Choosing Physical Tuning Options › Placement of Entities in Areas

Placement of Entities in Areas

Why separate entities?

To facilitate certain processing operations, you can instruct CA IDMS/DB to divide the database into separate areas. Each area can contain one or more entities. You place database entities in separate areas to:

• Minimize processing interruptions that might be caused by backup and recovery procedures. CA IDMS/DB provides standard system utility programs that allow the system operator to rollforward/rollback or dump/restore only those areas in a database that require backup and recovery. Before performing backup and recovery procedures, the operator typically varies each area or file that is currently held in update usage mode to retrieval (or offline mode). Once an area has been varied to retrieval or offline mode, further update processing is not allowed. By assigning entities to separate areas, you can ensure that backup and recovery procedures impact the minimal number of applications.

  For further information on backup and recovery, see CA IDMS Utilities Guide and CA IDMS Database Administration Guide.

• Reduce time required to perform maintenance activities (such as unload and reload by area). By separating entities into separate areas, you make the amount of data processed smaller, which, in turn, reduces the time required for the processing.
• Reduce cluster overflow. The impact of large cluster sizes can be reduced by separating one or more entity types into separate areas. This is especially effective if less-frequently accessed entities are separated.
• Improve efficiency of serial processing. If an entity (or entities) is to be retrieved mainly by area sweeps, that entity (or entities) should be assigned to a separate area.

Guidelines

Consider the following general guidelines for assigning entities to database areas:

• Whenever possible, place indexes in separate areas. If two or more indexes can be accommodated by the same page size, you can place the indexes in the same area. If using a non-SQL implementation, consider segregating each index in its own page range if they are in the same area or if the indexes are restricted to separate page ranges.
• In general, you should store only one type of entity cluster in each area of a database.
• Nonclustered entities can be placed together in a separate area or can be included in an area containing a cluster, provided that CALC overflow will not be a problem.