A) I have not observed this phenomenon.
B) I would question the assertion of 10 sec. vs 20 min. Can you go back to the
1.12 system and validate?
Things I would check (in no particular order).
1) Relative TSO region size
2) Was capping in effect at this time? (4 hr RA)
3) relative workload (Is it the same at 9 am as at 10 pm?)
4) other activity, (swaps, catalog backup,...)
5) system paging rate
6) other HSM activities (BACKVOL CDS, BACKUP, VOLUME DUMP)
7) DFDSS BACKUP/DUMP
8)
HTH,
<snip>
We just converted them from z/OS 1.12 to z/OS 1.13 and have found that
performance has degraded.
They are looking at a list of data sets - approx 98,000 under a HLQ, say ABCD
with the Initial view set to 2. SPACE
On z/OS 1.12 this list was retrieved in under 10 seconds On z/OS 1.13 this list
took 20 Minutes to return.
If the Initial View is changed to 1. Volumes then the performance is much the
same.
The view is paged right to get the space values, on the 1.12 system it comes
back in seconds
on the 1.13 system it takes minutes to come back.
Many of the 98,000 data sets are on TAPE or migrated.
Anyone recognise a feature that came with z/OS 1.13 that might cause this
change in behaviour?
</snip>
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
