Siebel Performance Tuning Guide > Tuning Siebel Remote > Tuning Siebel Remote Server Components >

Increasing Throughput for the Database Extract and Parallel Database Extract Components

This topic is part of Tuning Siebel Remote Server Components. It provides tips to help you improve the throughput for the Database Extract and Parallel Database Extract components.

Use Parallel Database Extract for extracting a regional database or extracting remote users who have very large visibility presence. Some of the characteristics of Database Extract and Parallel Database Extract follow:

• Parallel Database Extract can support only one task at a time. You cannot run multiple instances of Parallel Database Extract simultaneously, as you can do with Database Extract.
• Parallel Database Extract takes advantage of servers that have multiple CPUs and RAID disc array by using multiple threads.
• Database Extract by default uses a data file type of Bulk Copy format (BCP) to export data. This can lead to a faster time in extracting and initializing than using the data file type DAT.

The following list describes tips to improve the throughput for the Database Extract and Parallel Database Extract components:

• Run multiple concurrent instances of Database Extract

  You can increase throughput by running multiple concurrent instances of the Database Extract component. Each instance of the Database Extract component requires a temporary table. This table is called S_DOCK_INITM_N where N equals the value of the parameter TS Table Number (alias TSTableNum). TS Table Number specifies the number of the temporary table that serves the Database Extract component.

  For example, if TS Table Number equals 1, then temporary table 1 (S_DOCK_INITM_1) serves the instance of the Database Extract component that is currently executing.

  By default, 48 temporary tables are available for use. If you require additional tables, then you create them using Siebel Tools.

  The recommended number of temporary tables to use depends on the database platform in use. For example:

  • Oracle Database

    The number of temporary tables that you use depends on the size of the shared pool that this database server can access. If the size of the shared pool is less than 300 MB, then it is recommended that you use one temporary table and execute one instance of the Database Extract component. If the size of the shared pool is greater than 600 MB, then using one temporary table for each instance of the Database Extract component can increase throughput.

  • Microsoft SQL Server and IBM DB2

    Use one temporary table for each instance of the Database Extract component that you execute. For example, if you execute 11 instances of the Database Extract component, then use 11 temporary tables.

• Stop Transaction Router component

  Stop the Transaction Router component if you are running multiple instances of the Database Extract or Parallel Database Extract component in order to avoid table locking.

• Extract lists of users simultaneously

  Separate users to be extracted into groups of 50 to 100 users per list and extract these lists of users simultaneously using the Database Extract and Parallel Database Extract components. These components extract the Enterprise visibility data for all users in the list once.

• Set the Truncate TS table parameter to TRUE

  Setting this parameter (alias TruncateTSTable) to TRUE can improve performance, because deletions from S_DOCK_INITM_N tables are usually logged and can take a long time.

• Defragment tables

  Defragment tables to which you intend to extract data and defragment S_DOCK_INITM_N tables.

For more information about the components and parameters described here, see Siebel Remote and Replication Manager Administration Guide. For more information about performance issues for the Database Extract component, see 477174.1 (Article ID) on My Oracle Support. This document was previously published as Siebel Troubleshooting Steps 15.