*** Original Author: ADSM-L @ MARIST - ** Remote User **; 05/21/97 10:25am
>Message-ID: <ADSM-L%97052111205326 AT VM.MARIST DOT EDU>
>Date: Wed, 21 May 1997 10:10:42 -0500
>Reply-To: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
>Sender: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
>From: "Schumack, Wayne" <SchuW01 AT NORWEST DOT COM>
>Subject: Re: Diskpool configurations on MVS.
>To: ADSM-L AT VM.MARIST DOT EDU
>
Damon,
We currently have two DISKPOOLS defined, one backup and one archive. We
have 3 mod-3 devices defined for BACKUP and 1 mod-3 defined for ARCHIVE.
We have mapped the entire volume as one virtual volume.
We have experienced no DASD performance problem backing up approx 150
mixed nodes (340 file spaces). Some are large UNIX databases.
I insisted that our storage pool volumes be located on RAID-1
DASD (we are using EMC), both for performance (and especially)
reliability/availability.
Hope this helps.
> ----------
> From: Damon Taylor SMTP:dtaylor AT AMOCO DOT COM
> Sent: Wednesday, May 21, 1997 9:20 AM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Diskpool configurations on MVS.
>
>
> I've asked this question once before and received no response so
> I'm
> re-sending and begging this time as we are starting to see some
> performance problems.
>
> If you MVS ADSMers would be so kind as to enlighten me as to the
> following:
>
> 1. How many DISKPOOLS do you have configured?
>
> 2. How many virtual volumes do you have configured for each
> physical
> volume? What size?
>
> 3. How are your virtual volumes defined to the various DISKPOOLS.
>
> 4. Is anyone experiencing any performance or IO problems related
> to
> ADSM hits on the DASD? If so how are you dealing with them?
>
> We're desperately searching for some information so any light you
> can
> shed would be greatly appreciated. Thanks in advance.
>
> Damon Taylor
>
> Amoco Corp.
>
*** Comments From: STEWAJM - Stewart, Mike; 05/21/97 10:30am
We also have two disk pools. One backup and one archive.
We do very little archiving so that one is very small.
The backup pool is composed of six ADSM "volumes"/vsam files.
These six linear datasets are all full 3390-3 volumes.
This is for 1000 nodes or so.
Until recently we were running with only 3 3390-3 volumes.
But migration was kicking in at undesirable times. Also
I hope to improve co-location by allowing more client data to
collect on disk before it moves to tape by adding the additional
disk pool space.
These were on real 3390's for awhile, but are now on
a raid device. We've not really had any problems with
device/channel utilization. (Although when ADSM is in
the process of initializing his datasets he can do some
amazing things to device/channel utilization.)
I think basically our network link through our 3172-3 is
too slow to really drive the channels/devices very hard.
I have a performance problem open with IBM in another area
of ADSM. One suggestion they gave was to turn off
caching in the disk pools. They said there was a problem
where "dirty" pools can eventually develop causing excessive
IO/CPU usage. I wasn't happy at the prospect of turning
off caching, so instead they had me cycle through
deleting each disk volume to force ADSM to copy all the
disk pool data from one volume to another. This didn't
fix the problem I was having, but it may be applicable to
you.
|
