Reducing resource contention

Stress conditions are an indication that certain limit conditions have been reached and additional processing is required. The transactions involved must wait until resources are released.

The main limit conditions or constraints that can occur in a CICS® system include those listed in Identifying CICS performance constraints.

To summarize, limit conditions can be indicated by the following:
  • Virtual storage conditions (short-on-storage or SOS). This item in the CICS storage manager statistics shows a deficiency in the allocation of virtual storage space to the CICS region.

    In most circumstances, allocation of more virtual storage does not in itself cause a degradation of performance. You should determine the reason for the condition in case it is caused by some form of error. This could include failure of applications to free storage (including temporary storage), unwanted multiple copies of programs or maps, storage violations, and high activity of nonresident exception routines caused by program or hardware errors.

    All new applications should be written to run above the 16MB line. The dynamic storage areas above the 16MB line can be expanded up to the 2GB limit of 31-bit addressing. The dynamic storage areas below the 16MB line are limited to less than the region size, which is less than 16MB.

  • Number of simultaneous tasks (MXT and transaction class limit) reached (shown in the transaction manager statistics).
  • Maximum number of VTAM® receive-any RPLs in use (shown in the z/OS® Communications Server statistics).
  • Wait-on-string and associated conditions for VSAM data sets (shown in the file control statistics).
Check how frequently the limit conditions occur. In general:
  • If no limit conditions occur, this implies that too many resources have been allocated. This is quite acceptable if the resource is inexpensive, but not if the resource is both overallocated and of more use elsewhere.
  • Infrequent occurrence of a limit condition is an indication of good usage of the particular resource. This usually implies a healthy system.
  • Frequent occurrence (greater than 5% of transactions) usually reveals a problem, either directly or indirectly, that needs action to prevent more obvious signs of poor performance. If the frequency is greater than about 10%, you may have to take some action quickly because the actions taken by CICS itself (dynamic program storage compression, release of storage cushion, and so on) can have a perceptible effect on performance.
    Your own actions should include:
    • Checking for errors
    • Raising the limit, provided that it does not have a degrading effect on other areas
    • Allocating more resources to remove contention
    • Checking recovery usage for contention.