Re: Casual benchmarking OS performace with OpenLDAP

[Quanah Gibson-Mount <quanah@symas.com>]

--On Wednesday, March 30, 2005 2:55 PM -0500 "James F. Hranicky" 
<jfh@cise.ufl.edu> wrote:

> On Wed, 30 Mar 2005 11:16:13 -0800
> "Matthew Hardin" <mhardin@symas.com> wrote:
>
> > Hi James,
> >
> > Take a look at http://www.symas.com/benchmark.shtml. We've spent some
> > time doing this in a more formal manner, and it may save you some time.
> > We've also provided the tuning information that helped us achieve the
> > results we did.
> >
> > Hope this helps...
>
> I've looked at that, and actually run DirectoryMark, but I wasn't getting
> results that seemed to make much sense. I got 40 ops/s at one point,
> changed the number of processes, and dropped to about 5 ops/s. I figured
> I was doing something wrong as that wasn't adding up to being able to
> retrieve 80000 LDAP records in 4 seconds.
>
> Checking the config, I see you had your servers configured like so:
>
> 	Setting 		CDS
> 	Database Cachesize 	384 MB
> 	Entry Cachesize 	10,000
> 	IDL Cachesize 		30,000
> 	Access Logging 		Disabled
>
> Is my config substantially different? I didn't think so. At any rate,
> the differences in the platforms still doesn't add up. I figured maybe
> there was some tweak I was missing.
>
> Does the config I posted look reasonable?

Hi Jim,

I also want to note that the hardware used to run the benchmark also 
greatly influences the results.  The server running DM must be powerful 
enough to actually  process the results from the OL server in a fast enough 
fashion for the benchmark to have any meaning.

Note: AMD-244 results will be up next week (beats everything else so far).

--Quanah

--
Quanah Gibson-Mount
Product Engineer
Symas Corporation
Packaged, certified, and supported LDAP solutions powered by OpenLDAP:
<http://www.symas.com>