Benchmarking KSE and SMPng

[Jesse Guardiani]

culverk at yumyumyum.org wrote:

> Quoting Roderick van Domburg <r.s.a.vandomburg at student.utwente.nl>:
> 
>> > Just in case you don't know, there is no 5-STABLE yet. All the 5.x
>> Releases have
>> > been based on -CURRENT.
>>
>> Indeed, I stand corrected.
>>
>> > > Question is then which {application,theoretical} benchmarks to run
>> > > under which circumstances. Although I've heard of bonnie et al, I'm
>> > > new to the benchmarking business and welcome any suggestions.
>> > >
>> > Something that might be interesting to try is running mysql with
>> > libc_r, linuxthreads, libkse, and libthr, using the included mysql
>> > benchmarks that
>> come
>> > with mysql. I used those benchmarks to test mysql on a few of the
>> > machines
>> I
>> > needed to install it on at one point, and I know mysql utilizes
>> > threads.
>> Off
>> > the top of my head I can't think of anything other than databases that
>> > performed badly on FreeBSD before due to threading issues.
>>
>> How about Apache2 or locked subsystems (I/O and networking spring to
>> mind)?
>>
> Good call. I hadn't thought of that because I've always just used the
> preforking mpm in apache.

I'm very interested in hearing what you find from your benchmarks.

I've been wondering how much better KSE currently is in CURRENT ever since
I started running 5.1-RELEASE on my laptop a few months ago. But I don't
have the time to do extensive benchmarking, and I don't have a spare machine
to run them on. (hey, I need my laptop for work.)

I've also heard that 5.x ships with added kernel debugging and such. If that's
true, it might be a good idea to turn that off (not sure how.).

Also, running similar benchmarks on a 4.8 system would make a good comparison.

-- 
Jesse Guardiani, Systems Administrator
WingNET Internet Services,
P.O. Box 2605 // Cleveland, TN 37320-2605
423-559-LINK (v)  423-559-5145 (f)
http://www.wingnet.net