Re: slow run -- MIDRANGE-L

Muralidhar,

1) Is it running in a separate batch memory pool? (WRKSBS and check whether
the second column of Qbatch is a number such as 5 which is a pool on the
WRKSYSSTS screen and appears to have jobs running in it. Sometimes the
system is shipped with these jobs running in pool 2 (*Base) and this is not
so efficient, they can be affected by other work such as ODBC.)

2) If batch jobs have been separated into pool 5 or whatever, is Expert
Cache turned on? It should be. If so, F11 on WRKSYSSTS will show *Calc next
to the pool. If it shows *Fixed, change it to *Calc, but only if you are
sure that all the work in that memory pool is the same type (i.e. all batch,
or all interactive.)

3) Have the files used by the batch program been getting bigger? They may
need purging. Or they may contain 'hard deleted' records, if so you need to
do a RGZPFM on them to speed things up.

4) Does the batch program use SQL to access the files? If so, you may need
indexes creating over the files. You can run the database monitor to find
out.

5) Which model of iSeries are you running on? On some models there is a
'governor' program that kicks in when the workload reaches the 'knee' of the
response time curve, and slows everything down using an interrupt called
CFINTxx - you can only see this in action on WRKSYSACT.
This is more likely to happen with interactive than batch though.

6) Have you done a RCLSTG recently? Maybe some data got corrupted.

7) Any duplicate records in the files? Sometimes the software doesn't handle
this well.

8) Sometimes you can speed up a batch job by removing logicals and
recreating them afterwards.

Those are some things to be going on with! One of my clients had a batch job
that was taking 18 hours, and now runs in 35 minutes....

good luck,

Clare

----- Original Message ----- 
From: "Muralidhar Narayana" <Muralidhar_Narayana@xxxxxxxxxxx>
To: <midrange-l@xxxxxxxxxxxx>
Sent: Friday, January 14, 2005 3:27 PM
Subject: slow run


> We are having issues with a batch which is running slow , it doesnt used
take more than 30min, now-a-days it is taking more time around 2-3hr.One
reason we found was CPU utilization is getting increased.But unsual time of
2to3hr is nt acceptable even if CPU increased.This runs in separate library
so no clash with other jobs running at same time, and the library it runs it
doesnt have data mirror or mimix  or journals.
> Some one has suggested to use iDoctor for iseries to trace the jobs
accurately&effectively.
> I would request the esteemed folks here to comment and share your
expertize.
>
> Thank you very much,
>
>
> -- 
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.
>
>