Defining pools of volumes

To ensure that recalled non-SMS-managed data sets go to the correct volumes, you can organize your non-SMS-managed volumes into pools. You can define either pools that accept only specified data set names during recall or pools that accept for recall any data set that migrated from them. In this topic, you are first shown how to define the pools, and then the effects of the pools are described.

DEFINE POOL(GROUP1 VOLUMES(UG1001 UG1002 UG1003))

DEFINE POOL(GROUP2 VOLUMES(UG2001 UG2002 UG2003))

DEFINE VOLUMEPOOL(SAMPLE VOLUMES(VP0001 VP0002 VP0003))

The first two examples specify data set pools. The data set pool list specifies DFSMShsm primary volumes where DFSMShsm is to recall data sets whose set of initial characters match that pool ID. For the first example, data sets with an initial character sequence of GROUP1 are recalled to it. For the second example, data sets with an initial character sequence of GROUP2 are recalled to it.

The third example specifies a volume pool. A volume pool associates volumes from which data sets migrate with other volumes, to aid in the selection of target volumes for recall.

You can specify nonresident volumes in a volume pool, to allow recall of data sets that migrated from volumes that no longer exist in the system. Not all volumes in a volume pool are required to be managed by DFSMShsm. You can specify SMS-managed volumes in volume pools, although they will never be selected from the pool for non-SMS-managed data.

Data sets that migrate from a volume in a volume pool are recalled to an online volume that has been added to the list of volumes managed by DFSMShsm using the ADDVOL command. The target volume selection is based on the volume pool configuration at the time of recall. To change the definition of a volume pool (adding or deleting volumes from the pool), the DEFINE command must include all the volumes you want as part of the pool, including any volumes you selected with a previous DEFINE command if you still want them as part of the pool. The volume pool is limited to 140 volumes.

All volumes in a pool need to have the same geometry. For example, you could have a volume pool that included a 3390 Model 3, and a 3390 Model 9 because they share the same geometry. This means that a data set originally allocated on one model could be recalled to a different model.

A primary volume can be in any number of data set or volume pools and can also be available for general recall use if it is online, permanently resident or reserved, and has the AUTORECALL subparameter in its ADDVOL command.