Bookshelf Home | Contents | Index | Search | PDF |
Siebel Analytics Performance Tuning Guide > The ETL Process > Remove Unused Batches >
Example Company
For example, a company named "Data Warehouse Inc." only intends to use the Activity and Service Request fact tables. Then, the administrator needs to maintain only these two fact tables and the dimensions they reference.
NOTE: The examples in the following sections show you how to set up the initial load batches; you must also set up the incremental batches in a similar fashion.
Determine the Dimension Tables
To determine the dimension tables referenced by the two fact tables, look in Siebel eBusiness Data Warehouse Data Model Reference guide. According to this guide, the Activity and Service Request fact tables reference these dimension tables:
Define the Batches
Now that you have defined all the required dimensions and fact tables, you need to set up batches to load these tables. The following procedure rearranges the batches already defined in the Informatica repository. These batches were shipped with Siebel Analytics.
This example rearranges the sessions into two batches so they load only the dimension and fact tables listed above:
- DimensionStaging3
- DimensionStaging4
As shipped, these batches contain some of the required sessions as well as several sessions which are not required.
To set up these batches, you need to locate the sessions that load the required dimensions tables and move them into either batch three or four. You also need to remove the sessions which are not required.
To rearrange batches
- Login to the Informatica server manager and open Siebel_DW_Rep.
- Open Full_Extract_Siebel_DW batch as shown in the figure below.
- Locate the sessions required to load all the required fact and dimension tables.
Some of these sessions already reside in the DimensionStaging3 and DimensionStaging4, but several do not.
- Drag the names of the required sessions to the DimensionStaging3 and DimensionStaging4 batches as shown in the following table.
CAUTION: ActivityTemp batch must run before SDE_ActivityCost Fact 1. For a successful load, you must always keep dependencies among the sessions and batches in mind.
- Remove unnecessary sessions from the DimensionStaging3 and DimensionStaging4 batches as illustrated in the following table.
Rather than removing and destroying the sessions that are not required, they are moved to a batch that will be disabled. Then, if you need the session in the future, you can retrieve it from the disabled batch.
- Rename and configure each batch.
- Select the batch.
- From the Menu, select Operations > modify batch name as shown in the figure below.
- Change the names to StagingDWINC_1 and StagingDWINC_2.
- Check Concurrent.
This enables the mappings within the batch to run in parallel.
Disable the following batches, which are not required.
- DimensionStaging1
- DimensionsStaging2
- FactStaging1
- FactStaging2
- FactStaging3
- FactStating4
To disable a batch, clear the Enable box.
- Open Full_load_Siebel_DW-Dimensions batch.
- Rearrange the existing Dimension3 batch as shown in the following table.
CAUTION: Person dimensions have to be loaded after Househould dimension. In the present case, the interest is not in Household dimension; otherwise when moving you must make sure that you run one after the other.
- Open Dimension3, MapEnabledDimensions, and Map_dimensions batches.
- Disable the following batches.
- Vendor
- Territory
- Industry
To disable the batch, clear the Enable box.
- Edit MapEnabledDimensions as shown in the figure below.
- Check Concurrent if OLAP is Oracle; clear Concurrent if OLAP is DB2 or SQL Server 2000.
- Open Full_Load_Siebel_DW_Facts and rearrange the Fact2 and Fact3 batches as shown in the following table.
- Rename Fact3 to FactForDWINC.
- Disable Load_Hierarchy batch if you do not want Account Hierarchy.
- Disable the following batches:
- Dimensions1
- Dimensions2
- Fact1
- Fact2
- Facts
- Dimensions
- Load_KPI
CAUTION: The dependencies among sessions which must be run must be established. The dependencies of the DW ETL batch as shipped are listed below.
ETL Batch Dependencies for a Full Load
Batch names occur in italics.
- Session SDE_ActivityCost Fact1 has to run after sessions in ActivityTemp batch.
- Session SIL_HouseholdDimension1 has to run after session. SIL_PersonDimension1.
- Within the ProductAttributeName batch:
Session SIL_ProductAttributeNameDimension_Unspecified & session SIL_ProductAttributeNameDimension1 should run after the session SIL_ProductDimension1.
- Session SIL_SurveyFact1 should run after session SIL_ServiceRequestFact1.
- Session SIL_AgreeFact1 should run after session SIL_AgreeItemFact1.
- Do not alter any sessions before the session Extract_RestartNextBatch4 and in UpdateRowImage batch.
- Any session that is loading a Dimension Staging table or a Fact Staging table can run in parallel (concurrently with each other).
- Dimension tables must be loaded after their counterpart Staging tables are loaded.
- Fact tables and Helper tables have to be loaded after Dimension tables.
- Hierarchy tables have to be loaded after their counterpart Dimension Tables have been loaded.
- Load_Aggregates and Load_Pipeline batches have to be loaded after Full_Load_Siebel_DW_Facts batch is completed.
- Sessions in batches Facts and Dimensions (these batches are inside the Full_Load_Siebel_DW_Facts batch) have to be loaded after the ETL process has populated all the dimensions and facts tables. Sessions in the Dimensions batch have to be loaded after sessions in the Facts batch.
- KPI batch has to be run after all dimension tables have been loaded.
- Slowly Changing Dimension (SCD) sessions should run after their counterpart Dimension table has been loaded e.g. SIL_PriceListItemDimension_SCD1 should be run after SIL_PriceListItemDimension1.
Bookshelf Home | Contents | Index | Search | PDF |
Siebel Analytics Performance Tuning Guide Published: 18 April 2003 |