|Bookshelf Home | Contents | Index | PDF|
It is recommended that you partition EIM tables based on their U1 indexes, that is, the IF_ROW_BATCH_NUM and ROW_ID columns. This method of partitioning allows an EIM batch input to be assigned to one partition, thereby allowing multiple EIM batches to be run in parallel.
The sample storage control files shipped with Siebel 8.0 contain partitioning for a number of EIM tables based on their IF_ROW_BATCH_NUM column; these are the tables that are recommended for partitioning. You can accept this recommended partitioning approach or use the Database Storage Configurator to perform your own custom partitioning of EIM tables.
NOTE: If you choose to use the partitioned EIM tables in the Siebel partitioning schema, before using EIM, verify that you are using the correct key ranges, because key ranges depend on the batch numbers used.
To compute the optimal number of partitions for an EIM table, divide the number of rows in the EIM table by the number of parallel processes you intend to run. The result is the approximate number of partitions you need to create for the table.
The following examples illustrate how you might partition the EIM tables that are used to populate the S_ORG_EXT and S_OPTY target base tables. Using the columns shown ensures that the data is physically clustered based on the clustering index keys of the target tables.
For information on using the Database Storage Configurator to perform your own custom partitioning of EIM tables, see Modifying a Storage Control File Using the Database Storage Configurator.
|Implementing Siebel Business Applications on DB2 UDB for z/OS||Copyright © 2008, Oracle. All rights reserved.|