Restoring multivolume data sets and restoring data sets using multiple target volumes (spill volumes)

Multivolume data sets from a logical data set dump tape can be restored either to a single volume or to multiple volumes. When multiple target volumes are specified, DFSMSdss selects target volumes as follows:

If a target volume that has the same volume serial as the source volume is available and has adequate space, it is chosen.
If a volume of the same device type is available, and if it has adequate space, it is selected.
A volume of a like device type is selected if it has adequate space.
A volume of an unlike device type is selected if it has adequate space.

If you are restoring a multivolume data set from a physical dump, be sure the segments from all volumes are restored with successive RESTORE commands. Restoring a portion of a multivolume non-VSAM data set to a preallocated data set is allowed only if the volume sequence numbers of the source and target data sets are the same.

A VSAM data set that has its index component defined on more than one volume (that is, a multivolume KSDS defined with the imbed attribute) should always be processed logically. If it must be processed physically, it should be treated as an absolute track allocation data set, and its extents restored to their original location. This can be accomplished by performing either a full-volume restore, or a tracks restore of the relevant tracks. If this procedure is not done, the index may become unusable.

During logical restores of VSAM data sets whose data and index components are on different source volumes, DFSMSdss preserves the volume spread if enough target volumes of like device types are specified.

Note: DFSMSdss preserves the volume spread by placing the data and index components on separate devices only if all of the following are true:

The source data and index components reside on separate devices.
The target data set is preallocated with the data and index components on separate devices.
DFSMSdss does not need to scratch and reallocate the preallocated target data set.

For information about when DFSMSdss scratches and reallocates target data sets, refer to Restoring to preallocated targets.

Note: When you are copying or restoring multivolume data sets, be aware of the following considerations:

DFSMSdss does not preserve candidate volumes. However, for SMS-managed data sets, if you copy and do not specify any output volumes, DFSMSdss preserves the source volume count. If you copy and do specify the output volumes, DFSMSdss sets the volume count to the number of output volumes specified.
DFSMSdss does not ensure that the copied or restored data set is on the same number of volumes as the original data set, nor does DFSMSdss ensure that the copied or restored data set extents are the same as the original data set. Instead, DFSMSdss tries to allocate the new data set on as few volumes as possible. This may result in the copied or restored data set becoming a single-volume data set.
In addition, DFSMSdss tries to allocate each volume so that all data is contained in a single primary allocation of contiguous space with few, if any, of the secondary allocations being used.