The parallel index update follows the parallel data load in the same REORG function. The number of subtasks used to perform the parallel index update can range from 1 through 25. The default is 10. The physical tracks in the index data set are divided into the number of subtasks. Each subtask reads every block in its part, and updates every pointer as required. Blocks with a change are written. Blocks without a change are not written. Blocks are read and written as one or more tracks.
The parallel process first occurs for the index area IXX to completion. If the database has Multi-Dataset Index areas, the parallel process next occurs for each index area in turn.
The index update process is based upon the data received and the currently loaded index. It cannot validate or protect users from error conditions introduced by mishandling of data input files. For example, the utility does not sort input key values and compares them to the current index. If you were to add keys or change the fields of any current keys, missing or incorrectly indexed data would result. Be aware, however, that if a row or rows were deleted, as in a user program that purges data from the extract files, the utility does work, and you can take advantage of this fact, as required.
|
Copyright © 2015 CA Technologies.
All rights reserved.
|
|