ADSM-L

Re: OS390 SELFTUNEBUFPOOLSIZE yes/no?

2003-02-17 10:34:12
Subject: Re: OS390 SELFTUNEBUFPOOLSIZE yes/no?
From: "Seay, Paul" <seay_pd AT NAPTHEON DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Mon, 17 Feb 2003 10:12:13 -0500
I like to set it about 10% higher than the happy point and set it to yes on
AIX.  Then check it every once and a while to see if it needs adjusting.
The problem is you can create a lot of GETMAINs on MVS if you do not set it
high enough to begin with.  On MVS, you are probably best to set it to NO so
that you do not get unpredictable memory usage on the machine.  This is the
difference between using a dedicated machine for TSM and a general use
machine and needing to share with other workloads.  You have to tune TSM on
MVS like you would a TP monitor like CICS, IMS, etc.

Paul D. Seay, Jr.
Technical Specialist
Northrop Grumman Information Technology
757-688-8180


-----Original Message-----
From: MC Matt Cooper (2838) [mailto:Matt.Cooper AT AMGREETINGS DOT COM]
Sent: Monday, February 17, 2003 9:28 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: OS390 SELFTUNEBUFPOOLSIZE yes/no?


Hello all,
        I have been reading the dialog on OS/390 performance tuning.  I too
have found that lowering the size of the address space to 512MB has helped.
I have also seen improvements in my throughput by cycling TSM.  (I just
don't do it as often.)   One thing that I was wondering is if anyone has
done any research on an advantage to NOT USING SELFTUNBUFPOOLSIZE.  Right
now I set BUFPOOLSIZE to 32760 and SELFTUNEBUFPOOLSIZE yes.   From the looks
of things TSM seems to be able to cause some thrashing with MVS memory
management.  SO I wonder if SELFTUNEBUFPOOLSIZE should be set to NO and just
allocate a bigger fixed BUFPOOLSIZE, (like the 128M that was suggested)?
Matt

<Prev in Thread] Current Thread [Next in Thread>