Physical design for shadow tables

A good physical design for shadow tables considers physical requirements such as memory and I/O. It also considers guidelines about replication and use to gain the performance benefits that shadow tables provide.

Follow these guidelines when you design shadow tables:

Ensure that you have sufficient I/O subsystem bandwidth. For more information, see Latency and throughput of shadow tables.
Use dedicated buffer pools. For more information, see Memory management for shadow tables.
Populate the source table before you start the subscription to replicate data to the shadow table. This action ensures that you build an appropriate compression dictionary. For more information, see Compression dictionaries and shadow tables.
Set the InfoSphere® CDC fastload_refresh_commit_after_max_operations system parameter to the maximum cardinality of your shadow tables. For more information, see Compression dictionaries and shadow tables.
Ensure that you have shadow tables for all the tables that you reference in analytic queries. For more information, see Determining the set of tables to shadow.
Maintain shadow tables only when they are useful. For more information, see Determining the set of tables to shadow.