Re: NetWare/Oracle Restore
1996-05-13 08:31:22
Jerry,
What kind of processor is the Netware server running on - ?? We
recently did a disaster recovery test for our HPUX environment where we
restored some Oracle database backups. Running 2 restores concurrently
(a 1.1 GB and a 550 MB file) with compression enabled (like you, we were
seeing > 10 to 1 compression ratio), each restore process used 70% CPU
on a 2-way HP I70 (with 512 MB RAM). So... judging from the numbers we
observed on a pretty sizable machine, the processor utilization you were
seeing doesn't sound unreasonable.
Tim Pittson
tpittson AT himail.hcc DOT com
>----------
>From: Jerry Lawson[SMTP:jlawson AT ITTHARTFORD DOT COM]
>Sent: Monday, May 13, 1996 8:18 AM
>To: Multiple recipients of list ADSM-L
>Subject: NetWare/Oracle Restore
>
>Date: May 13, 1996 Time: 07:22
>From: Jerry Lawson
> ITT Hartford Insurance Group
> (203) 547-2960 jlawson AT itthartford DOT com
>------------------------------------------------------------------------
>-----
>Last Friday afternoon we had a decent sized recovery of an Oracle Data
>Base on
>a NetWare server. The major part of the restore was 17 files that had
>a
>combined size exceeding 600MB; the recovery time was approximately 1hr
>20 min.
> The bad news was that we discovered (after the first time we restored
>the
>files) that the backups were taking place while the data bases were
>online!
>This led to some interesting situations, including what appeared to be
>a
>skipped log backup. We were, however, able to get a set of files that
>matched
> and were usable once restored.
>
>There were two things about the restore that bothered me.
>
>1. From the beginning of the restore process, processor utilization
>hit 100%,
>and stayed there pretty consistently. The server had approximately 50
>active
>users (seen in the monitor screen). All of them complained of slow to
>no
>response. Fortunately, it was late Friday and they all wanted to go
>home
>early anyway. I was perplexed by the high utilization - the NetWare
>machine
>is a 3.12 machine, and was running version 2.1.0.2 of ADSM. I believe
>we have
>Processorutilization set at 25, so I was at a loss to explain why no
>one could
>gain access to the server. Is this value too high for a restore? I
>should
>add that there is a fair amount of compression on the files - the
>actual
>amount of data transferred was 46M to get 600M restored.
>
>2. There was a long delay at the end of the restore of several files -
>sometimes as much as 7 minutes. What we observed was the restore was
>proceeding - we were getting a string of dots across the screen - and
>then the
>process stopped - no more dots. Switching to the monitor showed
>continued
>high utilization. I got on the ADSM Admin, and a Q Session command
>showed the
>session in an Idle Wait condition. (There was no media waits - all of
>the
>files were on a single volume). This occurred most notable after
>several
>large files (100MB), but also on some smaller ones as well. Any ideas
>on what
>was happening?
>
>Jerry Lawson
>jlawson AT itthartford DOT com
>
>
>************************************************************************
>*****
>Jerry Lawson
>ITT Hartford Insurance Group
>jlawson AT itthartford DOT com
>
|
|
|