RE: mod_deflate plus Sander's patch benchmarks.

> From: Kevin Pilch-Bisson [mailto:kevin@pilch-bisson.net]
> Sent: 15 February 2002 16:21

> Just wanted to give a few details about svn with Sander's neon patch, and a
> server with mod_deflate.
>
> This was on a checkout of trunk rev 1270:
>
> Results from time:
>
> Normal
> real 5m44.375s
> user 1m10.340s
> sys 0m25.610s
>
> Deflated
> real 5m0.809s
> user 1m13.050s
> sys 0m23.680s
>
> That's a 12.6% improvement in time, and I don't think the network is my
> biggest bottleneck. :)

Some more timings from my system:

Checkout of trunk rev 1270:

Normal
real 7m31.412s
user 0m10.470s
sys 0m7.320s

Deflated
real 5m29.647s
user 0m10.580s
sys 0m6.950s

Update from rev 1270 to rev 1

Normal
real 2m16.380s
user 0m11.350s
sys 0m5.090s

Deflated
real 2m7.650s
user 0m11.400s
sys 0m4.990s

> Another interesting note is the size of the ethereal log files for the two
> different checkouts.
>
> kevin@pilchie:~/public_html [513]$ ll eth*
> -rw-r--r-- 1 kevin users 1.5M Feb 14 15:16 eth-deflate.og
> -rw-r--r-- 1 kevin users 13M Feb 15 10:05 eth-no-deflate.log
>
> Now purists (who examine the log), will see that eth-no-deflate does grab a
> few more files, but still this is a _HUGE_ savings.

I didn't capture sizes.

Sander

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:37:08 2006