Separating the data set name filter from the policy itself

While it is possible to create data set name filter VRSs that also contain retention and movement information, doing so will eventually result in many duplicate policy definitions. If you ever need to change any of your policies, you will need to locate and update many different copies of the same policy. Also, if these policies are separately defined, you run the risk of inconsistent policies being defined for data sets that should all be using the same policy.

You can avoid these problems by separating the data set name filter from the policy itself, which enables clear and well-defined service levels to be set up for tape management. These policy and service-level VRSs can then be easily modified as required without changing the filters that select them. You can achieve this separation by creating a data set name VRS that uses a COUNT of zero, so that the first VRS in a chain has no retention specification. The NEXTVRS in the chain and subsequent vital record specifications are then used to specify the complete retention policy. To do this when adding a new data set to VRS retention:

Select an existing service level represented by a VRS chain that starts with a retention NAME VRS.
Define the data set name VRS, specifying DSNAME, JOBNAME, and GDG or NOGDG as required, with COUNT(0) and NEXTVRS(name), where NEXTVRS(name) identifies the beginning of a subchain that specifies the desired policy.

For example, suppose you want to assign a new data set name to an off-site retention policy to retain data for 365 days and then until expired and that you have an existing policy specified by a VRS subchain named YEAROFF:

  RMM ADDVRS NAME(YEAROFF) COUNT(365) DAYS LOCATION(OFFS) NEXTVRS(UEX)
  RMM ADDVRS NAME(UEX) UNTILEXPIRED

To apply this policy to the new data set name, specify:

  RMM ADDVRS DSNAME(‘NEW.DATASET.**’) NOGDG JOBNAME(*) NEXTVRS(YEAROFF) COUNT(0)

Using COUNT(0) in a VRS specifies that the VRS in question cannot retain a data set, and that VRS is to continue with the next VRS in the chain, if any. In the same way, when you specify COUNT(0) as part of an ANDVRS group, the ANDVRS group cannot retain a data set. When you specify COUNT(0), any location or retention criteria you specify for the VRS is ignored by DFSMSrmm.

COUNT(0) can be specified with the ADDVRS or CHANGEVRS subcommands. You can also use the EDGPV200 panel (DFSMSrmm Add Vital Record Specification) to create a data set name filter by specifying YES in the "Add data set filter VRS" field. The added VRS will be set with COUNT=0 and chain type NEXT.

If you have existing data set name VRSs that also contain policy information and you wish to convert them to separate the data set name filter from the policy information, you can use the EDGRVCLN REXX procedure. The LIST(FILTER) parameter of EDGRVCLN will provide a list of data set VRSs that are candidates for using COUNT(0). The FIX(FILTER) parameter of EDGRVCLN will implement the changes recommended by LIST(FILTER). See EDGRVCLN REXX procedure to clean up vital record specifications for additional information on using EDGRVCLN.