Prepared for SG Heavy Industries

Example SQL Server Health Check Report

Prepared by
UpSearch, LLC Colleen Morrow
177 Front Street David Maxwell
Berea, OH 44017
Michael Fal
www.upsearch.com/sql/ Ben Miller
UpSearch, LLC © 2016
Proprietary and Confidential
This is an example report with recommendations.
Contents
Executive Summary....................................................................................................................................... 3
Current Environment Overview .................................................................................................................... 4
Server and Infrastructure .......................................................................................................................... 4
SQL Server Instance .................................................................................................................................. 5
Database Information ............................................................................................................................... 5
SQL Server Performance Overview ............................................................................................................... 6
Host Performance (CPU and Memory) ..................................................................................................... 6
Disk and I/O Subsystem Performance ...................................................................................................... 6
SQL Instance Activity and Performance .................................................................................................... 7
Query and Indexing Overview ...................................................................................................................... 8
Long Running and Expensive Queries ....................................................................................................... 8
Top Missing Indexes .................................................................................................................................. 8
Duplicate Indexes ...................................................................................................................................... 9
Analysis ....................................................................................................................................................... 10
Configuration .......................................................................................................................................... 10
Performance ........................................................................................................................................... 11
Disaster Recovery.................................................................................................................................... 14
Security ................................................................................................................................................... 15
Additional Concerns ................................................................................................................................ 16
Conclusions ................................................................................................................................................. 17
About the Authors ...................................................................................................................................... 18
About UpSearch .......................................................................................................................................... 22

UpSearch, LLC @ 2016 Proprietary and Confidential Page 2

Executive Summary

Scenario

SG Heavy Industries, LLC (“SG Heavy Industries” or “client”) engaged UpSearch, LLC (“UpSearch”) to
facilitate an assessment of SG Heavy Industries’ current database platform in order to identify
performance tuning and monitoring options.

Scope of Work

UpSearch provided SG Heavy Industries with Colleen Morrow, David Maxwell, Michael Fal and Ben Miller
to gather and analyze operating system and SQL Server configurations and performance metrics,
recommend performance changes based on SG Heavy Industries’ environment and industry best
practices, and subsequently present this report.

Test Strategy

UpSearch ran a series of PowerShell scripts between February 15 and 16, 2016 (assessment period), which
explored and gathered information from both the Windows operating system and from SQL Server on
server ENTERPRISE. In addition, UpSearch ran a set of queries against the SQL Server instance and against
the databases running on the server. UpSearch also collected performance counters for both the
operating system and SQL Server.

Conclusion

UpSearch provided SG Heavy Industries this SQL Server Health Check Report after evaluating current
configurations, performance metrics and existing maintenance procedures in order to better understand
current performance bottlenecks. The findings and recommended actions of this report are not only
intended to improve performance, but are also made with an eye towards simplifying and standardizing
SQL Server management, ensuring business continuity and managing security.

The assessment revealed several configuration and performance related findings. The rough order of
magnitude estimate for UpSearch’s time required to complete the ensuing recommendations is between
50 and 65 hours. See page 18 for time estimates by recommendation.

Next, UpSearch recommends;

1. SG Heavy Industries reviews UpSearch’s SQL Server Health Check report.

2. SG Heavy Industries and UpSearch reconvene to clarify findings, answer questions, consider short-
term configuration and performance issues, as well as future roadmap options.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 3

Current Environment Overview

The following overview represents ENTERPRISE current state as of the assessment period.

Server and Infrastructure

Machine Name ENTERPRISE

Scan Date (UTC) 2/15/2016
Manufacturer Dell
Operating System Microsoft Windows Server 2008 R2 Datacenter
Version 6.1.7601
Service Pack Service Pack 1
Install Date (UTC) 4/23/2014
Domain STARFLEET.com
Role Member Server
Physical Processors 1
Processor Cores 8
Logical Processors 16
Physical Memory (MB) 30720

Logical Disk Logical Disk Logical Disk Logical Disk Logical Disk Allocation
Caption Size (GB) Free Space (GB) Volume Name Unit Size (B)
C: 60.00 18.63 OS 4,096
D: 239.99 21.29 Data 65,536
E: 600.00 57.01 Backups 4,096
Z: 152.57 90.48 Temp Storage 4,096

UpSearch, LLC @ 2016 Proprietary and Confidential Page 4

SQL Server Instance

Instance Name ENTERPRISE

Scan Date (UTC) 2/15/2016
Install Date 4/10/2012
Startup Date 2/14/2016
Product Name Microsoft SQL Server 2008 R2
Product Edition Web Edition (64-bit)
Level SP2
Platform NT x64
Version 10.50.4339.0
Server Type Standalone
Clustered FALSE
Logical Processors 16
Total Server Memory (MB) 30720
SQL Server Memory In Use (MB) 26592

Database Information
Database Owner Compatibility Level Recovery Last Known DBCC Last Full Backup
Name Model Date
master sa SQL Server 2008 (100) Simple 02/14/2016 1:00:08 AM
model sa SQL Server 2008 (100) Full 02/14/2016 1:00:09 AM
msdb sa SQL Server 2008 (100) Simple 02/14/2016 1:00:11 AM
PICARD_db DEN_DATANOMIX\ckasten SQL Server 2005 (90) Full 07/27/2014 7:59:39 02/07/2016 5:09:49 AM
PM
tempdb sa SQL Server 2008 (100) Simple

Database Name Data File Log File Data File Size Log File Size Available Data Space (MB)
Count Count (MB) (MB)
master 1 1 4.00 0.75 1.06
model 1 1 2.25 2.00 0.94
msdb 1 1 35.06 31.69 3.69
PICARD_db 1 1 201,608.31 22,250.50 7,451.81
tempdb 8 1 32,768.00 1.50 32,391.81

UpSearch, LLC @ 2016 Proprietary and Confidential Page 5

SQL Server Performance Overview

Host Performance (CPU and Memory)

Host performance indicates no overall stress from a server level. Overall low CPU usage (average 8.58%)
indicates that the server is not under any CPU stress. Available memory averages at almost 2GB and does
not fall below 1GB, showing that the server itself is not being starved for memory and has available
resources for the operating system and other non-SQL Server functions. The low use of the paging file is
evidence that the applications are not having to swap memory to disk, which is a key indicator of
application memory pressure.

Metric MIN AVG MAX

Paging File - Percent Usage 0.04 0.08 0.10
Memory - Available MBytes 1314.00 1958.20 2094.00
Processor - Percent Processor Time 0.03 8.58 99.99

Disk and I/O Subsystem Performance

The disk subsystem’s latency is within Microsoft’s recommendations for operation. Typically, it is expected
that read operations should be less than 20 milliseconds and write operations be less than 10 milliseconds.
The maximum values for these metrics indicate periodic spikes. For volumes hosting database files, this
could be an indicator of specific queries that require tuning or a result of contention from the database
log and data files being on the same volume.

Metric MIN AVG MAX

C:\ Avg. Disk ms/Read 0.00 2.19 421.48
D:\(1) Avg. Disk ms/Read 0.00 1.88 302.22
D:\(2) Avg. Disk ms/Read 0.00 1.94 335.48
D:\(3) Avg. Disk ms/Read 0.00 1.79 308.60
D:\(4) Avg. Disk ms/Read 0.00 1.80 274.49
E:\ Avg. Disk ms/Read 0.00 0.88 55.47
Z:\ Avg. Disk ms/Read 0.00 0.89 20.18

Metric MIN AVG MAX

C:\ Avg. Disk ms/Write 0.88 2.12 120.63
D:\(1) Avg. Disk ms/Write 0.00 2.52 65.55
D:\(2) Avg. Disk ms/Write 0.00 1.63 441.16
D:\(3) Avg. Disk ms/Write 0.00 1.52 319.39
D:\(4) Avg. Disk ms/Write 0.00 1.33 264.08
E:\ Avg. Disk ms/Write 0.00 1.82 304.30
Z:\ Avg. Disk ms/Write 0.00 1.47 101.03

UpSearch, LLC @ 2016 Proprietary and Confidential Page 6

SQL Instance Activity and Performance

In general, SQL Server is operating as expected in relation to internal processing. Query Compilations and
Re-Compilations fall within expected ratios compared to the number of Batch Requests per second. The
Page Life Expectancy counter is and overall indication of SQL Server memory efficiency. It is good to
differentiate PLE for 7am-6pm on the server (business hours) versus the 24 hour average. In general, the
trend for PLE is to average 4496 during the day, but when the server enters its low use timeframe, the PLE
climbs steadily.

Metric MIN AVG MAX

Batch Requests/sec 0.07 1109.25 6425.05
User Connections 32.00 82.67 206.00
Buffer cache hit ratio 93.30 99.98 100.00
Page life expectancy 953 15315 39506
Page life expectancy (Business Hours) 953 4496 39506

SQL Re-Compilations/sec 0.00 0.62 33.40

SQL Compilations/sec 0.00 99.23 676.51

SQL Server operates on a "waits and queues" system. This means that when tasks within the engine must
"wait" for a resource to become available in order to complete, they are assigned a "queue" until that
resource becomes available. These waits are tracked and can indicate what resources within the SQL
Server instance are subject to performance issues.

77.40%

15.18%
2.73% 2.52% 2.16%

UpSearch, LLC @ 2016 Proprietary and Confidential Page 7

 Query and Indexing Overview

Long Running and Expensive Queries

SQL queries can perform poorly for a variety of reasons. Some factors are how they are written,
inadequate hardware, or missing indexes to support the workload. It is important to identify the queries
that are performing poorly based on:

 Number of Executions
 CPU cost
 Disk Impact (Reads and Writes)
 Duration

This is a listing of the most expensive queries currently running in the SG Heavy Industries environment,
based on these factors.

SQL Text Execs CPU Reads Writes Duration(ms)

{CURSOR} SELECT * FROM CUSTOMER_REG 295 2,223,620 2,130,748,982 2,222,518 2,454,634,124
{CURSOR} SELECT * FROM CUSTOMER_REG_EXTENSION 73 498,423 615,087,877 990,457 549,260,223
{CURSOR} SELECT PROCESS_OBJECT_HISTORY.PROCESS_EVENT_NAME 10 71,416 4,170,635 3 96,526,517
AS RAW_EVENT, PROCESS_OBJECT_HISTORY…

{CURSOR} SELECT SCRPT_DTA_OBJ_ID AS SDO, 1 13,182 290,561 9 30,535,488

MAX(PARENT_TRAN_DTA_OBJ_ID) AS MAXPDO FROM
TRAN_DTAOBJ_O…
{CURSOR} SELECT * FROM (SELECT DISTINCT ROW_NUMBER() OVER 4 1,218 243,913 100 22,423,025
(ORDER BY LNAME, FNAME ASC) AS ROWNUMBE…

{CURSOR} SELECT * FROM PARTY 3 12,682 9,916,748 10,640 21,673,880

INSERT INTO "PICARD_DB".."CUSTOMER_SIGNER" ("SIGNER_ID", 2 0 26 3 21,117,060
"SIGNER_CUSTOMER_ID", "SIGNER_FIRST_NAME"…

INSERT INTO EV_LOG (CASE_NUMBER, RECORD_TYPE, PARTY_ID, 1 0 11 2 20,347,660

LAST_NAME, FIRST_NAME, MIDDLE_INITIAL, MA…

{CURSOR} SELECT * FROM (SELECT DISTINCT ROW_NUMBER() OVER 1 1,357 184,378 99 15,742,859
(ORDER BY LNAME, FNAME ASC) AS ROWNUMBE…

UPDATE "PICARD_DB".."CUSTOMER_SIGNER" SET 2 0 28 3 12,778,330

"SIGNER_LINK"=@P1 WHERE "SIGNER_ID"=@P2 AND
"SIGNER_DOCU…

Top Missing Indexes

SQL Server tracks individual query statistics as they are executed. From these statistics, it compiles a list
of suggested indexes that could help query processing. Listed here is the top selection of these
suggestions. It should be noted that implementation of these indexes should be reviewed and tested for
impact, as SQL Server cannot take all factors of implementing these indexes into account.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 8

Database.Schema.Table Equality Columns Included Columns
[PICARD_db].[dbo].[REPORT] [account_rqst_id]
[PICARD_db].[dbo].[REPORT] [originating_party_id],
[archive_sts]
[PICARD_db].[dbo].[PROCESS_OBJECT_HISTORY] [PROCESS_EVENT_NAME]
[PICARD_db].[dbo].[REPORT] [originating_party_id]
[PICARD_db].[dbo].[REPORT] [originating_party_id],
[archive_sts]
[PICARD_db].[dbo].[PARTY] [PRTY_STS]
[PICARD_db].[dbo].[REPORT] [originating_party_id]
[PICARD_db].[dbo].[REPORT] [originating_party_id],
[archive_sts]
[PICARD_db].[dbo].[ACCOUNT_INFO] [ACCOUNT_RQST_ID]
[PICARD_db].[dbo].[REPORT] [tran_nbr], [reg_type_code]

Duplicate Indexes

There are times when an index is created with the same leading keys as another index. When indexes
have the same key columns in the same order, they are duplicates. Additionally, indexes that share the
same leading column order but do not share the same total columns are considered overlapping and could
possibly be consolidated. Listed here are indexes that can be considered partial duplicates.

Table Index Partial Duplicate

dbo.PARTY_EXTENSION PK__PARTY_EXTENSION__1DB2CC54 NonClusteredIndex-20130801-075027
dbo.WOTC_DATA PK__WOTC_DAT__57478927237692A2 NonClusteredIndex-20150612-081343
dbo.SCRPT_DTA_OBJ_VAR PK__SCRPT_DTA_OBJ_VA__6F2B50E7 ix_Scrpt_Dta_Obj_Var_DOC_ID
dbo.WOTC_CREDIT IX_WOTC_CREDIT IX_WOTC_CREDIT_1
dbo.WOTC_CREDIT IX_WOTC_CREDIT_2 IX_WOTC_CREDIT_3
dbo.PRTY_ASSN PK__PRTY_ASSN__58FC18A6 SBJCT_PRTY_ID
dbo.wotc_data_old PK_WOTC_DATA_new IX_WOTC_DATA
dbo.wotc_data_old IX_WOTC_DATA_2 IX_WOTC_DATA_3
dbo.CUSTOMER_REG RecommendedIndex1 RecommendedIndex2
dbo.VARIABLE PK__VARIABLE__5C63608D IX_VARIABLE
dbo.VARIABLE PK__VARIABLE__5C63608D ix_Variable_VAR_ID
dbo.VARIABLE IX_VARIABLE ix_Variable_VAR_ID
dbo.DASHCOLS PK_DASHCOLS HOW_WE_GRAB_ALL_FOR_VIEW

UpSearch, LLC @ 2016 Proprietary and Confidential Page 9

Analysis

The following findings are category specific and include detailed descriptions:

Configuration

Finding: Compatibility level 90 for PICARD_db

The PICARD_db is running at compatibility level 90, meaning that the database is using the rules and
features of SQL 2005. As this is not the current version of the SQL Server instance that is hosting the
database, it adds additional overhead to process queries due to the compatibility translation. Additionally,
it means your database can not take advantage of newer features available in SQL Server. It is a low risk
effort to upgrade the compatibility level, which will provide a fundamental support and performance
improvement.

 Action: Update compatibility level to 100 (2008 R2)

Finding: Filegrowth settings are set to percentage

The PICARD_db and tempdb databases have several files that are configured to grow by a percent. Use of
percentage growth settings for SQL Server can lead to unpredictable growth of the database files. There
can also be a performance impact, as SQL Server queries will wait when a file needs to be grown. It is
recommended that explicit file growth be set relevant to the size of the database.

 Action: Set explicit filegrowth settings for data and log files
 Action: Ensure TempDB filegrowth settings are the same for all data files.

Finding: Successful log backup messages in SQL Server error log

The SQL Server error log is a key resource for identifying and troubleshooting issues in the instance. To
make it easier to spot when problems are occurring, it should be kept free of non-critical messages as
much as possible. By default, every successful backup operation adds an entry in the SQL Server error log
and in the system event log. If you create very frequent log backups, these success messages accumulate
quickly, resulting in huge error logs in which finding other messages is problematic.

By enabling trace flag 3226, you can suppress these log entries. This is useful if you are running frequent
log backups and if none of your scripts depend on those entries.

 Action: Enable trace flag 3226 to suppress successful log backup messages from the SQL Server
error log.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 10

Performance

Finding: Data and Log files stored on the same volume

The PICARD_db data and transaction log files are housed on the same disk volume (D:). In a typical OLTP
environment the IO profile of a data file is composed of random IOs, split relatively evenly between read
and write operations. The IO of the transaction log file, on the other hand, is predominantly sequential
write operations. Housing data and transaction files on the same physical disk generally leads to IO
contention and a degradation in write performance to the transaction log file. Because of the critical role
the transaction log plays in overall database performance, best practices from Microsoft recommend
separating out this traffic between multiple volumes.

 Action: Create second volume to house log volumes

Finding: Inefficient storage configuration

Currently the D drive (PICARD_db data and transaction log files) is nonaligned with a starting offset of
32KB. Additionally, the Z drive (TempDB data and transaction log files) is formatted with a 4KB allocation
unit size.

All database LUNs should have the allocation unit size (partition cluster size) set to 64Kb (65,536 bytes)
for all partitions. Doing so will improve sequential write performance at a minimum and may reduce the
split I/O (inefficient I/O) count. If partitions are backed up, destroyed, and then recreated on a Windows
Server 2008 (or higher) system, the OS will automatically align the partitions to 1024KB. However, the
cluster size for the partition will still need to be manually set to 64Kb (65,536 bytes). Note: this step
requires new LUNs to be created in advance.

 Action: Move data files and tempdb to volumes that are properly aligned to 1024KB and
configured with a 64KB allocation unit size.

Finding: Transaction log has a high number of VLFs

The PICARD_db log is comprised of 892 Virtual Log Files (VLFs). When the transaction log grows (either
manually or via auto-growth), the new space allocated for the log is made up of virtual logs. When the log
grows frequently, a large number of virtual logs can exist, which can cause a performance issue during
startup, restores, and even inserts/updates/deletes. It is recommended to backup the log and then shrink
it to reduce the number of VLFs. Then, grow the file to the appropriate size in 8000MB increments. The
“appropriate size” for the transaction log will depend on the level of activity on the database and how
frequently the log is being backed up.

 Action: Perform a transaction log backup, shrink the transaction log file to 0, and manually grow
the file to the appropriate size. For log files larger than 8GB, grow the file in 8000MB
increments.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 11

Finding: No scheduled index or statistics maintenance

As data is inserted into tables and indexes, SQL Server creates additional pages to store this data in. These
new pages are not always efficiently stored for access, creating fragmentation in the database. Over time,
this fragmentation can have an impact on performance. Microsoft best practices are to run regular index
maintenance operations to rebuild or reorganize fragmented indexes. A maintenance job to reorganize
all indexes in the PICARD_db database exists, however it has not been executed since July of 2015.

There are 2 ways of defragmenting indexes in SQL Server, rebuilding and reorganizing. The index rebuild
process is highly IO intensive and generates a large amount of log activity. In addition, when an index is
rebuilt offline, as is the case in the SG Heavy Industries environment, it will also lock the underlying table
for the duration of the rebuild process. Performing this operation during peak periods can cause long-
term blocking and deadlocks, and be highly disruptive to transactional processing. However, rebuilding
an index is the most effective way to correct high-level fragmentation. Index reorganization is an online
and relatively low-impact operation which is appropriate for low levels of fragmentation.

Because of the potential overhead, determining which indexes are rebuilt and when should be carefully
considered. The level of fragmentation in an index should always dictate whether a rebuild is necessary.
Microsoft recommends that indexes with a fragmentation level > 5% and <= 30% be reorganized. Indexes
more than 30% fragmented should be rebuilt. Bear in mind that these values are general
recommendations. Depending on the size of the objects or the time available in your maintenance
window, you may wish to raise those thresholds in order to be more selective in the rebuilds being
performed. Note that smaller indexes (1000 pages or less) and indexes with very low levels of
fragmentation (<=5%) should not be addressed, as they generally do not see a benefit from rebuilding.

Count of Indexes
59

39

Nothing Reorganize Rebuild

In an environment such as SG Heavy Industries, we would recommend performing index maintenance on

a weekly basis, preferably during a less active period when the impact on transactional processing will be
minimized. For databases that are backed up weekly, with daily differential backups, the index
maintenance should be performed prior to the full backup, so as to minimize the size of subsequent
differentials.

However, it is the accuracy of statistics, rather than the level of index fragmentation, which has the biggest
impact on query performance. SQL Server relies heavily on statistics regarding the data distribution in a
table to form an optimal query plan. Outdated statistics lead to poor performance. To handle this, we
would recommend creating a job to update statistics on a daily basis. Updating statistics is a low-impact,

UpSearch, LLC @ 2016 Proprietary and Confidential Page 12

online operation that will not block transactional activity. Overhead can be further reduced by updating
statistics only where data has changed. For very volatile tables, it may be desirable to create additional
jobs to update statistics on those tables more frequently. Note that index statistics are updated when an
index is rebuilt (but not reorganized), so following an index rebuild with a statistics update is unnecessary.
However, it is still necessary to ensure that column statistics are updated, as well.

 Action: Create and schedule a weekly job to reorganize indexes with >5% and <=30%
fragmentation, and rebuild indexes with > 30% fragmentation. Limit the procedure to indexes
with more than 1000 pages (not based on row count).
 Action: Create and schedule a daily job to update modified index and column statistics with a
100% sample.

Finding: High CXPacket Wait Type

CXPACKET accounts for 77.4% of In general, CXPACKET indicates a heavy use of parallelism in queries
running on the SQL Server. This could be the result of several different factors, including (but not limited
to) improper parallelism configurations, inefficient queries, and a lack of indexing within the database.
There is no single answer to addressing this issue, so recommended actions will cover several options.

 Action: Set the instance level MAXDOP setting to 8

 Action: Set the instance level Cost Threshold for Parallelism setting to 65
 Action: Perform query tuning analysis on missing indexes and top poorly performing queries to
identify candidates for improvement.

Finding: Missing index on REPORT.account_rqst_id

Numerous high-read and high-duration queries captured during the assessment included a search
predicate on the account_rqst_id. Creating an index on this column will help to eliminate full scans on
the dashboard table or related indexes and facilitate seek operations instead, thereby dramatically
reducing the IO incurred by each execution of these queries. It is notable that this is also the most
recommended index based on SQL Server’s internal tracking mechanism.

 Action: Create a nonclustered index on REPORT.account_rqst_id

Finding: Use of “SELECT *” with no predicate

The three queries with the highest average and total read operations captured over an 8-hour period
during the assessment are shown in the table below.

Statement Executions Avg Avg Avg

Reads Writes Duration(ms)
SELECT * FROM ORDERS_TABLE 73 8,425,861 13,568 7,524

UpSearch, LLC @ 2016 Proprietary and Confidential Page 13

 SELECT * FROM REGISTRATIONS 295 7,222,878 7,534 8,320
SELECT * FROM JUNK_DATA 3 3,305,583 3,547 7,224

It appears that a cursor is being used to repeatedly read each of these tables, causing a large amount of
read IO. Wherever possible, it is recommended to select only the columns needed to satisfy the
applications needs, and to include a predicate to limit the number of rows returned, thereby minimizing
IO operations.

 Action: Review code to ensure only relevant data is being retrieved to avoid unnecessary IO.

Disaster Recovery

Finding: Integrity Checks currently not being performed

Database integrity checks (commonly referred to as CHECKDB), check the logical and physical structures
of a database for possible signs of corruption. The earlier corruption is detected, the greater the chance
of recovering from that corruption with minimal data loss. Therefore, it is critical to perform integrity
checks on all databases, including system databases, on a regular basis. The amount of I/O generated by
an integrity check can be extremely high (the entire database must be read into memory) and it is
recommended that these checks be performed during off hours, at least once a week.

 Action: Create a job or set of jobs to check database integrity on a daily or weekly basis.

Finding: Page Verify setting for PICARD_db is NONE

SQL Server uses several internal mechanisms to validate the data stored on disk. The Page Verify setting
defines how SQL Server will check for both logical and physical corruption within the data pages. When
set to NONE, this means SQL Server is not performing any of these checks. Changing this is low risk and
gives SQL Server the ability to detect and alert for corruption.

 Action: Change Page Verify for PICARD_db to CHECKSUM

Finding: Master, Model, and MSDB system databases stored on C:\

By storing the system databases on the OS volume, it is possible that SQL Server will halt if the OS drive
fills up. A secondary risk is any I/O operations done by the OS will also contend with these system
databases. It is a best practice to place the system databases on to another volume, either their own or
the volumes for the user database files.

 Action: Move system databases to a different volume

UpSearch, LLC @ 2016 Proprietary and Confidential Page 14

Finding: Default backup path targets the C:\

Having the default backup path targeting the OS volume introduces the risk that a backup could
accidentally be made to that volume. Database backups vary in size, but a large enough backup would fill
up the OS volume and potentially cause SQL Server to halt. It is recommended that the default backup
drive be set to the current backup volume for the instance.

 Action: Change default backup path to E:\Backups

Security

Finding: Logins and users with elevated privileges

At the time of the assessment every login on the ENTERPRISE instance was a member of the sysadmin
server role and the corresponding user was a member of the db_owner role in the PICARD_db database.
This means that every connection, including those from the web application, has the ability to perform
any operation in the SQL Server instance.

 Action: Review existing logins and users to determine if the level of access currently assigned is
appropriate.

Finding: BUILTIN\Administrators enabled

The BUILTIN\Administrators group is a login group that allows local Windows administrators on the host
server access. The group has sysadmin level access in SQL Server, meaning that anyone within the local
Windows Administrators group will have elevated SQL Server privileges. This is a concern because it makes
the separation of access and control more difficult. It is recommended that this group be removed and
sysadmin access be granted via another mechanism for better control.

 Action: Remove the BUILTIN\Administrators group and replace it with an Active Directory group
dedicated for SQL Server administrators.

Finding: PICARD_db is not owned by SA

Database ownership grants complete control to a database and the objects within it. Unless specifically
required, having a database owned by an individual login means that it cannot be secured completely
against the login that owns it. Best practice is to set database ownership to SA within SQL Server.

 Action: Change database ownership to SA

UpSearch, LLC @ 2016 Proprietary and Confidential Page 15

Finding: SQL Server Agent Jobs not owned by SA

SQL Agent Jobs can only be edited and properly managed by either the logins that own them or logins
with sysadmin level privileges. Because of this, it is generally recommended to set the ownership of Agent
Jobs to SA. This allows for more consistent job management.

 Action: Set SQL Agent ownership to SA

Finding: xp_cmdshell enabled

Extended stored procedures are a set of code within SQL Server that allows for it to interact with
components of the Windows environment outside of the database. The procedure xp_cmdshell allows
DOS commands to be run from SQL Server. The issue is that these commands have the potential to
execute as the SQL Server service account, which typically has elevated privileges in the OS. Because of
this, it is considered a security risk. It is recommended that, unless there is a specific need for this
functionality, it be disabled.

 Action: Disable xp_cmdshell

Additional Concerns

Finding: SQL Server 2008 R2 is not at the current service pack level

The SQL Server 2008 R2 installation is at Service Pack 2, Cumulative Update 13. Service Pack 3 is currently
available. Service Packs provide all the combined updates as well as additional functionality
improvements. In general, it is recommended to have the most recent Service Pack installed.

 Action: Upgrade installation to SQL Server 2008 R2 SP3

UpSearch, LLC @ 2016 Proprietary and Confidential Page 16

Conclusions

The performance metrics gathered during the assessment period indicate adequate processor and
memory resources to support the current workload. While occasional spikes were recorded, disk read
and write latency metrics, on average, were also within recommended parameters. It is believed that the
recommendations put forward in this report will further improve performance and concurrency within
the ENTERPRISE instance.

UpSearch provided the following rough order of magnitude time estimates to complete the
aforementioned recommendations. The preceding time estimates by recommendation are provided as a
courtesy and do not constitute a formal Statement of Work.

 Update SQL Server Configurations (backup locations, trace flags, parallelism settings)
Estimate: 1 hour

 Update Database Configurations (compatibility levels, database file growth settings, database
ownership, page verify settings)
Estimate: 2 hours

 Update SQL Server 2008 R2 to Service Pack 3

Estimate: 2 hours

 Reconfigure storage and relocate database files

Estimate: 5 hours

 Install, configure, and validate database maintenance scripts (index maintenance, integrity
checks).
Estimate: 10 hours

 Security review and remediation for elevated logins and configurations

Estimate: 10 hours

 Initial query tuning analysis and remediation

Estimate: 20 hours

UpSearch, LLC @ 2016 Proprietary and Confidential Page 17

About the Authors

Colleen Morrow is Principal Consultant at UpSearch.

Colleen is a database strategist, community advocate, author, blogger and public
speaker. She is passionate about helping technology leaders use Microsoft’s SQL
Server to protect, optimize and unlock data’s value.
Colleen has been working with relational databases for almost 20 years. Since
2000, Colleen has specialized in SQL Server and Oracle database management
solutions. She excels at performance tuning, troubleshooting mission critical SQL Server environments,
and training technology professionals and business users of all levels.
Since 2011, Colleen has maintained a SQL Server focused blog at http://colleenmorrow.com/. She is
Secretary of the Ohio North SQL Server User Group, as well as a volunteer for the Professional Association
for SQL Server (PASS).
Education & Certifications

 Microsoft Certified Solutions Expert: Data Platform (MCSE: Data Platform)

 Azure Technical Assessments: Data Analytics & Data Platform Solutions
 Bachelor of Science in Computer and Information Systems, Cleveland State University

Information Technology Skills

 Microsoft Azure  Microsoft SQL Server Reporting Services (SSRS)

 Microsoft SQL Server (through 2014)  Oracle (9i through 11g)

Publications

 What Changed? Auditing Solutions in SQL Server, Tribal SQL: New voices in SQL Server, Chapter
7 (http://www.amazon.com/Tribal-SQL-Tony-Davis-ebook/dp/B00H3JP4R0)

Speaking & Community Engagements

Colleen Morrow has presented at PASS Summit 2015, Enterprise Auditing with SQL Server Audit
(Enterprise Database Administration & Deployment), PASS Summit 2014, SQL Audit: from Introduction
to Automation (Enterprise Database Administration & Deployment), and at many SQL Saturday
conferences and user groups throughout the United States.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 18

 David Maxwell is a SQL Server Consultant at UpSearch.
David Maxwell is a database strategist, community evangelist and public
speaker. He is passionate about helping community and business leaders get the
most out of SQL Server.
Since 2000, David has served to protect, unlock and optimize data’s value within
such diverse environments as healthcare providers, insurance companies,
manufacturers and financial institutions. David has the unique ability to zero in
on a complex challenge quickly, and provide a long-term solution that fits both the business needs and
budget.
Since 2012, David has maintained a SQL Server focused blog at https://dmmaxwell.wordpress.com/. He
is a frequent presenter at local and regional events for the Professional Association for SQL Server (PASS)
and works with the Columbus, OH based PASS chapter.
In his free time, David is an avid musician who plays several instruments, as well as a lover of puzzle-based
games.
Education & Certifications

 Microsoft Certified Solutions Expert: Data Platform (MCSE: Data Platform)

 IEPTO1: Immersion Event on Performance Tuning and Optimization – Part 1
 IEPTO2: Immersion Event on Performance Tuning and Optimization – Part 2

Information Technology Skills

 Microsoft SQL Server (through 2014)  Systems architecture design,

 Microsoft SQL Server Reporting Services implementation, and administration
(SSRS)  TCP/IP Networking with DNS and DHC
 Microsoft SQL Server Integration
Services (SSIS)

Speaking & Community Engagements

Since 2012, David has presented sessions on SQL Server maintenance, performance and database
corruption at numerous SQL Saturday events throughout the Midwest, in addition to being a presenter
and director for his local Columbus, OH based PASS chapter. David has also presented for several online
events such as the PASS DBA Fundamentals and Performance Virtual Chapters, and 24 Hours of PASS. He
is also the winner of the PASS Summit 2015 Speaker Idol competition.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 19

 Michael Fal is a SQL Server Consultant at UpSearch.
Mike is a specialist in data management technologies. As a community advocate,
public speaker, and blogger, Mike is a practicing thought leader for data and
automation. He is passionate about DevOps and data, building platforms to
optimize, protect, and use data efficiently.
Since 1999, Mike has been working in the database field, focusing primarily on
SQL Server and specializes in automating data solutions to improve the reliability
and efficiency of his environments. Mike has the unique ability to create robust, repeatable solutions to
resolve a variety of technical challenges.
Mike has been working with databases for more than 15 years. He continues to blog at
http://mikefal.net since 2011, writing about a wide range of topics that center around data management.
He is an active member, and former board member, of the Denver SQL Server User Group as well as a
volunteer for the Professional Association for SQL Server (PASS).
Education & Certifications

 Microsoft Certified Solutions Expert: Data Platform (MCSE: Data Platform)

 Bachelor’s Degree, University of Colorado at Boulder

Information Technology Skills

 Microsoft Azure  High Availability & Disaster Recovery

 Amazon Web Services (AWS)  DevOps & Data Management
 Microsoft SQL Server (through 2014)  Performance Tuning
 Powershell  NoSQL / Non-Relational Data
 Database Development & Design Management

Publications

 DevOps and the DBA, Simple Talk by RedGate

 A Month of SQLPS Blog Series, Art Of The DBA
 Guest Posts – Hey Scripting Guy TechNet Blog

Speaking & Community Engagements

Mike has presented at PASS Summit 2015, Powershell and the Art of SQL Server Deployment, IT/Dev
Connections 2015, The Scalable SQL Server Enterprise, PASS Summit 2013, Monitoring Methodologies –
The Hierarchy of Database Needs, and at many SQL Saturday conferences and user groups throughout the
United States.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 20

 Ben Miller is a Principal Consultant at UpSearch.

Ben is a database strategist, community advocate, author and public speaker.

Ben’s passion to automate runs deep. He exemplifies how to work smarter and
automates whenever possible in order to get more done with less effort.
Ben is a Microsoft Certified Master (SQL Server) and Microsoft Data
Platform MVP. He holds numerous advanced industry certifications. Ben is also
an active member of Professional Association for SQL Server (PASS) and leads
the Utah County Chapter SQL Server User Group. He earned a Bachelor of Science in Information
Technology from Western Governors University.
Ben is a lifelong bowler and data professional. He has been bowling since he was 12 and has bowled a
sanctioned 300 game putting him in the Bowler’s Hall of Fame. He loves helping others get more out of
life through education and volunteers regularly in his local area.
Awards

 Microsoft Data Platform MVP (2011 – Present)

Education & Certifications

 Microsoft Certified Master (MCM) : SQL Server 2008

 Microsoft Certified Solutions Master (MCSM) : Charter Data Platform
 Microsoft Certified Database Administrator (MCDBA)
 Microsoft Certified Applications Developer (MCAD) : .Net 2.0
 Bachelor of Science, Information Technology, Western Governors University

Information Technology Skills

 Microsoft Azure  Data Warehouse & ETL Design

 Amazon Web Services (AWS)  Database Development & Design
 Microsoft SQL Server Engine Internals  High Availability & Disaster Recovery
 Microsoft SQL Server Encryption  DevOps & Data Management
 Microsoft SQL Server Partitioning  ETL Architecture
 Microsoft SQL Server Availability Groups  Performance Tuning
 Microsoft SQL Server Clustering  Powershell

Publications & Presentations

 SQL Server MVP Deep Dives, Vol. 2, Chapter 26 – SQL Server Filestream : To Blob or not to Blob
 PowerShell Deep Dives, Chapter 23 – SQL Server Provider in PowerShell
 SQL Server PowerShell Stairway, SQL Server Central

Speaking & Community Engagements

Ben has presented at many SQL Saturday conferences throughout the United States. He helped plan and
organize Salt Lake City SQL Saturday events in 2012 and 2014.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 21

About UpSearch

UpSearch began supporting Microsoft SQL Server in 2009. Microsoft subsequently

confirmed UpSearch as a Certified Gold Partner, its most highly accredited independent
support honor. Today, UpSearch’s Microsoft Practice provides full life-cycle SQL Server
and Azure support in the United States and Canada.
We offer the following services:
 SQL Server monitoring and companion DBA support, which bundles our own proprietary tools,
SQL Sentry’s Power Suite Plus and a predetermined number of dedicated monthly support hours.

 UpSearch’s Block of Hours Model provides support against a pre-negotiated block of hours. The
larger your annual commitment, the lower the bill rate. Our clients typically assemble twelve
months’ worth of projects and support to lock-in the best bill rate.

Our team will complement your team across a broad spectrum of database initiatives, such as:

 Migrations to Azure
 Health Checks and Baseline Assessments
 High Availability & Disaster Recovery
 Database Design and Development
 DevOps and Data Management
 Performance Analysis and Tuning
 Business Intelligence
 Conversions to SQL Server
 Server Consolidations / Upgrades
 Virtualization

On-premises or in the cloud, UpSearch can help you protect, unlock and optimize your data’s value.

UpSearch, LLC @ 2016 Proprietary and Confidential Page 22