Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.

com
Contact : Adam Patrick Oracular, Inc.
Data Warehousing With Oracle
By Muhammad Ahmad Shahzad
Abstract
With the emergence of data warehousing, Decision Support Systems have evolved to its best. At the core of these
warehousing systems lies a good database management system. Database server, used for data warehousing, is responsible
to provide robust data management, scalability, high performance query processing and integration with other servers.
Oracle being the initiator in warehousing servers, provides a wide range of features for facilitating data warehousing.
This paper is designed to review the features of data warehousing conceptualizing the concept of data warehousing and
lastly, features of Oracle servers for implementing a data warehouse.
Data Warehouse A Conceptual Overview
Definition of Data Warehouse
W.H. Inmon, father of data warehousing, defined data warehouse as: A data warehouse is a Subject Oriented,
Integrated, Non-volatile, and Time-variant collection of data in support of managements decisions.
With the advancement in the computing technology, the fall in the computer hardware and change in the nature of
business the value of information have raised dramatically. The need of making decisions on the basis of large amount
of data, which has the property of diversification along with the hugeness, have raised to a level not comparable to any
phase throughout the history of Information Technology. Supplementing was the betterment of server operating systems
and the explosion of Internets and Web based applications. The more organized Information database is the better is
the performance of the company. This indispensable requirement to store enormous amount of data lead to the Analytic
Systems which in turn gave birth to the idea of Data Warehousing.
Data warehousing is about molding data into information, and storing this information based on the subject rather than
application. As mentioned by W.H. Inmon, in one of his articles, the data warehouse environment is the foundation of
DSS Decision Support Systems.
Going back to the definition of data warehouse, the warehouse is a Subject Oriented, Integrated, Non-volatile, and
Time-variant collection of data.
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Figure 1
Subject-Oriented
In data warehousing the prime objective of storing data is to facilitate decision process of a company, and within any
company data naturally concentrates around subject areas. This leads to the gathering of information around these
subjects rather than around the applications or processes.
Integrated
Though the data in the data warehouses is scattered around different tables, databases or even servers but the data is
integrated consistently in the values of variables, naming conventions and physical data definitions.
Nonvolatile
Being the snapshot of operational data on a given specific time, the data in the data warehouses should not be changed or
updated once its loaded from operational system. As the snapshot shows operational data at some moment of time and
one expects data warehouse to reflect accurate values of that time frame. There exist only two operations the time-
based loading of data, accessing the loaded data.
Time-variant
The value of operational data changes on the basis of time. The time based archival of data from operational systems to
data warehouse, makes the value of data, in the data warehouses, being function of time. As data warehouse gives
accurate picture of operational data for some given time and the change in the data in warehouse is based on time based
change in operational data, data in the data warehouse is called time-variant.
From the operational systems to the requirement of DSS, to designing of data warehousing, to Implement to ongoing
support, data warehousing does not use some alien concepts and is more or less based on the typical System
Development Life Cycle (SDLC) concept.
Data warehouses possess a degree of multi-dimensioning in there nature. The advocates of Relational Modeling say that
Multi-dimensioning of data is just another way of representation of data in two dimensional relational models. If we
agree to the above rationale then the data warehousing comes in the umbrella of traditional RDBMS application
development process. Yet indeed, there are some major differences when building a warehouse, including features like
hugeness of data or accessibility or providing dynamic access etc. The most important difference is of course the way data
is placed in data warehouses, its more like summarized, referenced, de-normalized representation. In short what ever or
how ever we develop a data warehouse it should at least be capable of providing ad hoc complex, statistical, and analytical
queries to facilitate decision making process.
Data Warehouse
Collection of Data
Subject Oriented,
Integrated,
Non-volatile,
and Time varian
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Architecture of data warehouse
As repeatedly mentioned in this paper, the prime concern of providing a separate set of data the data warehouse, is to
facilitate Business Analysts in the process of Decision Making. Essentially data warehousing is the warehousing data
outside operational systems and this has not significantly changed with the evolution of data warehousing systems. Prime
reason of this separation is that the evaluation and analysis, done by analysts, require complex and analytic queries - the
effect of which is the performance degradation of operational systems. Another important feature is the combination of
data from more than one operational system to provide the ability of cross-referencing.
Figure 2
Most of the data warehousing done, posses three-tier architecture.
The base level from which data is extracted is operational system (OLTP) and the legacy systems, from which data is
transformed and loaded into the warehouse database. So the middle level is the data warehouse and the top most level is
the analytic system (OLAP) and Decision Support System (DSS). OLAP systems utilize the data warehouse to provide
multi-dimensional view. Functionally a data warehouse can be divided into following:
Data Extraction
Transformation and Scrubbing
Storing and Cataloging
Data Access
Data Delivery
All of the above functions are self-explanatory. The process starts with the extraction of data from operational system and
legacy systems, then comes the transformation and cleaning of data during this process summarization and aggregation is
also done. Data Storage represents the process of storing transformed and cleaned data in a relational database. Data
access holds the query processing, multi-dimensional analysis and data mining. Lastly comes the function of data delivery
to the end-users, which may be the part of data warehouse or can come under the umbrella of OLAPs.
A data warehouse, being unique in the class of applications, possesses a structure, which is different from other database
applications. Being used for analytic purpose it is designed in a way so that it can facilitate complex queries. Mostly the
business analysts focus on the summarized data, time variant data. So the data warehouses are designed to facilitate the
above process. Data warehouses hold different levels of summarization and details. It also has two groups of detail data,
the current detail data and older detail data.
Legacy
System
Operational
System
OLAP
DSS
OLAP
Data
Extraction
Data
Transform
Data
Scrubbing
Data
Access
Data
Delivery
Data Storing
Data
Cataloging
Operational
System
Data Warehouse
A
N
A
L
Y
S
T
S
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Current detail data, reflecting the most current happening in the organization, is highly voluminous and is always stored
on disk storage. It may reach space as much as gigabytes or even terabytes. The reason of being so sizable is that it asserts
lowest level of granularity.
Old detail data, as the name shows, is the data, which is not that frequently, used. Due to the infrequent requirement
this data is stored on some cheaper storage mediums like tape cartridge.
Then comes the level of summarization, the difference between lightly summarized and highly summarized is quite
obvious. Lightly summarized data is the summary of detailed granulized data. Whereas highly summarized data is more
compact than summarized and is based on lightly summarized figures. Both of these reside on disk media, as these are
accessed very frequently.
Meta data, being very important data repository, resides on different dimension than other data classes. As it may be
accessed by any of the other layers and work as a linkage warehouse and operational environment.
Warehouse Database Server Its Role in Data Warehousing
Data in the data warehouse database is organized by subject rather than applications or processes, and this data is
extracted and refreshed from operational system on a periodic basis. We have already discussed the three-tiered
architecture in which first tier is the operational system, middle is the data warehouse database server and last one is
front-ended client applications, including DSS and OLAP applications. In three-tiered architecture the warehouse
database server works as the heart of warehouse application. Though a simpler form of data warehouse applications exist,
in which the architecture is two tiered tier1 includes the Operational System as well as Warehouse database and tier2 is
the client front-end Decision Support applications.
One cant disagree to the fact that Database servers are at the core of every application that supports business decisions,
specially data warehouses providing robust data management and scalable, high-performance query processing.
Figure 3
Current Detail
Data
Older Detail
Data
Lightly Summarized
Data
Highly Summarized
Data
Meta Data
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Warehouse servers are categorized in two types, RDBMS (Relational Database) and MDD (Multi-dimensional
Database) the choice is based on type of data stored in warehouse.
RDBMS is based on the concept of mathematical relation operation. The implementation of RDBMS is based on two-
dimensional relationship of related data called the tables. Whereas, MDD can be viewed as cube, where information is
pilled on various axes of cube. Taking as an example, the case of Sales production of a company Sales are related to
salespersons, the geographical region, and some time frame, this result in three-dimensional view of data. The cross-
section of these three can give the required data. However MDD just work with finite set of data and information which
is highly related to each other.
Relational database technology has an edge on MDD, when we are considering huge data storage capacity or portability
issue or security. RDBMS is an old and proven technology in data storage and recovery. MDD is popular for its Instance
Response, Implementation ease, and integration with Meta-data. Either we choose MDD or RDBMS in both cases a
database server has a very central role in the data warehouse architecture.
Data modeling Star Schema as choice
Data modeling, the process of making data models is not unique for warehousing; in fact we use this tool in the
development of all kinds of database applications. The reason in data warehousing is pretty much the same.
Defining the scope of data warehouse
Viewing the complexity of the relationship between data
Recognizing and controlling redundancy
Decision-makers during the analysis generally formulate complex queries, which are based on multiple dimensions. As
data warehousing is done to facilitate this type of multi-dimensional queries for decision making, the modeling of data
also tends to bear multiple dimensions. Data warehousing done using relational database technology generally holds
modeling in star schema.
Star schema is the implementation of multiple dimensions in the relational modeling. This schema addresses data
navigation difficulty and its dimensions are the categories by which analysts organize the information. Star schema at the
lowest level is the relationship between tables, but with the expansion of scope of warehouse the model tends to become
more and more complex. So its a good practice to aggregate data into levels of hierarchy. The relationship among
different objects is provided by introducing fact tables tables having primary key compounded by primary keys of all
dimensions.
The fact table is the central table in star architecture containing the data links or points to establish dimensions in
different entities. Technically, this table is just intersection of entity primary keys.
Fact
Table
Dim 1 Dim 3
Dim 4 Dim 2
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Meta Data the data about data
Meta data provides data repository. Providing both technical and business view of data stored in the data warehouse. It
lays out the physical structures that includes:
Data elements and their types
Business definition for the data elements
How to update data and on which frequency
Different data elements having same meaning
Valid values for each data elements
Meta data plays very important rule in the definition, building, management and maintenance of data warehouses. In a
data warehouse Meta data are categorized into Business and Technical Meta data. Business Meta data describes whats in
the warehouse, its meaning in business terms. The business Meta data lies above technical Meta data, adding some
more details to the extracted material. This type of Meta data is important as it facilitates business users and increases the
accessibility. In contrast, technical Meta data describes the data elements as they exist in the warehouse. This type of
Meta data is used for data modeling, initially, and once the warehouse is erected this Meta data is frequently used by
warehouse administrator and software tools.
Using the Internet Technology
With the technology provided through Internets, the transfer of information has become very easy. On the other hand is
the requirement of accessing data warehouse globally. Putting both together gives a very effective solution to give the
access to data warehouses on the global scale. To provide global access to a data warehouse using web is like giving easy
access to data on the whenever, whoever basis. However, there are some issues which has to be sorted before the effective
utilization of internet technology, like the usage of web server along with the database server, security issues, and some
issues like providing ways for query and report purpose.
Oracle a choice for Implementing Data Warehouse
Since the conceptualization of data warehouse, many database venders have tried to mold their database systems for
accommodating it. Amongst which was Oracle that systematically evolved to address specific needs of warehousing.
When considering a data warehouse implemented in a RDBMS, there are some technology requirements like query
processing, data storage, scalability, integration with other systems and lastly the security management.
Query Processing
Queries in a data warehouse generally involve very large amount of data. Also its not rare to find complex operations like
multi-table joins, sorting and aggregation in data warehouse queries. These operations are generally set-oriented;
operating on some groups of records based on specified criteria. Most of the queries in decision making process are
multi-dimensional in nature, based on star schema. Another important feature in query processing of data warehouse is
that queries are not pre-defined and are based on the business-users runtime criteria.
Features like query optimization, access and joining methods and parallel execution of queries are very vital for
performance of data warehouse.
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Data Management and Scalability
This is the way data is loaded, organized, stored, accessed and maintained in a database. The database operations such as
data loading, enforcing constraints, building indexes, collecting statistics on the data, reorganizing tables and indexes,
building aggregates or summaries, and data purging are included in data management. Its not unusual to find very large
databases when implementing a data warehouse, also the growth of a warehouse is in big data leaps. The database
operations, listed above, are functions of database size.
To effectively meet the needs of data warehousing, the database server has to provide capacity to deal with large data
volumes and data operations should also be tuned for the same reason.
While the scope of data warehouse is not at all limited, this feature leads to the scalability of both users and data. With
the globalization of organizations number of end-users, requiring to use warehouse, have increased dramatically. The
supporting of this population of users is the responsibility of database server. This include supporting wide range of
hardware, operating systems, and clients trying to access data warehouse from widely apart physical positioning.
When considering scale of data, server has to support data volumes of gigabytes, terabytes or even beyond. The scalability
doesnt merely mean the capacity to store immense data; it encompasses the ability to efficiently process queries, the
capability to perform data management operations, and delivering business-critical availability, all at huge scale.
Integration with other systems
In the process of decision making, the analysts have to access data even beyond the boundaries of operational data and its
not always wise to transfer each bit of data from systems like this to data warehouse. So database servers should provide
provisions to link the warehouse application to systems like SAP, BAAN or PeopleSoft.
Security Management
With the physical size of data warehouses and number of users requiring to access data warehouse in the process of
decision making the security of organizations critical data is at stake if database server is not able to manage security
properly.
Oracle Server Where does it stand
Oracle has been amongst the earlier database management systems extending its features to accommodate data
warehouse related features. It was the era of Oracle7 when the concept of data warehouse came and Oracle-corporation
right away recognized its importance. Oracle v7.3 provided features like parallel query execution, parallel data
management, cost-based query optimization, efficient bitmap indexing and hash joining embedded in query execution.
Then came Oracle8 enhancing the features, already provided by Oracle v7.3. Linking the server with tools like Oracle
Discoverer and Oracle express have made Oracle the must viable option for data warehousing. Below we are going to
discuss features provided in Oracle to enhance the server capabilities for the implementation of data warehouses.
Query Processing
Oracle7 advanced its architecture to improve the Query Optimizer as well as the execution of query.
QUERY OPTIMIZATION: The main task of query optimizer is to choose the most efficient way to execute a SQL
statement the DML (Data Manipulation Language) are considered for optimization. Oracle produces an execution
plan for the optimization purpose. Oracle Optimizer takes following steps for the selection of best execution plan:
Evaluation of query expression and modification as per required. The optimizer assesses expressions construct and
whenever required introduces some modification to enhance the speed and reduce resource utilization. Some examples of
which are given in Figure 5.
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Figure 5
(Bold shows the
optimized query)
Transformation of complex and symbiotic queries into equivalent joins statements. In the process of transformation,
optimizer modifies two types of queries; queries containing OR to UNION ALL and complex queries into join
statements.
For queries having views the optimizer merges the query statement with that of view. Examples of such optimization is
given in Figure 6
Figure 6
Selection of Optimization approach from Rule-Based Optimization and Cost-Based Optimization. Rule-based approach
chooses execution path based on heuristically ranked operations. When more than one execution paths exists, rule-based
approach selects path with lower rank. Cost-based approach optimizes a query based on following steps:
Firstly, all potential execution plans are predetermined by optimizer plans are based on access paths.
Then, optimizer estimates the cost of each execution plan based on the data distribution and storage
characteristic statistics the statistics are based on table structure, indexes and clusters, I/O and CPU time, the
available memory .
Lastly, optimizer compares the cost of execution plans and selects one with lowest cost.
{Where Clause}
cl_name like xyz
cl_name = xyz
cl_name in (a, b, c)
cl_name=a or cl_name=b or
cl_name=c
cl_name > any (select amount
from payment
where place = xxx)
exists (select amount
from payment
where place = xxx
and cl_name > amount)
cl_name > all (select amount
from payment
where place = xxx)
create or replace view view1
as
select cl_1, cl_2
from table1
where cl1 > 10;
{when selecting from view1)
select cl2
from view1
where cl2 > 15;
{optimizer modifies the query into}
select cl2
from table1
where cl1 > 10
and cl2 > 15;
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
The selections of appropriate access path when a query is based on more than one table. Generally there exists more than
one access paths when the table-data is accessed. The optimizer chooses the most appropriate access path based on the
Rule-based or Cost-based approach.
Figure 7
When joining more than two schemas, optimizer decides which pair to join first.
QUERY EXECUTION: Introducing the intra-query parallelism via Parallel Query option, Oracle7 provided parallel
execution of complex queries having SQL operations like; SELECT, sub-queries in INSERT, DELETE OR UPDATE,
CREATE TABLE based on sub-query, and CREATE INDEX commands. The parallel Query option improves the
performance of data manipulation operations in very large databases, like warehouses. Best performance can be viewed
on SMP (Symmetric Multiprocessor) and MPP (Massively Parallel Processing) machines. The query writers have to
implicitly command the parallel query option and also declare degree of parallelism. Figure 8 explain how parallel query
works in Oracle.
Figure 8
(Parallelism of degree 3)
Select cl_1, cl_2, cl_3
From table1
Where cl_4 in (select cl_5
From table1, table2
Where table1.pk=table2.pk)
{Access path of above query is as follows}
Index
Primary
key
(table1)
Filtering
the
selection
Index
Primary
key
(table2)
Join
Operation
(table1 &
table2)
Access
table
(table1)
Access
table
(table2)
Select *
From table1
Table1
{query with parallel query
option}
Select *
From table1
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
In order to provide parallelism in query execution many initial parameters have to be configured. Once the system is
configured to run queries with parallel query option, it is the task of Query Coordinator Process to initiate parallel query
servers and coordinate between the results from these query servers. The number of query servers, running in parallel to
complete one operation, is called Degree of Parallelism.
Data Management and Scalability
Oracle provides a data management architecture in which physical database structure is encapsulated by logical structure.
By hiding the physical structure of database Oracle provides a level of manipulation through which one can introduce
modification in the physical structure without hindering the usual workload on the DBMS. Logically database is based
on Table Spaces users access the table space to access data. Table Spaces are made of different Data Files which
physically resides on Fixed Storage. To support very large space, specially for DSS systems, data base administrator can
introduce very large table spaces creating multiple data files for each table space.
Oracle7 introduced parallelism in data management operations like data loading, indexing and creation of summary
tables. V7.3 also incorporated bit-mapped indexing as integrated server capability, adding to the already available
indexing schemes like B-trees, clustered tables and hash clusters.
Oracle7 reputed itself in reliably managing wide range of users. Special features like replication of database and
partitioning on distributed systems have been introduced in Oracle to support the scalability of users.
Database Security
With the provision of being multi-user database system, Oracle provides sound controls for the security of database. The
controls include unauthorized access to database as well as individual schema objects, assessment of environment
parameters such as disk usage and system resource usage. Oracle provides a set of privileges and a user is restricted by the
grants given by these privileges. For effective management of privileges there exists roles grouping privileges together
and given some unique name. For most effective security management Oracle provides Trusted Oracle. Trusted Oracle
provides a multi-level secure database management and mandatory access control (MAC). To monitor the user actions
on database Oracle endure auditing of users.
Oracle8 Object Relational Data Server
Most of the features described in the previous section are related to older versions of Oracle, with the parturition of
Oracle8 the Object-relational database, Oracle took a major step from the world of relational database technology to
object-relational technology. In addition to other enhancements Oracle8 has advanced in some key areas related to data
warehousing. Here we are just going to cling with the server enhancements for data warehousing.
Scalable Query Processing
Typically in data warehouse, queries are the most critical operations. No decision-maker can be expected to decide
without the help of complex data retrieval. data warehouses, in deed, exist to support these decision-makers.
QUERY OPTIMIZATION: Based on the strong grounds provided by Oracle7, Oracle8 improved both - the query
optimizer and query executor. The Oracle8 query optimizer mainly uses Cost-based approach; in addition it is also aware
of parallelism and partitioning in the data. During query transformation, the optimizer rewrites the queries based on the
techniques provided in Oracle7 optimizer enhancing it by introduction of sophisticated cost-based query rewrites such
as the Star Query Transformation and Anti-joins as well as Semi-joins.
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Generally data warehouse design is based on star schema, which is characterized by one or more very large fact tables that
contain the primary information of dimension tables, each of which contains information about the entries for a
particular attribute in the fact table. The star query is based on the join between the fact table and dimension tables.
These queries are based on star join, which is primary-key to foreign-key join of the dimension tables to a fact table. To
have most effective performance the fact table is indexed using bit-mapped indexing. Oracle8, while executing star
queries; firstly, retrieves filtered data from fact tables fact tables are indexed using bit-mapped indexing, secondly, joins
the fact table with dimension tables and retrieve the result. An example is shown below:
Figure 9
QUERY EXECUTION: Oracle8 advancing in its best-of-both-worlds parallel query architecture making use of the
SMP as well as MPP architecture, introduced data partitioning. Oracle8s parallel query execution runs on two levels;
parallel execution across partitions and parallel execution within partitions. The introduction of parallelism within data
partitions have made Oracle8 a unique database management server as without this feature majority of database
management servers were unable to utilize parallelism properly.
Data Management
Oracle8 improved the parallelism feature and indexing of tables. Oracle8 introduced partitioning of very large table as
well as large indexes. In a non-partitioned index, a parallel index scan is done when a full table scan is required. Whereas
in paralleled access on partitioned-indexes, one query slave is assigned to scan each partition of an index then the result of
each of those are gathered together. It is the responsibility of DBA to set partitioned tables the CREATE TABLE
command have the provision for setting partitions based on partition key and a range for that key. Example of
partitioning is given in Figure 11.
Oracle8 brought up support for bulk insert, update and delete operations in parallel. These data management operations
in parallel provide efficient utilization of hardware resources.
Select *
From fact, dim1, dim2, dim3
Where fact.dim1pk = dim1.pk
And fact.dim2pk = dim2.pk
And fact.dim3pk = dim3.pk
And dim1.key1 = 1000
And dim2.key1 in (A, B, C)
And dim3.key1 = XXX
{Star Query executor will run it as
following query}
select * from fact
where fact.dim1pk in (select pk
from dim1
where key1=1000)
and fact.dim2pk in (select pk
from dim2
where key1 in (A,
B, C))
and fact.dim3pk in (select pk
from dim3
where key1=XXX)
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Figure 10
Figure 11
{TABLE1 partitioned into 3 partitions}
{Inter Patition Parallelism}
Select * from
Table1
{Intra-Partition parallilism}
Table1
5K
rows
Table1
95K
rows
Table1
400K
rows
Select *
from Table1
{Due to Intra-partitioned parallelism query
can access partition 3 in parallel}
Table1
5
Table1
5K
rows
Table1
95K
rows
Table1
400K
rows
Select *
from
Table1
{Partitioning on Purchase-Date based on
Quart1, quart2, quart3 & quart4}
Purchase
1M rows
Purchase Table
Partitioned in
Q1, Q2,
Q3& Q4.
Pur
Pur Pur
Pur
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Oracle8, for supporting large objects, introduced enhanced functionality to support new data-types:
BLOB for binary large objects
CLOB for character large objects
NCLOB for character large objects stored in the national character; analogous to NCHAR
BFILE for binary files stored outside of the Oracles universal data server
There exists functionality to store large objects within as well as outside database BFILE is used as a locator within the
database table pointing to OS file.
Scalability
Oracle8 provides database replication and partitioning along with management tools like Oracle Enterprise Manager and
Oracle Names to support scalability. All the above features enhance Oracles capability to manage very huge amount of
clients and data. Database replication across distributed environment allows the provision of enormous amount of users
accessing the database. The provision of partitioning ensures the database size to grow beyond terabyte boundary. Lastly
Enterprise Manager and Oracle Names enhances the management and configuration of thousands of users.
To support network-computing Oracle8 database server has been re-designed. The key strengths of Oracle8 are:
Ability to support tons of thousands of users both entry level and business-critical users
Ability to support huge amount of data
Fastest data server in industry
New features have been introduced in server to enable data availability even in the cases of partial failures and
maintenance. Oracle8 also managed to introduce improvements for Internet Users Internet applications can now
directly communicate with the server.
Database Security
With the support of tons of thousands of users connecting the server as direct users, client/server users or Internet
users, Oracle8 have to enhance its security measures to secure the integrity of data. The power of Oracle8s security
measures are more prominent when considering highly networked environment, where most of the data is traveling
through the network wires.
Oracle8 provides Discretionary Access Control (DAC), regulating all users access to database objects based on privileges.
With the usage of Trusted Oracle B1-level data security can be guaranteed based on Mandatory Access Control (MAC).
Integration with other systems
To entertain the needs of linking the Oracle server (v7.3 or v8) Oracle has provided tools like Oracle Open Gateway and
Oracle Transport Gateway.
Oracle8i The Internet Database
With the advancement of Internet applications and support given by Internet to networked environment more and more
organizations are planning to use this technology in the production of internet/intranet applications. Special
consideration to this environment is given, when developing the data warehouses providing access to thousands and
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
thousands of users on global scale. Oracles newest server in marked Oracle8i, is database-resident Java Virtual
Machine for storing and executing Java code on the server, based on Oracle8 server architecture. This integration
between Oracle8 features and Java helps in developing large scaled, internet-savvy applications like data warehouses.
Oracle8i provides features to store and execute Java code within database, and to create stored procedures, database
function and triggers in Java. With the use of Oracle8i the development, deployment and updation of internet
applications have become very simplified.
Following are some of the features available in Oracle8i for the management of data warehousing.
To enhance the data management of very large databases like data warehouses, the partitioning option have been
enhanced in Oracle8i. Some of the advantages of partitioning would be:
Recovery of individual partitions
Only relevant partitions are considered during query execution
Option of reorganization, addition and deletion of individual partitions without affecting the data
Oracle8i introduced new types of partitioning Hash Partitioning and Composite Partitioning. In hash partitioning
Oracle8i uses a hash function to generate almost random numbers for partitioning key, this random number is latter
used to access the partition in which the row is stored. The composite partitioning is the combination of key range
partitioning and hash partitioning first table is partitioned using key range partitioning, then another partition is
created based on some different key using hash function.
The need of summarization and aggregation is very high while accessing a database for decision making. To facilitate this
requirement Oracle8i uses Materialized Views a view for which database server runs the view defining query and stores
the results in database for future usage. Supplementing is the concept of Automatic Query Rewriter using this rewriter
query optimizer can decide when to use materialized views and when to use the tables directly. Now by using
materialized views the database administrator can make summary tables the most frequently used data type in data
warehouses.
Another interesting feature introduced in Oracle8i is the Transportation Tablespace. This is used to transfer some data
from one Oracle system to another Oracle system from operational system to data warehouse. Oracle8i gives the
facility to copy the transportation tablespace without any unloading or reloading of tablespace.
Oracle Warehouse Architecture
This architecture is designed using the RDBMS server and tools provided by Oracle. The Oracle warehouse can be
developed using two-tiered or three-tiered architecture. The two-tiered architecture involves the database server at back-
end and front-end decision support tools. A more complex warehouse involves separate tiers for data access from
operational source, data storage and presentation of data for decision support.
Tier 1 Accessing Source Data
Data can be accessed from multiple sources; including operational systems, Legacy systems and other Oracle
applications. Utilities like SQL*Loader, export/import Oracle schemas, SQL Stored Procedures can be used for the data
transfer. For transferring data from legacy systems and a wide range of other systems Oracle Transparent Gateways are
used.
Tier 2 The Server for Warehouse
The server for data warehousing can be of RDBMS or MDD type. Oracle provides solution for both options. If one
decides to use Multi-Dimensional Database Architecture then there exist Oracle Express Server, and for Relation Database
Architecture the option is Oracle7, Oracle8 or Oracle8i. Warehouse can be designed by integrating the two.
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Tier 3 Decision Support System
To entertain Business Analysts, both DSS and OLAP tools can be provided in the data warehouse. With tools like Oracle
Reports, Oracle Discoverer and Oracle Express, users can have access on data warehouse on the whenever and however
basis.
Figure 13
Oracle Warehouse Toolkits
Figure 14
Considering the importance of data warehousing and reducing the effort required by data warehouse designers, Oracle
provide set of toolkits. Using these tools it would be very easy to transfer operational data from other sources; including
SAP, PeopleSoft and BAAN. Then just connect this warehouse to DSS or OLAP tools and the Decision support system
is ready.. Oracle Warehouse Toolkits provides features like:
Access to updated Operational data
Easy-to-use, graphical analytical tools
Fast and flexible analysis of information
Legacy
System
Warehouse Database
Management Server
DSS
OLAP
DSS
Operational
Systems
OtherOracle
Systems
SQL*Loader
Export/Imort
Stored
Procedures
Oracle
Transparent
Gateway
Oracle
Discoverer
Oracle
Express
Oracle
Reports
Legacy
System
Warehouse Database
Management Server
ORACLE
WAREHOUSE
TOOLKITS
SAP
PeopleSoft
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Figure 15
In addition with the Ideal servers for warehouse data, the tools for DSS and OLAP, like Oracle Discoverer and Oracle
Express, makes Oracle warehouse the ideal analytical environment. Oracle Discoverer, the GUI based reporting tool,
gives an excellent interface for querying and reporting purposes. Typical queries like what and how can be
entertained in Oracle Discoverer. Whereas OLAP tool Oracle Express, provides graphical interface for answering
what-if questions.
Generally, warehouse designers use CASE tool such as Oracle Designer/2000 for designing Oracle Warehouse model.
Using this designer tool, RDBMS warehouse design can be implemented in Oracle8 or Oracle7, whereas MDD
warehouse design in Oracle Express Server. The combination of RDBMS and MDD servers can also be designed to get
the most optimum performance out of data warehouse.
RDBMS
Server
MDD
Server
OLAP/
DSS
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
References
W.H.Inmon known as father of Data Warehousing
What is Data Warehouse?
Vivek R. Gupta Senior Consultant, Services corporation, Chicago, Illinois.
An Introduction to Data Warehousing
David Heise CIO Andrews University
Data Warehousing at Avondale College
Dr. James Goodnight CEO SAS Institute Inc.
Data Warehousing: Understanding Its Role in a Business Management Architecture
Oracle Publications
Oracle7 Server Concept Manual
Oracle Publications
Oracle7 Server Tuning Manual
Oracle Publications
Oracle8 Concepts
Oracle White Paper, June 1997,
Oracle8 for Data Warehousing
Oracle White Paper, June 1997,
Oracle8 Enabling Decisions in the New Business Era
Data Warehousing With Oracle
Oracular, Inc. 317 City Center Oshkosh, WI 54901 920.303.0470 www.oracular.com
Contact : Muhammad Ahmad Shahzad Oracular, Inc.
Oracle White Paper, June 1997,
Oracle8 The Database for Network Computing
Oracle White Paper, June 1997,
Star Queries in Oracle8
Oracle White Paper, June 1997,
Oracle8 The Hub of Oracle Warehouse
Winter Corporation White Paper
Large Scale Data Warehousing with Oracle8i
Oracle White Paper, November 1998,
Oracle8i The Database for Internet Computing
Oracle White Paper, March 1998,
Oracle Warehouse Toolkits