Networker

Re: [Networker] Restore times excessive?

2003-01-07 16:58:02
Subject: Re: [Networker] Restore times excessive?
From: "Herrmann, Dave" <daveh AT DELTAFAUCET DOT COM>
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Date: Tue, 7 Jan 2003 16:55:20 -0500
Yes the file is fairly compressible, however turning compression on or off
has little affect the backup and restore times. Granted this is just one
file. But we're using it as a test because we are seeing excessively long
restore times when we do a disaster recovery test. And seeing this file take
literally 4 times longer to restore than to backup tells me something's not
right.

I performed a similar test on the same file copied over to my Windows 2000
based Networker server. The results on it was that the restore only took
about twice as long as the backup which is more in line with what we expect.
So the problem seems to reside only on my AIX based Storage Nodes.

Dave

-----Original Message-----
From: lemons_terry AT emc DOT com [mailto:lemons_terry AT emc DOT com]
Sent: Friday, December 13, 2002 11:28 PM
To: daveh AT DELTAFAUCET DOT COM
Cc: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: RE: [Networker] Restore times excessive?


Hi Dave

Here are some thoughts.  On the backup side, backing up 10 GByte in 10
minutes is 16.7 MB/s, which I think is pretty good performance.  I'm
assuming that your text file is fairly compressible (you can check that with
a Zip program, for instance).

I can't explain why the restores are taking so much longer than the backup.
Are you looking only at the time that the data is actually moving between
the tape and the disk?

Thanks
tl

-----Original Message-----
From: Herrmann, Dave [mailto:daveh AT DELTAFAUCET DOT COM]
Sent: Friday, December 13, 2002 1:09 PM
To: NETWORKER AT LISTMAIL.TEMPLE DOT EDU
Subject: [Networker] Restore times excessive?


I'm in the process of trying to optimize recovery speed for our backup
system and am not having much luck. I've got a Windows 2000 based NetWorker
server with a StorageTek L180 and 3 IBM LTO drives attached. There are also
3 AIX based storage nodes, each of which is SCSI attached to 1  IBM LTO
drive each in the L180. So there's a total of 6 drives in the library, 3
directly attached to my server and 1 each to the 3 storage nodes.

Here's the results of the tests I've ran the past couple days:

Backups were performed via a test group (testaix) that I started from the
Networker server. I configured the group to perform a clone upon completion
so I could test the recovery from both the backup as well as the clone. The
test file is a 10 gigabyte text file residing on the storage node that we're
performing the test from. The tape drive is configured for compression
(/dev/rmt2.5) and during this test had a blocksize configuration of zero. No
other backups or restores were running during the tests.

The backup took 10 minutes to perform
The recovery from the backup tape took 40 minutes to complete. The recovery
from the clone tape took 41 minutes to complete.

We then tried setting the blocksize to 256K and here are the results:

The backup took 9 minutes to perform
The recover from the backup tape took 40 minutes.

Not really much difference!

I also copied the file over to the Windows 2000 based Networker server and
tried backups and restores, again via a test group, with and without
compression for comparison purposes. Here's what I found:

Backup without compression took 7 minutes
Restore from this took 23 minutes.
Backup with compression took 14 minutes
Restore from this took 25 minutes.

It's curious to me that the Windows based backups were 3-4 minutes faster or
slower depending upon compression setting, but in both cases the restore
time on the windows based test was almost half that of the test on the AIX
storage node.

In all cases, I would think the restore should be able to be completed in no
more that twice the time it takes to back up. Has anyone got any guidelines
for what I should be looking at to try and get better results on the
restore?

David D. Herrmann
Novell Network and Backup System Administrator
Delta Faucet Co.
55 E. 111th St.
Indianapolis, IN 46280
Phone: 317-848-0641
Fax:     317-573-3479
Email:   <mailto:daveh AT deltafaucet DOT com> daveh AT deltafaucet DOT com

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 the system manager.
Please note that any views or opinions presented in this email are solely
those of the author and do not necessarily represent those of the company.
Finally, the recipient should check this email and any attachments for the
presence of viruses. The company accepts no liability for any damage caused
by any virus transmitted by this email.

--
Note: To sign off this list, send a "signoff" command via email to
listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can also view
and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
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 the system manager.
Please note that any views or opinions presented in this email are solely
those of the author and do not necessarily represent those of the company.
Finally, the recipient should check this email and any attachments for the
presence of viruses. The company accepts no liability for any damage caused
by any virus transmitted by this email.

--
Note: To sign off this list, send a "signoff networker" command via email
to listserv AT listmail.temple DOT edu or visit the list's Web site at
http://listmail.temple.edu/archives/networker.html where you can
also view and post messages to the list.
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=

<Prev in Thread] Current Thread [Next in Thread>