security for 27 See also delta storage
SQL Server login 33 storage deadband 356
simple replication storage locations
about 331 about 116
SimpleReplicationNamingScheme system alternate 117
parameter 39 buffer 118
slope retrieval mode 203 circular 117
slope, retrieval modes 203 modification effects 74
slow network 89 path 116
snapshot actions 355 permanent 118
specific value detectors 351 storage process 42
SQL Server storage rate
authentication 28 cyclic storage 113
See Microsoft SQL Server store-and-forward
SQL Server logins as manual data 23
See logins HCAL 95
SQL statements IDASs 84
for actions 354 stored procedures
SQL templates 351 OLE DB queries 318
SQL-based detectors 351 string tags
time intervals 352 about 22
stairstep interpolation 241 SuiteLink 17, 79
StartTime variable 363 about 53
state summary replication SuiteLinkTimeSyncInterval system
about 334 parameter 39
statement permissions 31 summary actions
StateSummaryHistory table 139 about 356
StateSummaryTypeAbbreviation system summary data 16
parameter 39 summary history
statistical average 182 storage duration 40
storage summary operations 356
about 101 summary replication
active image 132 about 332
data conversions 114 summary tags
data versioning 129 about 22
dynamic configuration 132 SummaryCalculationTimeout system
forced storage 103 parameter 40
methods 102 SummaryReplicationNamingScheme system
processes 126 parameter 40
real-time data window 127 SummaryStorageDuration system
parameter 40
reserved values for classic storage 129

Index373
swinging door deadband 104 time synchronization

benefits 105 I/O Servers 91
examples 108 IDASs 91
options 107 MDAS clients 93
real-time data window 128 Wonderware Historian 40, 91
requirements 129 time zone 35, 246
SysPerfTags system parameter 40 time-in-state 255
system driver time-in-state retrieval mode 211
about 44 time-in-state, retrieval modes 211
acquisition 79 timestamp rule 244
service 43 parameter 40
System Management Console 70 timestamping 323
system messages TimeStampRule parameter 40
about 41 TimeSyncIODrivers system parameter 40
categories 41 TimeSyncMaster system parameter 40
system parameters time-weighted average retrieval mode 182,
about 36 200
modification effects 74 topics
system tags I/O Servers 80
about 44 modification effects 75
tracking
T See modification tracking
tags trailing edge detection 270, 274, 275
about 21 Transact-SQL 17, 94, 154, 224
allocating memory 37
conventions 283 U
information in memory 133 UPDATE statements 94
mixing in a query 288 user account
modification effects 74 See logins
replication configuration 341 users
sources of values 23 Wonderware Historian 32
types 22 UTC 35, 246
threads 52
event actions 360 V
event detector 359 value deadband 104, 236, 284
event system 358 event action 356
tiered historian option for swinging door 107
about 327 value deadband, retrieval 236
time deadband 104, 232, 284 ValueState retrieval mode 211
event action 356 ValueState, retrieval modes 211
time deadband, retrieval 232 variables 320
time domain extensions 17, 154, 224 events 363
time handling 35 variant type data 289
query resolution 318 version
synchronization 39 stored data 129
time interval version of data, retrieval 239
SQL-based detectors 352 versioned data 239
time series data 17 views

374Index
using 142 storage 17

VTQ 21 subsystems 19
time synchronization 40, 91
W users 32
WHERE clause 148 version 37
wide tables Wonderware Indexing service 42, 43
delta retrieval 280 wwAdmin login 30, 33
limitations 145 wwAdministrators role 33
querying 283 wwCycleCount column 154, 287
using variables 320 wwdbo user 30, 33
WideHistory table 139 wwEdgeDetection column 154
querying 279 about 266
Windows authentication 28 wwFilter column 154
Windows login wwInterpolationType column 154
See logins wwParameters column 154, 224
Windows operating system wwPower login 30, 33
security 27 wwPowerUsers role 33
Windows security groups 29 wwQualityRule column 154
Windows Server 2003 53, 79 wwResolution column 154, 287
Windows Server 2008 53 wwRetrievalMode column 154, 287
Windows Vista 53 wwStateCalc column 154
Wonderware Historian wwTimeDeadband column 154
architecture 19 wwTimeStampRule column 154
documentation set 11 wwTimeZone column 154
integration with Microsoft SQL Server 18 about 246
logins 30, 33 wwUser login 30, 33
processes 42 wwUsers role 33
roles 32 wwValueDeadband column 154
security 26, 27, 28 wwValueSelector column 154
starting 36 wwVersion column 154

WONDERWARE 2013 Historian Concepts PDF

Uploaded by

Copyright:

Available Formats

WONDERWARE 2013 Historian Concepts PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

WONDERWARE 2013 Historian Concepts PDF

Uploaded by

Copyright:

Available Formats

What is the purpose of the document?

What is the purpose of the document?

What are some of the main components of the Wonderware Historian solution?

What are some of the main components of the Wonderware Historian solution?

Wonderware

© 2002-2010, 2012-2013 by Invensys Systems, Inc. All rights reserved.

Invensys Systems, Inc.

Chapter 1 Introduction .............................................. 15

Chapter 2 System-Level Concepts and Functionality ...... 21

Wonderware Historian Concepts Guide

Time Handling ............................................................................... 35

Chapter 3 Configuration Subsystem ............................ 69

Wonderware Historian Concepts Guide

Chapter 4 Data Acquisition Subsystem ........................ 77

Chapter 5 Data Storage Subsystem........................... 101

Wonderware Historian Concepts Guide

Chapter 6 Classic Data Storage Subsystem ................ 125

Chapter 7 Data Retrieval Subsystem ......................... 135

Chapter 8 Data Retrieval Options .............................. 157

Wonderware Historian Concepts Guide

Integral Retrieval ...................................................................... 200

Chapter 9 Query Examples....................................... 277

Wonderware Historian Concepts Guide

Using the COUNT() Function ..................................................... 292

Chapter 10 Replication Subsystem ............................ 327

Wonderware Historian Concepts Guide

Continuous Operation ............................................................... 342

Chapter 11 Event Subsystem ................................... 345

Wonderware Historian Concepts Guide

Wonderware Historian Concepts Guide

This guide provides information about the general architecture of the

Wonderware Historian Documentation Set

• Wonderware Historian Concepts Guide (HistorianConcepts.pdf).

• Wonderware Historian Administration Guide

• Wonderware Historian Database Reference

Wonderware Historian Concepts Guide

• Wonderware Historian Glossary (HistorianGlossary.pdf). This

In addition, the Invensys License Manager Guide (License.pdf)

Convention Used for

Initial Capitals Paths and file names.

Wonderware Historian Concepts Guide

Wonderware Historian Concepts Guide

Wonderware Historian Concepts Guide

The Wonderware Historian, formerly known as IndustrialSQL

Wonderware Historian Concepts Guide

The Wonderware Historian Solution

Wonderware Historian Concepts Guide

Wonderware Historian Benefits

Wonderware Historian Concepts Guide

Integration with Microsoft SQL Server

Support for SQL Clients

Wonderware Historian Concepts Guide

• Clients developed specifically to access data from historian.

Wonderware Historian Subsystems

Wonderware Historian Concepts Guide

Wonderware Historian Concepts Guide

System-Level Concepts and

Some concepts and functionality apply across the entire Wonderware

Wonderware Historian Concepts Guide