When the extraction is complete, the keys are read back in and sorted, and then the index is built from the sorted key/RID pairs.
However, using the NOWORKDDN option allows the extracted keys to go directly into a sort process. This strategy avoids writing and rereading the key work data set, and no work data set is required to hold the extracted keys. While restartability may be affected (see the table in, Restarting a BMC AMI Recover job), the I/O reduction significantly reduces the time that is required to rebuild the index.
If you do not specify WORKDDN, the NOWORKDDN option is the default. NOWORKDDN is required for the rebuild of the partitioning index when used together with the RECOVER UNLOADKEYS strategy for the nonpartitioning indexes.
For more information, see NOWORKDDN.