What I have gleaned from *SM studies is that you should have enough disk
cache to hold one day's worth of backups. Most installations will backup AND
migrate important data. But this is an installation's decision. In doing so,
it may be beneficial to backup the disk storage pool before migrating it.
Otherwise, you have to copy from tape to tape rather than from disk to tape.
(Of course, even after migrating one must backup the tape to tape pool to be
sure you get everything!)
Any optimum diskpool size will depend on how much data you normally receive
on any given day. With 700 NT servers, you have to do some math and come up
with a number to suit your needs. I think 45GB is a very low number. At
our installation, we receive much more data than that from fewer NT servers.
Again, it depends on the nature of the data and how much is backed up each
day. Have a look at that and I think you will find a solution that fits you.
And don't be too willing to give back the disk cache! You may need more for
the OS2 boxes!!! :-)
Regards,
John G. Talafous IS Technical Principal
The Timken Company Phone: (330)-471-3390
P.O. Box 6927 Fax : (330)-471-4034
1835 Dueber Ave. S.W.
Canton, Ohio USA 44706-0927
[EMAIL PROTECTED] http://www.timken.com/
-----Original Message-----
From: Gerard M Hanway [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, October 24, 2000 10:54 AM
To: [EMAIL PROTECTED]
Subject: Optimum diskpool size
I look after the companies TSM environment. Approx 700 servers back up an
average of 60-70gb per night. The ADSM server runs on OS/390 with 500 OS/2
and
200 NT clients. I have 2 primary disk pools, one of 45GB for the NT
servers(colocated) and one of 4GB for the OS/2 servers(non colocated). HIMIG
is
set to 80% on the large pool and usually results is migration starting
between 2
and 3am each morning and completing around 9am.
However the storage people have just been in touch wanting to know what I am
doing with so much disc in the large pool. Basically they want 30gb back
leaving
leaving me with just 15gb. I suspect that things should continue to work
just
with migration kicking it a lot earlier however I would welcome opinions.
Is there a recommended ratio of disk pool size to amount of data backed up.
What
ratio do most people use. Are there quantifable performance impacts in
running
migration at the same time as backups. I do know from experience that with
migration starting significantly earlier I can expect a lot of additional
tape
mounts. But it might be argued so what.
Are there any other gotcha's
Gerard
**********************************************************************
This document is strictly confidential and is intended for use by
the addressee unless otherwise indicated.
Allied Irish Banks
**********************************************************************
