Bonnie++ for deskside JackRabbit

This is a 15 drive JackRabbit unit (under $6500 USD the way we have it configured), where we carved 2 drives out for OS, and built a RAID6 across 12 drives, with 1 hot spare. Just finished the other tests. Pretty pleased with the results. Still have to do driver and kernel updates, but I want a simple baseline test. So here it is.

root@crunch:~/jr# bonnie++ -d /big -u root -f
Using uid:0, gid:0.
Writing intelligently...done
Rewriting...	done
Reading intelligently...done
start 'em...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version 1.03b       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
crunch.scala 32152M           411990  60 186080  37           560387  55 466.2   0
                    ------Sequential Create------ --------Random Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 18714  94 +++++ +++ 19290  93 14966  74 +++++ +++ 18523  96
crunch.scalableinformatics.com,32152M,,,411990,60,186080,37,,,560387,55,466.2,0,16,18714,94,+++++,+++,19290,93,14966,74,+++++,+++,18523,96

This uses Ubuntu 8.04 … we should likely go back to 7.10 for our tests for apples to apples. The 2.6.24 kernel is known to have some performance regressions.

Viewed 8762 times by 2465 viewers

2 thoughts on “Bonnie++ for deskside JackRabbit

  1. As someone working in the area of HPC on Linux also, I’m curious as to how to keep up to date on news such as “The 2.6.24 kernel is known to have some performance regressions” – are you watching the LKML for this kind of news or are there other forums dedicated to performance testing recent Linux kernels – or have you got some dedicated staff testing new releases of the Linux kernel?

  2. @Stephen:

    This is due to dedicated testing of the 2.6.24 kernel on a system that behaves very well with the 2.6.23.14 kernel we support. We have seen some serious direct IO and buffer cache issues that we have reported on lkml. No responses from the relevant groups though. Linus is known to dislike the direct IO system.

    We are more hopeful on 2.6.25 and 2.6.26 kernels. .25 has performance tweaks and fixes for a number of the regressions in .24. Someone also was trying to push the 4k stacks in .25/.26 and thankfully this was removed.

Comments are closed.