This section provides information about reduced direct access optimization.
Reduced optimization in single systems may occur due to the following:
Resources may be taken away from underperforming Sets and allocated to better performing Sets.
For specific files, command profiles of applications may change over time resulting in reduced optimization rates.
Buffer parameters (for example, Maximum Jobs) may be limiting the number of jobs/clients able to connect to the Fastpath buffer. Lesser performing jobs may be connecting
rather than potentially higher performing jobs.
Jobs that are too volatile (high rate of updates, deletions, or insertions) being defined for Fastpath optimization.
The possible reasons for reduced optimization in a multi system are the same as in a single system with one additional reason.
If an individual COR daemon within a COR group shuts down either accidentally, normally (for example, due to an IPL of the LPAR), or abnormally, update notifications from Adabas databases local to that COR daemon can no longer be forwarded to remote Fastpath buffers operating in peer COR daemons. Under these circumstances, direct access optimization will be turned off for those files that rely upon the forwarding of such update notifications.
It is important to note that even when the COR daemon is restarted, direct access optimization will remain turned off for those files.
An AFP RESTART operation is required to restart full direct access optimization.
For more information, refer to the AFP RESTART information in
the section Operator Commands, paying particular attention to the WARNING note in
relation to multi system implementations.
