Harold Zbiegien wrote:
We have a performance problem I am trying to improve.
When we do our Disaster recovery tests we wind up do a large amount of
delete noscratch against our user catalogs. We basically uncatalog all
of our disk datasets leaving just the tape datasets in the catalogs.
The catalogs are restored along with full pack restores of certain
volumes. We then selectively restore our needed disk datasets, almost
always to different volsers than in production.
Well we generate the IDCAMS delete noscratch statements in about 20
seconds, but then running the acutal IDCAMS deletes takes upwards of
80 minutes to uncatalog 92,000 entries from our largest user catalog.
We delete them in alphabetical order.
Do you know of anyway I can SPEED up this process??
The job runs unconstrained. It pretty much is the only thing running
in the system, it (CATALOG address space) must be doing a huge amount
of IO.
I though perhaps deleteing things in reverse alphabetical order might
improve things but that is just a wild guess.
---------------------------------------------------------------------
While I can't speak for reverse order with respect to catalog entries,
it makes a HUGE difference in deleting PDS members.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html