z/OS DFSMSdfp Storage Administration
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Cluster and storage facility image considerations

z/OS DFSMSdfp Storage Administration
SC23-6860-01

The DFSMSdss data set fast replication function requires that all volumes of a multi-volume data set reside in the same storage facility image (SFI). Data set fast replication is most efficient when both source and target data sets reside in the same cluster. To help you take advantage of data set fast replication, SMS volume selection:
  • When allocating or extending an SMS-managed multi-volume data set that has point-in-time copy volumes requested, prefers candidate volumes that are in the same cluster, and, if that cannot be honored, prefers candidate volumes that are in the same SFI. Point-in-time copy volumes are requested when the ACCESSIBILITY parameter in the storage class is set to CONTINUOUS or CONTINUOUS PREFERRED.
  • When allocating the target data set for data set fast replication, prefers volumes in the same cluster as the source data set, and, if that cannot be honored, prefers volumes in the same SFI as the source data set.

To make efficient use of the fast replication function, ensure that enough volumes are in one SFI to accommodate the likely increase in allocations to volumes in these storage groups and SFIs. For optimum efficiency, ensure that enough volumes are in one cluster to accommodate the likely increase in allocations to volumes in these storage groups and clusters.

The preference for volumes that are in the same cluster or SFI applies to all multi-volume allocations with ACCESSIBILITY = CONTINUOUS or CONTINUOUS PREFERRED. However, there are a few situations where SMS volume selection will not take the cluster or SFI attribute into consideration. These situations are:
  • The volumes specified by the user for a guaranteed space request are not in the same SFI.
  • No clusters or SFIs have a sufficient number of volumes to meet the number of volumes required for allocation.
  • No clusters or SFIs have a sufficient number of unique controllers to meet the stripe count for an striping allocation.
  • During extend processing, the volumes that the data set resides on are not in the same SFI.
  • Space Constraint Relief (SCR) processing is in effect.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014