ADSM-L

Restore performance

1998-10-22 18:42:04
Subject: Restore performance
From: Julie Phinney <jphinney AT HUMANA DOT COM>
Date: Thu, 22 Oct 1998 17:42:04 -0500
We have been trying to do a restore of a directory on a Nov 4.11 server.
There are  probably around 300,000 files in that directory.  Almost all are
there and don't need to be restored.  The ADSM client is 3.1.0.3 and the
server is MVS ADSM 3.1.2.0.  The connection is FastEthernet 100mbs to an
OSA/2 adapter on the mainframe... but bandwidth isn't the problem, there's
nothing to restore anyway.  We are doing   restore rep=all -IFN.   It shows
skipping files  and continues skipping for 20 hours  then we have to stop
for one reason or another, we haven't been able to get through it yet this
week.  It turns out, you can't use restart restore with -IFN.  IBM is
opening an APAR on that.
My question is, when you do an incremental restore like that, on a server
with that many files, do you see that kind of performance?
I would like to know, when you've done a restore with REP=ALL and  -IFN,
and the number of files was in the 100,000s what have you seen for restore
times?  And what is your server platform?  I'm interested in any that are
like mine and any that are not.  The repercussions are going to be such
that I may have to recommend a different server platform or a different
solution.
Mgmt's idea was to use ArcServe on the weekends, then ADSM during the week.
Then in a disaster, use Arcserve for the big restore, the ADSM to just
bring it up to current with -REP=ALL and -IFN.   But that unacceptably slow
on big clients.  Or is it my server platform?  I'd really like to know
others' experiences  with that type of restore.  One of your avg big ones,
the avg time and your platform.  Thanks to any who respond!  You can
respond directly..  I appreciate your time,
Julie
JPHINNEY AT HUMANA DOT COM
<Prev in Thread] Current Thread [Next in Thread>
  • Restore performance, Julie Phinney <=