[Grok-dev] Grok raw performance

Mon May 11 07:28:10 EDT 2009

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hanno Schlichting wrote:
> Michael Haubenwallner wrote:
>> Raw performance comparison of grok and competing web frameworks
>> method: ab -n 1000 http://localhost:port/
>> system: ubuntu8.4, python2.6
>>
>> # grok (wsgi) == 100%, bigger numbers are better
>>
>> 0.8 ... grok (zopectl, zodb)
>> 1.0 ... grok (wsgi, zodb)
>> 2.3 ... zope2.12 (zodb)
>> 7.3 ... repoze (zodb)
>> 7.3 ... pylons (no db)
>> 8.6 ... repoze (no db)
> 
> This comparison is highly dependent on what the particular framework
> ships with as their "default page". Numbers are also highly dependent on
> the exact way that page is setup, for example using some templating
> language or just a browser view / controller returning a plain HTML
> string. To be fair you should also try to disable any kind of debug-mode
> or similar functionality to get numbers that matter for deployment.
> 
> Chris has some good instructions at http://www.plope.com/whatsitdoing2
> on how to get some halfway comparable numbers.
> 
> If I change the default persistent index_html page template from Zope
> 2.12 to be "<html><body><h1>Hello world</h1></body></html>" and render
> that in production mode, I get a consistent 500 req/sec vs. 320 on the
> default quick start page.
> 

You are right, things get even worse when using no template engines and
4 concurrent requests.
The Grok - Repoze (no db) relation drops from 1:8 to 1:12.

Regards
Michael

- --
http://blog.d2m.at
http://planetzope.org

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKCAvKl0uAvQJUKVYRAs12AJ9UAQshzHqpluzEbMXgCaKDKLkt/ACfWpss
LEjcEFYoU28Od1HdCqUfeHk=
=n46K
-----END PGP SIGNATURE-----