[Linux-cluster] GFS performance

[Kamal Jain]

Hello and Happy Holidays!

This is sort of a follow-up to a posting I sent last week about GFS and Cluster Suite performance and stability.  I ran some throughput (kBps and IOPS) tests on a server running RHEL4.5 with RHCS and GFS.  The goal was to complete a brief set of tests to show some basic performance differences between GFS, direct iSCSI (using EXT3 on the LUN) and EXT3 on a local SAS disk.

There weren't many test runs so certainly there's room for error and differences, but in general, the ~35% and ~60% performance degradation on GFS versus the local disk did manifest itself in some runs we did with our own applications, in places where we were reading or writing heavily.

Does this performance difference make sense to you?  The iSCSI-direct was on the same EqualLogic appliance as the GFS volumes were on, so no change in the storage array or GbE switch.  It was also the same server in all the tests.  Local disk was JBOD on Dell PERC5/i.

- K


[cid:image005.png at 01C847C4.02D477E0]


[cid:image006.png at 01C847C4.02D477E0]



--
Kamal Jain
kjain at aurarianetworks.com
+1 978.893.1098  (office)
+1 978.726.7098  (mobile)

Auraria Networks, Inc.
85 Swanson Road, Suite 120
Boxborough, MA  01719
USA

www.aurarianetworks.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20071226/8b4bc9a3/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image005.png
Type: image/png
Size: 70363 bytes
Desc: image005.png
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20071226/8b4bc9a3/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image006.png
Type: image/png
Size: 77296 bytes
Desc: image006.png
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20071226/8b4bc9a3/attachment-0001.png>