The Process Engine (PE) data for all isolated regions are stored together in the database in various tablespace areas. To facilitate database administration or to enhance performance, some database administrators, especially those with large databases, might separate isolated region data (such as indexes, rosters, and so on) into different tablespaces. However, this separation still leaves some PE region data in system tables in the shared tablespace.
NOTE In this document, the term tablespace refers to both tablespace for Oracle and DB2, and filegroup for Microsoft SQL Server.
To enable recovery of all the data for one isolated region independent of other regions in the system, use the Enable Region for Recovery feature in Process Task Manager. The feature separates the global PE data for the region into the region-specific tables. This makes it possible to back up all the data for one isolated region, to roll back only this region, and to roll forward without stopping Process Engine or rolling back all of the isolated regions.
In the illustration below, Region 1 data are split out from both the system tables and the region tables and moved to the custom TS1 tablespace.
To separate the data for one isolated region from the data for other regions, use the following general steps. See links to specific procedures in step 3 below.
TIP It is most efficient to do these two steps before you initialize the isolated region to avoid moving existing data from one table to another.
Once you have separated the data and enabled a region for recovery, you can then back up the data for that region independent of other regions. Your strategy depends on your requirements, tolerance of downtime, and most important, sensitivity of your data.
Use the following general procedure to restore or roll forward an isolated region.
NOTE For Point in Time Restore (PITR), do not select a time in the middle of the transfer. Verify that the restore time occurs either before or after (not during) the transfer.