kevin beckford dijo [Mon, May 18, 2009 at 11:16:48AM -0700]:
> > My point was that a 1.6k static file really isn't what makes up a typical
> > website - so why would you use it to benchmark?
> 
> Agreed.  Something simple like a django/rails/catalyst/ app to , i don't
> know, pull out the mailing list archives or something would be a better
> benchmark.
> 
> Of course benchmarks are not really that important, cherokee's admin is why
> i'm using it really.  It's a (bit) easier to use than nginx right now, and
> that is the thing for me.

Still, benchmarking such an application places the bottleneck at the
wrong spot - In my (somewhat limited) experience, you won't have much
difference between Cherokee and other web servers on simple setups
handling a Rails application - as Cherokee will be limited to two
functions:

- Serving static content (everything inside public/). This should be
  seen as what has already been benchmarked elsewhere.
- Serving dynamic content. The bottleneck is your application server
  (i.e. Mongrel running Rails). The real, main application.

Of course, you can set up a big Mongrel cluster - some tens of
processes, having Cherokee round-robin between them. But... That will
be misleading IMHO as a Cherokee benchmark - You will be measuring the
throughput of a gang of Mongrels. You could as well just measure
Cherokee as a proxy for any content. 

Greetings,

-- 
Gunnar Wolf - [email protected] - (+52-55)5623-0154 / 1451-2244
PGP key 1024D/8BB527AF 2001-10-23
Fingerprint: 0C79 D2D1 2C4E 9CE4 5973  F800 D80E F35A 8BB5 27AF
_______________________________________________
Cherokee mailing list
[email protected]
http://lists.octality.com/listinfo/cherokee

Reply via email to