ADSM-L

Successes with Solaris/Veritas disk performance with TSM

2005-02-04 10:06:40
Subject: Successes with Solaris/Veritas disk performance with TSM
From: David McClelland <David.McClelland AT REUTERS DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 4 Feb 2005 15:05:46 +0000
Guys,

Not so much a question, but some sharing of my experiences with using
TSM on Solaris with Veritas Volume Manager - I hope these may help some
others who might be having similar experiences, and perhaps there might
be some improvements on what we've been doing:

Background are various SAN attached TSM Servers (versions 5.1.6.7 -
5.2.3.3) running on Solaris with Veritas Volume Manager and Veritas
Cluster Services on Sun V480/V440 hardware. *Very* poor performance when
backing-up/archiving from clients to our SAN diskpool - even on an
uncontended 100Mb link, we were seeing only between 1.5MB/s to 3MB/s
throughput to disk. 

So, performance troubleshooting time. When pointing the client to a tape
pool instead (LTO2), the data flew through at the full 11MB/s (i.e. the
100Mb/s LAN was my bottleneck). When creating a temporary disk stgpool
in /tmp on the TSM server, the data flew in as well at 11MB/s (/tmp is a
memory area on Solaris, virtual disk), so it wasn't TSM writing to any
old disk that was the problem. Writing to a local disk (/opt) and not
our SAN disk, we were still seeing 1.5MB/s, so it didn't appear to be a
SAN/FC/HBA related issue either. Finally, FTP from client to TSM server
was consistently rating at the full 11MB/s over the LAN, which suggested
that it was *something* to do with the way that TSM was interacting with
the disk layer, rather than general slow disk performance.

Anyway, a little Veritas Volume Manager tuning followed, and the
following settings were applied:

vxtunefs -s  -o discovered_direct_iosz=512 <mountpoint> (e.g. vxtunefs
-s -o discovered_direct_iosz=512 /stgpool/tsma)

Our 'discovered_direct_iosz' was previously around 256000. These were
applied (after some trial and error and help from a VxVM-man here), and
our disk write performance has picked up no-end (from a local client,
backing up a file in /tmp, seeing 40MB/s plus instead of 1.5MB/s!), and
so I understand from our support guys, has everything else.

Bear in mind that, in order to ensure this vxtune takes effect between
restarts/failovers, a file called /etc/vx/tunefstab, which has the
following contents:

/dev/vx/dsk/tsmlog_itsma_dg/tsmdb_itsma_vol02
discovered_direct_iosz=512

for each filesystem which you want to apply this to.

I hope this helps someone out there - does anyone else have any
improvements on the above or experiences of similar tweaks they'd like
to share with the list? I'd like to try going with raw volumes next.

Rgds,

David McClelland
Tivoli Storage Manager Certified Consultant
Infrastructure Backup and Recovery Development
Shared Infrastructure Development
Reuters
85 Fleet Street
London EC4P 4AJ


-----------------------------------------------------------------
        Visit our Internet site at http://www.reuters.com

Get closer to the financial markets with Reuters Messaging - for more
information and to register, visit http://www.reuters.com/messaging

Any views expressed in this message are those of  the  individual
sender,  except  where  the sender specifically states them to be
the views of Reuters Ltd.

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