Tom Smith <smith@cag.zko.hp.com> wrote:
> Rick Jones wrote:
>>
>> Step one - eliminate variables. Run a bandwidth test that does not
>> involve a filesystem - eg netperf - http://www.netperf.org/ - and
>> the TCP_STREAM test.

> And do the same with your filesystem(s)/disks(s). Run a bandwidth
> test that does not involve either a network or another filesystem.
> For example, dt (
> http://home.comcast.net/~SCSIguy/SCS..._Tools/dt.html )
> will tell you the write bandwidth with a test that takes only a few
> seconds. For read bandwidth tests, if you are using buffered IO, you
> have to remember to umount and remount the filesystem before the
> test to flush the cache (on most OSs).

A very good point. I'll mention http://www.iozone.org/ as another
possible benchmark.

rick jones


--
portable adj, code that compiles under more than one compiler
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...