Re: Cut MVS CPU Utilization
2000-09-08 17:23:54
Darlene,
Yes, the option is active in my options file, also here is part of 'q opt' -
adsm> q opt
Server Option Option Setting Server Option Option Setting
----------------- -------------------- ----------------- --------------------
CommTimeOut 240 IdleTimeOut 60
CommTimeOut 240 IdleTimeOut 60
BufPoolSize 131072 LogPoolSize 1024
...
...
PDMConnect PDM TAPEIOBufs 9
After thinking more about your great results (I know in a later post
you said you lost the performance), I think it was due to your change in
TXNGROUPMAX. The default is 40 so your increased it by more than 6 times.
What this did is to make the aggregates 6 times as large, depending on the
client setting for TXNBYTELIMIT. So for similar backups before the restart and
after
you decreased the number of aggregates by 6 times. Think of aggregates
as physical files; ba stg, migration & move data work on aggregates.
After the restart you had 6 times fewer trips through the code to backup
the disk storage pool.
Perhaps your later lose of performance is because you backed up a lot of
client that have a low txnbytelimit. For maximum performance you
should force all the clients to use txnbytelimit=25600. You can force this
from the server with client option sets.
--
--------------------------
--------------------------
Bill Colwell
Bill Colwell
C. S. Draper Lab
Cambridge, Ma.
bcolwell AT draper DOT com
--------------------------
In <NY3cbe3b-5f0a0eb1 AT sficorp DOT com>, on 09/08/00
In <NY3cbe3b-5f0a0eb1 AT sficorp DOT com>, on 09/08/00
at 05:10 PM, Darlene E Wnukowski <Darlene.Wnukowski AT SFICORP DOT COM> said:
>Have you checked your TAPEIOBUFS line in your server option file? It comes set
>to 9 BUT commented out. I had to uncomment it. The PTF has nothing to do with
>performance. I only referenced it because it was the first place I saw that
>this option could be used for 3490's. I have no 3590's.
>Darlene E. Wnukowski
>Lead Systems Programmer
>Schreiber Foods, Inc.
>darlene AT sficorp DOT com
>920-455-6268
>------------------( Forwarded letter 1 follows )--------------------
>Date: Thu, 7 Sep 2000 13:12:00 -0400
>To: ADSM-L AT VM.MARIST DOT EDU
>From: Bill.Colwell[bcolwell]@DRAPER.COM
>Sender: owner-adsm-l AT vm.marist DOT edu
>Reply-To: ADSM.Dist.Stor.Manager[ADSM-L]@VM.MARIST.EDU
>Subject: Re: Cut MVS CPU Utilization
>John,
>TAPEIOBUFS came in with 3.7. All my storage pools are
>on stk 9840s but they are emulating 3490's. I have tapeiobufs set
>to 9 anyway; after the storage pools were in use, we
>got some 9840's in 3590 mode. I hoped that I could do the dbb
>to the 3590s and it would be faster but it wasn't.
>The only performance knobs I know of are bufpoolsize,
>TxnGroupMax, MoveBatchSize and MoveSizeThresh.
>Darlene's results are a mystery to me; I haven't seen the same
>performance improvement. PQ38991 doesn't have anything to do with it
>(unless she has an interim testfix for it)
>since there isn't a ptf yet which fixes it, and the text is vague about
>what exactly IBM plans to do.
>--
>--------------------------
>Bill Colwell
>C. S. Draper Lab
>Cambridge, Ma.
>bcolwell AT draper DOT com
>--------------------------
>In <5DA5491C2D46D311BACE00508B0AC0F1024D1597 AT ctnhemail02.corp.timken DOT
>com>, on 09/07/00
> at 01:12 PM, "Talafous, John G." <Talafous AT TIMKEN DOT COM> said:
>>Does anyone know when the TAPEIOBUFS parameter was introduced to *SM? I did
>>a search of my ADSM 3.1 documentation and didn't find this parameter. I see
>>the PTF on IBMLINK and it looks like, maybe, 3.7. Can anyone confirm? I
>>could really use some CPU relief on OS390 with ADSM 3.1.
>>John Talafous
>>Information Systems Technical Principal
>>Global Software Support - Data Management
>>telephone: (330)-471-3390
>>e-mail: talafous AT timken DOT com
>>http://www.ctnvm.inside.tkr/~talafous/
>>http://www.cis.corp.inside.tkr/networkstorage/
>>Ask me about free firewood!
>>-----Original Message-----
>>From: Darlene E Wnukowski [mailto:Darlene.Wnukowski AT SFICORP DOT COM]
>>Sent: Thursday, September 07, 2000 11:26 AM
>>To: ADSM-L AT vm.marist DOT edu
>>Subject: Cut MVS CPU Utilization
>>I have been able to cut the MVS TSM server's use of CPU from a high of 60%
>>of
>>a 7060-H50 to under 10% of a 7060-H50 during backup of tape storage pools. I
>>did it by changing two parameters in the server parameter file. One of them,
>>TXNGROUPMAX, is not the one that made the difference but I am including it
>>because I changed it as well to its maximum value of 256. The parameter that
>>I
>>believe had the greatest impact is TAPEIOBUFS. It is documented as being
>>used
>>for 3590 tapes but I found a PTF on IBMLink that implies it may also work
>>for
>>3480 and 3490 users. The PTF is PQ38991. I would have included it but I'm
>>not
>>sure how IBM would feel about that. We have only 3490's for ADSM use and I
>>set
>>TAPEIOBUFS to 9. As I said, it made a dramatic difference. I apologize if
>>this
>>is old news but I searched the ADSM listserve search site and found no other
>>reports on this. Let me know if you find this useful. Thank you.
>>Darlene E. Wnukowski
>>Lead Systems Programmer
>>Schreiber Foods, Inc.
>>darlene AT sficorp DOT com
>>920-455-6268
>>Darlene E. Wnukowski
>>Lead Systems Programmer
>>Schreiber Foods, Inc.
>>darlene AT sficorp DOT com
>>920-455-6268
>>**************************************************************
>>This email and any files transmitted with it are confidential
>>and intended solely for the use of the individual or entity
>>to whom they are addressed. If you have received this email
>>in error please notify Postmaster AT sficorp DOT com.
>>This footnote also confirms that this email message has been
>>swept by MIMEsweeper for the presence of computer viruses.
>>**************************************************************
>**************************************************************
>This email and any files transmitted with it are confidential
>and intended solely for the use of the individual or entity
>to whom they are addressed. If you have received this email
>in error please notify Postmaster AT sficorp DOT com.
>This footnote also confirms that this email message has been
>swept by MIMEsweeper for the presence of computer viruses.
>**************************************************************
|
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- Cut MVS CPU Utilization, Darlene E Wnukowski
- Re: Cut MVS CPU Utilization, John . G . Talafous [Talafous]
- Re: Cut MVS CPU Utilization, Talafous, John G.
- Re: Cut MVS CPU Utilization, Darlene E Wnukowski
- Re: Cut MVS CPU Utilization, Bill . Colwell [bcolwell]
- Re: Cut MVS CPU Utilization, Richard Sims
- Re: Cut MVS CPU Utilization, Darlene E Wnukowski
- Re: Cut MVS CPU Utilization,
Bill Colwell <=
- Re: Cut MVS CPU Utilization, Darlene E Wnukowski
- Cut MVS CPU Utilization, Darlene E Wnukowski
- Re: Cut MVS CPU Utilization, Bill . Colwell [bcolwell]
- Re: Cut MVS CPU Utilization, Darlene E Wnukowski
- Re: Cut MVS CPU Utilization, Darlene E Wnukowski
- Cut MVS CPU Utilization, Darlene E Wnukowski [mailto:Darlene.Wnukowski
|
|
|