Bookshelf Home | Contents | Index | Search | PDF

Upgrade Guide for Microsoft Windows > Preupgrade Tasks >

Preparing Tables for Upgrade

---

Take the following measures to prepare your tables for the upgrade.

Preserving Custom Indexes on Tables

Consider the following implications and carefully plan your upgrade to preserve custom indexes.

• Custom indexes against extension columns on obsolete tables. Careful planning is necessary if you have custom indexes against extension columns on obsolete tables. You must migrate data to new extension columns before you upgrade the Siebel database schema. For assistance, contact Siebel Technical Support.
• Custom indexes that were not defined through Siebel Tools in the Siebel repository. If your DBA created custom indexes in your current schema that were not defined through Siebel Tools in the Siebel repository, these indexes are dropped during the upgrade process.
• Custom indexes on interface tables. Custom indexes on interface tables are not re-created during the upgrade, but they can be restored when the upgrade is complete.
• Custom indexes on base tables. Release 7.x upgrade automatically drops and re-creates custom indexes on base tables.

NOTE:  Custom indexes may need to be changed to reflect schema changes. You should reevaluate custom indexes for applicability in the new release.

For more information about applying custom indexes, see Siebel Tools Reference.

Considerations for Clustered Indexes

If you created clustered indexes on base tables and Release 7.x introduces a different clustered index on the same table, the upgrade process re-creates custom indexes as nonclustered and creates the Siebel index as clustered.

For IBM DB2 UDB, indexes that reside on tables that have been defined with append mode enabled are re-created as nonclustered indexes during the upgrade. Tables that have been created with append mode enabled cannot contain clustered indexes.

Disabling Customized Triggers

Release 7.x does not support customized triggers. If you have created customized triggers on your Siebel base tables, disable them before you perform the upgrade. You must re-create them after the upgrade is finished.

Dropping Customized Views

If you have created customized views on your Siebel base tables, you must drop them before you perform the upgrade. If they are still applicable after the upgrade, you must re-create them after the upgrade is finished.

Identifying DB2 UDB Long Columns for Truncation

In Release 7.5, the maximum length for DB2 UDB long columns with a type of varchar has reduced to 16350 from 16383. The upgrade truncates long columns of type varchar that exceed 16,350.

To prevent a data truncation error that may cause transaction processing (txnproc) or transaction routing (txnroute) to fail, perform the following steps to identify these columns and reduce the data in these columns.

To identify and reduce the length of long varchar columns

1. From any shell, open the script chk16350.bat, and edit the following parameters as appropriate for your deployment:

   SRC_USR = username of the source database

   SRC_PSWD = password for the source database

   SRC_TBLO = tableowner of the source database

   SRC_TBLO_PSWD = tableowner password for the source database

   SRC_ODBC = ODBC data source name of the source database (edit the value "CHANGE_ME")

   SRC_REPOSITORY_NAME = repository name of the source database

   DBSRVR_ROOT = directory where you installed the dbsrvr component of the Siebel Server (edit the value "CHANGE_ME")

   SIEBEL_ROOT = directory where you installed the siebsrvr component of the Siebel Server (edit the value "CHANGE_ME")

   VALID_RESULTS_DIR = directory where you want the output files to be generated (edit the value "CHANGE_ME"); this must be an existing directory

   This script produces two files:

   • long_trunc_cols.rpt. This report identifies all long varchar columns that are longer than 16350 characters.
   • update_trunc.sql. This SQL file generates update statements that truncate identified columns to 16350 characters.
2. Reduce the data in these columns using either of the following methods:
   • Manually review the columns in the long_trunc_cols.rpt report and manually reduce the size of each column identified.
   • Run update_trunc.sql using the DB2 command line processor.

CAUTION:  If you do not truncate or otherwise reduce the data in these columns, a "data truncated" error will occur, and transaction processing and transaction routing may fail.

Updating Statistics on DB2 UDB

DB2 UDB databases use statistics about tables and indexes to compute the most efficient access plans. When statistics become inaccurate, which can happen for tables with high insertion rates, high deletion rates, or both, and for associated indexes, the performance of database operations can degrade dramatically.

To update statistics, run updatestats.sql or use DB2 UDB specific statements to refresh statistical information in the Siebel database.

Bookshelf Home | Contents | Index | Search | PDF