CentOS - Oct 2006 - server disk subsystem benchmarks, bonnie++ and/or others?

Greetings

I've searched to no avail so far... there is bound to be something more
intelligible out there...???

I am playing with bonnie++ for the first time...

May I please get some advise and list experience on using this or other disk
subsystem benchmark programs properly with or without a GUI ?

Test system in this case is a Compaq DL360 with 2 to 4 Gig DRAM and qty (2)
36Gig 10k drives in hardware raid1 (raid1 default 128k) and a smart array
5304 controller with 128 meg dram

The machine screams with this controller... yet I want to compare tests with
it to the standard built in smart array controller etc

Ive read bonnie++ stuff till I cant stand it no more and am just not seeing
the relevance in the end result vectors to something tangible.

Please advise and thanks in advance...

 - rh

--
Robert - Abba Communications
   Computer & Internet Services
 (509) 624-7159 - www.abbacomm.net

> 
> Ive read bonnie++ stuff till I cant stand it no more and am just not seeing
> the relevance in the end result vectors to something tangible.
> 
> Please advise and thanks in advance...
You have got this part right. bonnie++ was irrevelant...was because the 
guy has finally added an option to disable write buffering.

Try bonnie++ without write buffering...

However, that still does not make the results very useful...the best 
thing would be to simulate what your application would do to your disk 
subsystem and filesystem.

Here is an example of a good benchmark for those doing mail storage with 
the Maildir format:

http://untroubled.org/benchmarking/2004-04/

Complete with the tools needed for you to perform your own tests.

Email Lists wrote:
> May I please get some advise and list experience on using this or other
disk
> subsystem benchmark programs properly with or without a GUI ?
I've always used iozone to much satisfaction.  Just make sure your file
sizes are larger than your caching.  The automatic options will reveal
with the real boundaries are to your caching and then you can perform
custom tests from there.  Keep in mind to get good, relevant results,
you're looking at very large files that'll take awhile to complete.

<http://www.iozone.org>

-- 
Jason McCormick <jason at devrandom.org>

  "The object of life is not to be on the side of the majority,
  but to escape finding oneself in the ranks of the insane."
    - Marcus Aurelius, Emperor of Rome