Planning to Manage Temporary Data

Properly managing temporary data involves determining which data sets can be directed to VIO or to the primary or large storage groups. This helps improve performance and reduce I/O controller and device workload. VIO simulates I/O to the VIO UNIT in processor and auxiliary storage that you specify in the VIO storage group.

Review the following advice on VIO usage:
  • Do not direct DFSORT temporary SORTWKnn data sets to VIO.

    SORTIN and SORTOUT data sets might be good candidates, especially during the offshift processing periods. However, we do not recommend allocating DFSORT temporary work data sets as VIO data sets. Instead, consider using the Hipersorting function of DFSORT. Hipersorting improves DFSORT elapsed time and reduces execute channel programs (EXCPs) by using Hiperspace™ in place of some or all of the SORTWKnn space. DFSORT queries the system about paging activity and, based on the information returned, allocates Hiperspace as long as it does not adversely affect the performance of other programs using expanded storage. For more information about Hipersorting, refer to z/OS DFSORT Application Programming Guide or to z/OS DFSORT Tuning Guide.

  • Do not direct OS COBOL II Interactive Debug for SYSUT5 to system-managed storage. An unmovable data set is dynamically allocated. This data set cannot be system-managed.
  • Assess your systems' paging activity and the size of central and expanded storage to determine the maximum data set size to be directed to VIO:
    • Temporary data sets between 2 MB and 6 MB are potential candidates for VIO. IBM® recommends 2 MB as an initial threshold.
    • Evaluate the amount of expanded and central storage on all CPUs if you have multiple processors.
    • Create a specific VIO storage group for each CPU based on its storage size. You can tailor the VIO unit and maximum size in each storage group definition and set the status to DISALL for the other systems in the multi-processor complex.