Bookshelf Home | Contents | Index | Search | PDF

Siebel Enterprise Integration Manager Administration Guide > Running EIM > Optimizing EIM Performance >

Run-Time Optimization for EIM

---

This section describes the ways you can optimize EIM performance at run time.

Parallel Processing

Run independent EIM jobs in parallel. Two or more EIM processes can be started simultaneously by using the Siebel Server Manager.

A special setup is not required to run EIM processes in parallel. For parallel processing, the following conditions must be met:

• No duplicate unique keys between runs for inserts.
• No duplicate updates or deletes between runs.
• No lock escalations on either EIM tables or target tables can be tolerated. Set LOCKLIST and MAXLOCKS as high as necessary to prevent this.

NOTE:  If you run EIM jobs in parallel on the same base tables, you might encounter unique constraint errors if you have the same values for the unique index fields in batches being processed by two different EIM jobs.

CAUTION:  Running EIM processes in parallel on a DB2 database may cause a deadlock when multiple EIM processes access the same EIM table simultaneously. To avoid this potential problem, set the UPDATE STATISTICS parameter to FALSE in the EIM configuration file. See Parameter Settings Optimization for EIM for other suggestions.

For more information on parallel processing, see Running EIM Tasks in Parallel.

Transaction Logging

Consider switching off transaction logging during the EIM run. Disabling transaction logging will definitely improve performance; however, this benefit must be balanced with the need for mobile users to reextract afterward. To disable transaction logging, complete Step 2.

Bookshelf Home | Contents | Index | Search | PDF