Previous Topic: Specifying an Appropriate Amount of Virtual StorageNext Topic: Evaluating the Use of LLA and VLF

Tuning CA PMO › Addressing Special Tuning Concerns › Examining Paging Rates

Examining Paging Rates

CA PMO minimizes the number of pages that it accesses during hash table searches. Normally, CA PMO hashes to a single page within the table. However, as with any application that references pages, CA PMO may increase paging on your system. To see if paging is a problem, examine the effect CA PMO has on user response time and system throughput. If they improve, the impact of CA PMO on your system has been positive. If you want to lower the paging rate, follow these steps:

  1. Eliminate clone libraries that you created to reduce bottlenecks caused by library directory searches on DASD. CA PMO needs only one hash table to handle all directory searches of a particular private library. The pages of this table will be referenced more often and it will have a higher probability of being resident in processor storage when CA PMO accesses it. In addition, a lower number of virtual pages will be competing for processor storage.
  2. Since CA PMO provides I/O savings, you can shift some of your I/O resources from libraries to paging.
  3. To ensure that the managed list is always resident in processor storage, specify FIX=YES (the default). If paging is posing a very serious problem, you may want to specify FIX=NO so that less frequently used pages of the managed list can be written to auxiliary storage. However, this could easily result in increasing the average elapsed time for managed list searches.