Tailoring for Performance › Using Select/Omit Maintenance

Using Select/Omit Maintenance

Select/omit maintenance filters records that match the specified selection or omission maintenance. The two types of select/omit maintenance are: static and dynamic.

If the select/omit maintenance is dynamic, all records, regardless of the select/omit set, are included in the access path. The system filters the records as the program reads them. The program returns only the records that match the select/omit maintenance.

If the select/omit criteria is static, only those records that satisfy the select/omit maintenance are included in the access path. As each record is added or changed, the system determines if it should be included in the access path. As the data is read, no filtering is required since the access path maintenance has performed the filtering.

Note: In the DDL generation mode, irrespective of the select/omit criteria setting being static or dynamic, only those records that satisfy the select/omit criteria are included in the access path.

For more information about the relation between static/dynamic maintenance and the generation modes, refer to Select/Omit in DDL Index.

Because of this difference, any access path with static select/omit criteria usually has a separate internal active index, whereas any access path with dynamic select/omit maintenance can share an active index with other similarly keyed access paths even if the select/omit criteria differ. The provision and use of dynamic select/omits increases the possibilities of sharing active indexes.

Note: For join logical files in CA 2E (files with virtual fields) with select/omit criteria, the static select/omit maintenance cannot be used. Use the dynamic select/omit maintenance.

The type of data file is again important. For non-volatile (master or table) data files the logical files should, if possible, have static select/omit criteria. The static select/omit maintenance requires an extra active index to be maintained. But the frequency of change to the data is low and therefore maintenance does not occur often. For volatile (transaction) data files, the type of select/omit criteria depends on the actual number of records being read and the frequency of use of the active index.

If the active index is used infrequently or in batch, the overhead of having dynamic select/omit criteria may might be acceptable.