Allocating XRF data sets

If your installation is using the IMS Extended Recovery Facility (XRF), in which an alternate IMS system acts as a backup to an active IMS system, there are a number of considerations for allocating data sets. Some data sets must be shared between the two systems, others must be replicated, and others can optionally be replicated.

Three main XRF requirements for placing your IMS data sets are:

Mandatory sharing of system data sets

Use of XRF requires that some IMS system data sets, such as the system logs, be available to both the active and the alternate IMS subsystems during the tracking phases. Use of XRF requires that others, such as the DEDB data sets, be present immediately at takeover.

The following data sets must reside on DASD that active and alternate IMS subsystems share:

  • CRITICAL DL/I DATABASE (DFSMDA definitions)
  • DEDB AREA
  • DFSOLPxx (DFSMDA definitions are recommended)
  • DFSOLSxx (DFSMDA definitions are recommended)
  • DFSWADSx (DFSMDA definitions are recommended)
  • IMSRDS
  • IMSRDS2
  • MODSTAT
  • MODSTAT2
  • MSDBINIT
  • RECON1 (DFSMDA definitions are recommended)
  • RECON2 (DFSMDA definitions are recommended)
  • RECON3 (DFSMDA definitions are recommended)

These data sets must be accessible to both subsystems through the catalog structure. Also, do not store an OLDS, WADS, or restart data set (RDS) on volumes containing data sets (IMS or otherwise) that can be subject to a RESERVE operation. Keep such data sets separated.

Mandatory replication of data sets

Certain IMS execution data sets contain information unique to only one subsystem. Replicate these data sets, so each active and alternate IMS subsystem has its own unique data sets. Store these data sets on local, non-shared DASD, and define them in a separate catalog structure. The data sets in this category are:

  • IMSMON
  • LGMSGx
  • LGMSGL
  • MSDBCP1
  • MSDBCP2
  • MSDBCP3
  • MSDBCP4
  • MSDBDUMP
  • QBLKS
  • QBLKSL
  • SHMSGx
  • SHMSGL
  • SPOOLx
  • SYSABEND
  • SYSUDUMP

If your XRF configuration requires that both IMS subsystems be executable on either CPC, these data sets must be on shared or switchable DASD, and in a catalog structure accessible to both subsystems.

Optionally replicating data sets

To avoid single points of failure, you can duplicate certain other IMS execution data sets and store them in non-shared local DASD. Data sets in this category are:

  • DBDLIB (used by DL/I batch)
  • FORMATA
  • FORMATB
  • IMSACBA
  • IMSACBB
  • IMSTFMTA
  • IMSTFMTB
  • JOBS (used in the IMSRDR procedure)
  • MODBLKSA
  • MODBLKSB
  • PGMLIB
  • PROCLIB
  • PSBLIB (used by DL/I batch)
  • SDFSRESL
  • SDXRRESL
  • TCFSLIB
  • OTHER STEPLIB DATA SETS

If your XRF configuration requires that both IMS subsystems be executable on either CPC, these data sets must be on shared or switchable DASD and in a catalog structure accessible to both subsystems.

Other data sets impacted by XRF

When planning your XRF configuration, it is important to consider the possible impact on the other IMS data sets. Also examine the impact on activities other than online execution, such as IMS system definition and the application of SMP/E service. Table 1 provides information about data sets in this category, including descriptions and whether they are managed by SMP/E.

Table 1. Other data sets impacted by XRF
Data set Description Managed by SMP/E
IMS.FORMAT message format library No
IMS.MODBLKS created by SYSDEF Yes
IMS.OBJDSET created by SYSDEF No
IMS.PROCLIB created by SYSDEF No
IMS.REFERAL used with FORMAT No
IMS.SDFSMAC created by SMP/E Yes
IMS.SDFSRESL created by SYSDEF and SMP/E Yes
IMS.TFORMAT test message format library No

Some of these data sets appear in lists in other related topics. You must avoid possible synchronization conflicts.

It is essential to synchronize the DFSMDA members in the IMS.SDFSRESL, or associated libraries, across the XRF complex.

RACF profile data sets

The Resource Access Control Facility (RACF®) profile data sets must be on DASD shared by the active and alternate IMS subsystems. To avoid single points of failure, use the RACF backup facility to keep a second copy of these data sets also on shared DASD.

For additional information about RACF, see z/OS® Security Server RACF General User’s Guide.