BATCH_RCLMIGDS infos
Article ID: 29778
Updated On:
Products
Issue/Introduction
The option can be set via SETALLOC or in the ALLOCxx Parmlib member using
parameter BATCH_RCLMIGDS=SERIAL|PARALLEL
SERIAL indicates the system will allow the CATALOG LOCATE function to recall
data sets on its behalf, resulting in serial recall processing.
PARALLEL indicates the system will note which data sets are migrated and
recall them in parallel, potentially reducing job execution time.
If VANTAGE - ARM is installed on the partition:
ARM can start one or more DMSAR servers, optionally
dedicating them to either tape or disk restore work, and leave them active,
waiting for more work. So if there are 5 or more DMSAR servers defined in
ARM, then above sample is covered by ARM.
We feel that ARM is a superior solution rather than Parallel Batch Restore
(PBR). The z/OS solution will restore the data sets for a step where ARM
and Exit6 will restore the data sets for the job prior to the job being
started. In both ARM/Exit6 and PBR we are limited by the media that the
archive is stored on (single tape, multiple tapes, DASD archives) but ARM
has the higher potential to parallel process the requests since it will
likely be processing more data sets the entire job rather than a single
step. I think that HSM is also limited in the number of parallel recalls
that can be done, but I don't know what the default number is, or any other
information about it. I do believe that you can limit how many tape drives
can be used at once, just as we can do with ARM.
If we don't have VANTAGE - ARM installed on the partition:
The parameter should still be left to SERIAL (which should be the default) for CA Disk.
PARALLEL would enable HSM to recall data sets in parallel for a Batch job.
Environment
Component:
Feedback
