Space announcement


This space provides the same content as before, but the organization of the home page has changed. The content is now organized based on logical branches instead of legacy book titles. We hope that the new structure will help you quickly find the content that you need.


The rebuilding of large nonpartitioned indexes on partitioned table spaces is a major recovery issue.

To recover these objects with a conventional strategy requires a sequential read pass of all partitions of the table space. In some cases, the amount of time that is required to read the partitioned space and rebuild the index with a conventional strategy is prohibitive. To facilitate the recovery of these spaces, BMC AMI Recover provides concurrency in the extraction of keys from multiple partitions.

For example, if you have a large 16-partition table space with a nonpartitioned index that must be recovered (perhaps because of media failure), BMC AMI Recover provides an alternative strategy that uses the RECOVER UNLOADKEYS and RECOVER BUILDINDEX command statements. You can submit up to 16 jobs to extract keys from multiple partitions simultaneously. Sixteen jobs are not required. You can submit up to one job per partition. If you have only eight available initiators, you can submit eight jobs, with each job reading two partitions sequentially. Any combination (10-2-4, 4-4-4-4, and so on) of partitions per job is acceptable.

Specify the RECOVER UNLOADKEYS command statement for the nonpartitioned index and specify the PART option to indicate which table space partition is to be used for this key extraction. Include a DD statement in the JCL to specify the work data set to hold the keys. The extracted keys files are sorted and are then input to the RECOVER BUILDINDEX job where the keys are merged to build the index.

The improved concurrency can drastically reduce the elapsed time that is required to rebuild a nonpartitioned index. Because RECOVER UNLOADKEYS performs the sorting, particularly for a large, partitioned space, you may see performance improvements because a number of smaller sorts are done.

You can extract the keys for RECOVER UNLOADKEYS indexes during the table space recovery process, further reducing total elapsed time for combined table space and index recovery. For an example of a combined LOGSORT/MERGE strategy and UNLOADKEYS/BUILDINDEX strategy, see Example 6: Extracting nonpartitioned index keys.

Was this page helpful? Yes No Submitting... Thank you