Pageable frames

SRM attempts to ensure that enough pageable central storage is available to the system. SRM monitors the amount of pageable storage available, ensures that the currently available pageable storage is greater than a threshold, and takes continuous preventive action from the time it detects a shortage of pageable storage until the shortage is relieved.

When SRM detects a shortage of pageable frames caused by an excess of fixed or DREF storage, SRM uses event code ENVPC055 to signal an ENF event. When the shortage is relieved, SRM signals another ENVPC055 event to notify listeners that the shortage is relieved. SRM does not raise the signal for the “shortage relieved” condition until a delay of 30 seconds following the most recent occurrence of a fixed-storage shortage. The intent of signalling this event is to give system components and subsystems that use fixed or DREF storage an opportunity to help relieve the shortage.

Regardless of the cause of a shortage of pageable storage, SRM takes these actions:

Inhibits LOGON, MOUNT, and START commands until the shortage is relieved
Prevents initiators from executing new jobs
Informs the operator that a shortage of pageable storage exists

Further SRM actions to relieve the shortage depend on the particular cause of the shortage.

The following system conditions can cause a shortage of pageable storage:

Too many address spaces are already in storage.
Too many address spaces in storage does not usually, of itself, cause a shortage of pageable storage because SRM performs MPL adjustment and logical swap threshold adjustment, which generally keep an adequate amount of fixed storage available to back the address spaces.
Too much page fixing is taking place.
One or more address spaces are using substantial amounts of storage either through explicit requests to fixed virtual storage or by obtaining virtual storage that is page fixed by attributes such as LSQA or SQA.

There are different types of pageable storage shortages:

A shortage of pageable storage below 16 megabytes.
A shortage when pageable storage has reached a threshold.
A shortage when fixed and DREF allocated to CASTOUT=NO ESO Hiperspaces has reached a threshold.

If too much page fixing is the cause of the shortage of pageable storage, SRM:

Identifies the largest swappable user or users of fixed storage. If any of these users own more frames than three times the median fixed frame count, SRM begins to physically swap them out, starting with the user with the largest number of fixed pages. SRM continues to swap users out until it releases enough fixed storage to relieve the shortage.
Begins to physically swap out the logically-swapped out address spaces if swapping out the largest users of fixed storage does not relieve the shortage of pageable storage.
Decreases the MPLs for those domains that have the lowest contention indices if physically swapping out the logically-swapped out address spaces does not relieve the shortage of pageable storage. This MPL adjustment allows the swap analysis function of SRM to swap out enough address spaces to relieve the shortage.

SRM takes the following additional actions if the shortage of pageable storage reaches a critical threshold:

Informs the operator that there is a critical shortage.
Repeats all the steps described above that are applicable to the cause of the shortage.
Prevents any unilateral swap-in, except for domain zero.

When the shortage of pageable storage is relieved, SRM waits for a delay of 30 seconds following the most recent occurrence of a fixed shortage, and then:

Allows new address spaces to be created through the LOGON, MOUNT, and START commands.
Notifies the operator that the shortage is relieved.
Allows initiators to process new jobs.
Allows unilateral swap-ins.

Note: Those address spaces that SRM swapped out to relieve the shortage of pageable storage are not swapped back in if their storage requirements would potentially cause another shortage to occur.