Before z/OS V2R1, SMS, when allocating or extending a multi-volume data set, preferred the candidate volumes in the same storage facility image (SFI) if the storage class accessibility attribute was set to CONTINUOUS or CONTINUOUS PREFERRED. Beginning with z/OS V2R1, SMS prefers volumes that are in the same cluster. Similarly, when allocating the target data set for the data set fast replication function, SMS now prefers volumes that are in the same cluster as the source data set. If it is not possible to honor the preference of volumes in the cluster, SMS reverts to preferring volumes in the same SFI.
Data set fast replication operation is more efficient when the entire data set resides in the same cluster. If you wish SMS to select volumes in the same cluster, you should configure your clusters accordingly.
Before z/OS V2R1, SMS, when allocating a striped data set, allocated the stripes across logical control units (LCUs). SMS now attempts to allocate the stripes across separate extent pools, to ensure more uniform performance. If this is not possible, SMS continues to allocate the stripes across LCUs. If you wish SMS to separate the stripes across extent pools, you should configure your storage groups accordingly.
For details on SMS volume selection, refer to the topic about SMS volume selection in z/OS DFSMSdfp Storage Administration.