[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: 1.7 Performance via HTTP

From: Justin Erenkrantz <justin_at_erenkrantz.com>
Date: Thu, 12 May 2011 10:57:00 -0700

On Thu, May 12, 2011 at 7:34 AM, Mark Phippard <markphip_at_gmail.com> wrote:
> On Thu, May 12, 2011 at 10:31 AM, Justin Erenkrantz
> <justin_at_erenkrantz.com> wrote:
>> On Thu, May 12, 2011 at 7:23 AM, Mark Phippard <markphip_at_gmail.com> wrote:
>>> I re-ran the results with your latest changes and posted the new numbers.
>>
>> Is that with the commit as well as the patch I posted (but did not commit)?
>
> Just the commit.
>
> Wasn't the patch for mod_deflate?  And just to fix the memory leak
> when a client that does not support gzip connects?

Are you by chance using SSL?

I'm seeing something like a 2x perf drop on ra_serf with SSL to an SSL
server on the other side of the US from where I am now. But, over
HTTP, ra_serf is pretty close to ra_neon. (And, the server is 1.6 -
not 1.7.)

Basic tests:
| 1.7.0-dev (Serf HTTPS) | rNNNNNNNN | 1:05.224 | 0:24.992 | 0:17.997
| 0:00.047 | 0:00.057 | 0:39.892 | 0:00.074
| 1.7.0-dev (Neon HTTPS) | rNNNNNNNN | 0:21.495 | 0:14.227 | 0:04.769
| 0:00.046 | 0:00.057 | 0:07.892 | 0:00.073
| 1.7.0-dev (Serf HTTP) | rNNNNNNNN | 0:26.921 | 0:12.001 | 0:02.992 |
0:00.043 | 0:00.057 | 0:05.898 | 0:00.072
| 1.7.0-dev (Neon HTTP) | rNNNNNNNN | 0:21.076 | 0:11.001 | 0:03.999 |
0:00.044 | 0:00.059 | 0:08.895 | 0:00.069

I've got a hunch it has to do with OpenSSL buffering and memory use:

http://journal.paul.querna.org/articles/2011/04/05/openssl-memory-use/

I've been meaning to try zeroing out the compression bits in
OpenSSL... -- justin
Received on 2011-05-12 19:58:05 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.