About Choosing a Clustering Index
In previous versions of DB2, the partitioning index was automatically the clustering index. In DB2 for z/OS Version 8 and later releases, data can be clustered by any index. Data in a partitioned table space no longer needs to be clustered by the partitioning index, although in many installations, this might be the most efficient option. For example, it is recommended that child tables are partitioned so that all records related by the parent table ROW_ID are contiguous within the same partition and a similar approach is recommended for intersection tables. By choosing the most appropriate clustering index for each of the Siebel-partitioned tables, you can optimize query performance.
In choosing a clustering index, consider both day-to-day online access to the Siebel tables and processes such as EIM, Siebel Remote, Assignment Manager, and Workflow. The recommended clustering index for each of the Siebel-partitioned base tables is shown in the following table.
Only one clustering index can be defined for a table. When you have altered the index to make it clustering, run the REORG utility on the table space and then run RUNSTATS.