You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Chuck Murcko <ch...@n2k.com> on 1997/01/17 20:11:44 UTC

Re: server comparison (fwd)

sameer liltingly intones:
> 
> fyi.. It's likely that zeusbench is optimized for zeus, but the 1.2b2
> vs 1.1.1 slowness is worrisome. thoughts? perhaps there are some
> buffer copies that can be optimized away. I'll look through the code
> to see if I can find anything to optimize away.
> 	(Would an optimization belong in 1.2? I would like to put it
> in, if I find anything, but that does sound too much like a feature.)
> 
> ----- Forwarded message from Renee Walker -----
> 
> Using Zeusbench as his comparison tool: 20k byte files, 400 times with 40
> simultaneous connections+keep alive he found: 1.1.1=2700kps
>                                               1.2b2=1700kps
>                                               Zeus =8000kps
> 
> 
> ----- End of forwarded message from Renee Walker -----
> 
I would contend that this degradation in performance is a bug, and would
be a bug fix, not a feature.

I did mention this should probably be done (profile/optimize/tune) some
weeks ago.

chuck
Chuck Murcko	N2K Inc.	Wayne PA	chuck@telebase.com
And now, on a lighter note:
Newton's Fourth Law:  Every action has an equal and opposite satisfaction.

Re: server comparison (fwd)

Posted by ra...@mail1.bellglobal.com.
> I would contend that this degradation in performance is a bug, and would
> be a bug fix, not a feature.

If performance has been cut in half from 1.1.1 to 1.2, then it is a
showstopper problem.  We should hold everything and concentrate all efforts
on evaluating performance issues.

I will try to do some performance testing this weekend.

-Rasmus

Re: server comparison (fwd)

Posted by Brian Behlendorf <br...@organic.com>.
On Fri, 17 Jan 1997, Chuck Murcko wrote:
> > fyi.. It's likely that zeusbench is optimized for zeus, but the 1.2b2
> > vs 1.1.1 slowness is worrisome. thoughts? perhaps there are some
> > buffer copies that can be optimized away. I'll look through the code
> > to see if I can find anything to optimize away.
>
> I would contend that this degradation in performance is a bug, and would
> be a bug fix, not a feature.

Hmm, are these pages being served SSI's?  Remember that we did notice that our
new SSI code is a fair bit slower and could be optimized.  

mmap() instead of fopen(), anyone?

	Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com  www.apache.org  hyperreal.com  http://www.organic.com/JOBS