Sybase Technical Library - Product Manuals Home
[Search Forms] [Previous Section with Hits] [Next Section with Hits] [Clear Search] Expand Search

Chapter 16:  Aggregation and Partitioning [Table of Contents] Chapter 18:  Creating a WAM for Sybase Adaptive Server IQ

WarehouseArchitect - User's Guide

[-] Part 3 WarehouseArchitect Model
[-] Chapter 17: Generation for OLAP Systems

Chapter 17
Generation for OLAP Systems

About this chapterThis chapter explains how to generate for Online Analytical Processing (OLAP) systems.

What is an OLAP system?

An OLAP (On-line Analytical Processing) system is a decision support system. It is an interface between a decision maker and the data warehouse. It enables a decision maker with complex analysis requirements to query the data warehouse using familiar business terms.

To make informed decisions, decision makers need to analyze data from a number of different perspectives (business dimensions). OLAP systems make it possible to formulate queries using business dimensions. For example, in a query that requests sales by product, or sales by region, product and region are business dimensions.

What is a CUBE?A cube is a representation of multidimensional database data in a dimensional model. The cells of a cube contain data values The intersection of one member on each dimension represents a particular data value.

What is a MOLAP?A MOLAP is a multidimensional OLAP. It relies on precomputed cubes of data to speed responses to queries.

What is a ROLAP?A ROLAP is a Relational OLAP that enables a decision maker to assemble into a cube a subset of related data points that have been stored in a relational database.

Choosing a target OLAP engine

Interfacing with OLAP enginesWarehouseArchitect defines and presents information in a form expected by OLAP engines. The interface between WarehouseArchitect and OLAP engines ensures that the multidimensional database design in WarehouseArchitect is correctly implemented in the target OLAP engine.

The data files created by the rebuild and generate cube function are used by Powerplay and TM1. There is no direct interface between PowerDesigner and these MOLAP tools.

You can change the target OLAP engine into which you generate data.

OLAP engine toolsWarehouseArchitect supports several OLAP engine tools. Cube generation from the WAM creates CSV or TXT flat files containing data for use by MOLAP engine.

OLAP tool

Type

Requires cube generation

Generates structure from WAM interface

Express

MOLAP

(

--

Powerplay

MOLAP

(

--

Essbase

MOLAP

(

(

DSS Agent

ROLAP

--

(

Impromptu

Query and reporting tool

--

( *

PowerDimensions

ROLAP

--

( *

TMI

MOLAP

(

--



* Requires plug-in.

ODF and EXA filesThe OLAP engine dropdown listbox only displays OLAP engines for which there is an OLAP definition file (ODF). Some ODF files are associated with an extended attribute file (EXA). The EXA file is loaded automatically when you select the associated OLAP engine.

Rebuilding and generating cubes

Interface tools which use MOLAP engines require cubes to be rebuilt and generated. Generating a cube creates the data file or files required by the MOLAP to answer business queries.

ROLAP tools do not require data files, you generate from the cube rebuild and generate functions.

Rebuilding a cube

Rebuilding a cube creates a query and inserts it into the SQL Cube box of the fact property sheet. You can view the updated fact property sheet from the multidimensional hierarchy or from the List of Facts dialog box.

You can select only one attribute per dimension to rebuild.

Generating a cube

Generating a cube executes the query created by the cube rebuild and places the query results into CSV or TXT files. The filenames include the name of the fact table that you generate in the cube.

Generation optionsGeneration options indicate the format of OLAP scripts.

Option

Result of selection

Uppercase

Script contains all uppercase characters

Lowercase

Script contains all lowercase characters

No accent

Non-accented characters replace non-accented characters in script

Codes

Script refers to objects by their codes

Names

Script refers to objects by their names

Windows (ANSI)

Script uses ANSI character set

DOS (OEM)

Script uses DOS character set



Generating for PowerDimensions

PowerDimensions uses a server side OLAP architecture that includes multidimensional caching and query optimization features. PowerDimensions client tools include a schema explorer to import metadata and a graphic modeler's workbench to assemble and modify model components.

The PowerDimensions plug-in is installed by the WarehouseArchitect Setup program. You access the plug-in from the WAM menu in the WarehouseArchitect interface. Using the plug-in interface, you can generate the following types of objects:

PowerDesigner object

PowerDimensions object

Table

Table (imported from data source)

Reference

Join path

Model

Model

Dimension

Dimension

Dimension hierarchy

Hierarchy

Attribute

Level

Metric

Measure



Importing tablesStrictly speaking, tables are not generated from PowerDesigner. When you generate a schema with the PowerDimensions plug-in, the table codes in PowerDesigner are passed along to PowerDimensions.

PowerDimensions uses these codes to indicate the data source tables it needs to build a schema. Therefore the data source table codes must match the table codes in the PowerDesigner data model.

You can ensure matching table codes by generating the data source database from the PowerDesigner model, or by reverse engineering the PowerDesigner model from the data source database.

Unsupported objectsIn PowerDimensions, hierarchies are not supported for facts or attributes. Dimension hierarchies can only be created under a dimension. A level can only be created under an hierarchy.

Facts in PowerDesigner are not generated as PowerDimensions objects, but any metrics they contain are generated as PowerDimensions measures.

Login attributesThe PowerDimensions plug-in requires you to specify the following login options:

Login attributes

Description

User Name

PowerDimensions server connection user name. The default user name is superuser

Password

Password of the user. No password is required for the superuser login



Schema generation optionsPowerDimensions builds a schema from tables in the data source database. You can build a schema manually in PowerDimensions, or you can generate it from the PowerDesigner data model.

Database schema options

Description

Generate Schema

Generates the PowerDimensions database schema corresponding to the WAM

Database Server

Name of the database server

Database Type

Name of the database driver. You can select any driver supported by PowerDimensions*

Database Name

Name of the database. This name must be the same as the ODBC data source

User Name

Name of the database user as defined in the ODBC interface

Password

Password of the database as defined in the ODBC interface



* If you are generating PowerDimensions for Sybase Adaptive Server IQ, you must select either ODBC (Generic) or ODBC (Sybase SQL Anywhere) as Database type.

Model optionsIf you select the Generate Schema and the Generate Model options from the PowerDimensions plug-in interface, the schema is built before the model is generated. If you do not select the Generate Schema option, the schema must already exist in PowerDimensions before you can generate the model.

Model option

Description

Generate Model

Generates the multi-dimension model in PowerDimensions

Build Hierarchy

Generates the data elements for the dimension



Extended attributesInformation supported by PowerDimensions, but not supported by PowerDesigner, can be generated via the extended attribute mechanism. Values you assign to extended attributes are generated at the same time you generate a model to PowerDimensions.

For information on PowerDimensions extended attributes, see the section PowerDimensions extended attributes

Generating for Oracle Express

Oracle Express is a relational OLAP system. It provides querying and reporting of multidimensional data generated from a WAM file using Express.

You can generate 3 different types of file from the WAM to use with Oracle Express:

If you generate all files to the same directory, the program file can automatically load the data files that you create with cube generation.

You need to change the target OLAP engine to Express.

Generating for Essbase

Arbor Essbase is a MOLAP data server optimized for planning, analysis, and management reporting applications.

Extended attributesWhen you change target OLAP engine to Essbase, default extended attributes are loaded in the WAM. You can view these attributes in the List of Attributes dialog box for facts, dimensions, and so on.

You must install Essbase on your local machine or copy the Essbase DLL files to your local machine to be able to access an Essbase server.

You need to change the target OLAP engine to Essbase.

Generating for DSS Agent

Microstrategy DSS Agent is a relational OLAP system (ROLAP). It accesses data directly in the data warehouse.

You can generate:

Generating a script for DSS Agent

You can generate a script file that you can run in your DSS Agent interface environment. You must first generate a project in DSS.

The script you generate inserts SQL queries into the structure metadata database. These queries can then be used by DSS Agent to extract data from the data source.

You can generate for DSS without changing the target OLAP engine.

Generating metadata for DSS Agent

Metadata refers to the multidimensional objects that OLAP engines use to formulate queries of data warehouses. The metadata has the same business dimensions defined in the multidimensional model associated with the WAM.

You can generate metadata directly in the OLAP database.

Generating for Cognos Impromptu

Impromptu is a business-driven, interactive query, and report building tool that enables you to build reports on a database.

You must install the Impromptu plug-in from the WarehouseArchitect Setup program for the plug-in to work properly. You generate a structure file from the WAM plug-in interface that you can use in Impromptu to obtain your query results.

From Impromptu you must first create a logical database name for a data source. You must also close the database catalog in Impromptu before you generate a catalog file from the WAM.


Uninstalling the Impromptu plug-in You must uninstall the Impromptu plugin before you uninstall WarehouseArchitect. To uninstall the Impromptu plugin, you must run PDOBJECT.EXE with /unregserver. You can do this from DOS or from the Start bar Run window.

OLAP extended attributes

Extended attributes values can be specific to a particular OLAP engine. These attributes and their default values are loaded automatically into your WAM when you change the OLAP engine.

PrefixesEach extended attribute has a prefix, automatically added to it when it is created. The assigned prefix corresponds to the target OLAP engine that is currently selected.

If you access an extended attribute list from a property sheet you will not see the prefix. You can view the complete list of extended attributes, with their prefixes, from the Dictionary-->Extended Attributes-->List of Attributes menu in WarehouseArchitect.

The prefixes automatically added to an attribute are:

OLAP tool

Prefix

Essbase

ess_

Express

exp_

DSS Agent

dss_

Impromptu

imp_

PowerDimensions

Prefix not applied



Assigning values to OLAP extended attributes

You can assign values to extended attributes in the Multidimensional Hierarchy window from the property sheet of a fact, dimension, hierarchy, attribute, or metric. You can also create new extended attributes.

PowerDimensions extended attributes

This section lists extended attributes with their descriptions and data types for Sybase PowerDimensions generation. Extended attributes are only provided for dimension attributes that are generated as levels in PowerDimensions.

Extended attribute

Data type

Description

AliasColumn

String

Column name for alias

KeyColumns

String

List of key columns separated by a comma

SortColumn

String

Column name for sort



Example 1In cases where you want to display a month by name, but sort it by number, you can assign Month_No as the value for the SortColumns extended attribute. Both Month_No and Month_Name columns must exist in the table you are generating.

Example 2You can also generate a list of key columns to a PowerDimensions level using the KeyColumns extended attribute. For instance, you might generate Year, Month_No to a Month_Name level, or Year, Quarter to a Quarter level.

Essbase extended attributes

This section lists extended attributes with their descriptions, data types, and default values that are necessary for Arbor Essbase OLAP generation.

Fact extended attributes

Extended attribute

Data type

Description and default

ess_application

String

Application name

No default

ess_database

String

Database name

No default

ess_login

Boolean

Login name

Default=FALSE

ess_password

String

Password

No default

ess_server

String

Server name

No default

ess_username

String

User name

No default



Dimension extended attributes

Extended attribute

Data type

Description and default

ess_category

Integer

Dimension category

Default=ess_CAT_NONE

ess_share

Integer

Share option

Default=ess_SHARE_DATA

ess_storage

Integer

Dimension storage type

Default=ess_DIMTYPE_DENSE



Dimension Hierarchy extended attributes

Extended attribute

Data type

Description and default

ess_category

Integer

Dimension category

Default=ess_CAT_NONE

ess_share

Integer

Share option

Default=ess_SHARE_DATA

ess_storage

Integer

Dimension storage type

Default=ess_DIMTYPE_DENSE



Metric extended attributes

Extended attribute

Data type

Description and default

ess_consolidation

Integer

Unary consolidation type

Default=ess_UCALC_ADD

ess_expense

Boolean

Expenditure

Default=ess_FALSE

ess_skip

Integer

Time balance skip option

Default=ess_SKIP_NONE

ess_timebalance

Integer

Time balance option

Default=ess_TIMEBAL_NONE

ess_twopass

Boolean

Calculation order

Default=ess_FALSE




Chapter 16:  Aggregation and Partitioning [Table of Contents] Chapter 18:  Creating a WAM for Sybase Adaptive Server IQ