ADSM-L

Re: Read block size discrepancy between LV image and B/A client backups

2004-05-01 23:50:26
Subject: Re: Read block size discrepancy between LV image and B/A client backups
From: Orville Lantto <orville.lantto AT DATATREND DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Sat, 1 May 2004 22:46:04 -0500
Take a look at the vmtune parameters for read-ahead. (-r and -R).  These
can have a very large influence of file level throughput.  They can be
changed on the fly.  Watch with topas as you change them to see how far to
go.  The defaults are very low.  I have improved even old, slow SSA by
factors of 5 to 10 by bumping these from the default of 4 or 8 for minimum
read-ahead to 128, 2546, or even 512 blocks.  Such large read-ahead levels
will kill your random and/or small file access, so some compromise will be
necessary.  Other kernel parameters can be adjusted to improve throughput,
but read-ahead has the most immediate impact.

Orville L. Lantto
Datatrend Technologies, Inc.  (http://www.datatrend.com)
IBM Premier Business Partner
121 Cheshire Lane, Suite 700
Minnetonka, MN 55305
Email: Orville.Lantto AT datatrend DOT com

CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
for the sole use of the intended recipient(s) and may contain confidential
and privileged information. Any  unauthorized review, use, disclosure or
distribution is prohibited. If you are not the intended recipient, please
contact the sender by reply e-mail and destroy all copies of the original
message.




Paul Ripke <stix AT STIX.HOMEUNIX DOT NET>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
05/01/04 09:18 PM
Please respond to "ADSM: Dist Stor Manager"

        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Re: Read block size discrepancy between LV image
and B/A client backups


On Friday, Apr 30, 2004, at 05:33 Australia/Sydney, Ted Byrne wrote:

> AIX client v. 5.1.5.5
> AIX server v. 5.1.8.0
> AIX version 5.2.0.0
>
> We  have a customer who is experiencing discrepancies in performance
> between large-file cold backups of Oracle data and LV image backups,
> both
> sent over the SAN to LTO-2 tape drives.  On the lv image backups, they
> are
> getting about 70 MB/s sustained.  For the B/A client backups, they are
> getting less than 30 MB/s.
>
> <snip>
>
> I have a PMR open with IBM regarding this issue.  Is anyone aware of a
> way
> to control the block size used for reads by the TSM client?

To the best of my knowledge, you can't. AIX Filesystem I/Os, where
DIRECTIO is not involved, go through the AIX VM subsystem - and may
get split, joined and otherwise mangled on the way through. Some
of this is tuneable through /usr/samples/kernel/vmtune, although
much there is black magic.

I'm assuming DIRECTIO is not involved (as an AIX mount option)?

Just a hunch, but what do your filesystem parameters look like on the
client ("lsfs -q /blah")?

Cheers,
--
Paul Ripke
Unix/OpenVMS/TSM/DBA
I love deadlines. I like the whooshing sound they make as they fly by.
-- Douglas Adams

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