Implementing Siebel eBusiness Applications on DB2 UDB for OS/390 and z/OS > Migrating Data Using Siebel EIM > Setting Up EIM for DB2 >

Planning Data Mapping

Data mapping—an important part of preparing to run EIM—is particularly important on DB2 due to the way DB2 stores data in tables and indexes. Your data mapping should take into account such factors as partitioning, look-ups during the load phase, and searches for online transactions.

Only Load Necessary Data

The Siebel eBusiness Applications Data Model provides for many possible business needs and configurations. Many of the tables, columns, and indexes provided with Siebel eBusiness Applications might not be relevant to your business. Because DB2 stores information in all columns, including NULL columns, you should carefully consider which data should be imported into your Siebel application as part of your initial planning. Unnecessary data increases the size of tables and indexes, which can affect performance.

For information on dropping columns and indexes and reducing column sizes using Siebel Tools, see Customizing a Development Environment.

Do Not Load MS_IDENT

The MS_IDENT column exists in every EIM table and is specific to DB2 UDB for OS/390 and z/OS. It allows DB2 to automatically generate unique and sequential values for each row in a table.

This column is the unique identity column created using the "Generated Always with Identity" option.

CAUTION:  If you load data into this column using load utilities, your import might fail. This can particularly be an issue when migrating data from a Siebel application on DB2 for UNIX or Windows to a Siebel application on DB2 UDB for OS/390 and z/OS.

Presorting Data During Initial Load

When loading the EIM tables with legacy data, sort this data on the partitioning key that will be used on the target base table. When sorting string data, use the same character-collating sequence used in the target database. During this load, you should also preassign and load batch numbers corresponding to the partitioning scheme of the target base tables when possible. This is particularly useful for those processes such as Accounts or Contacts, where the partitioning key is based on business data rather than on a generated key. This generally improves EIM throughput, as the data is then loaded sequentially and by partition into the target base table.

TIP:  To further improve performance, once the EIM tables are loaded with partitioned data, you can unload this data in clustering order, then reload it.


 Implementing Siebel eBusiness Applications on DB2 UDB for OS/390 and z/OS 
 Published: 18 April 2003