web2py benchmarks

[mdipierro]

I run this test for web2py on a Ubuntu VM Machine running on the
latest PoweMac (the VM machine use one core only). I used the built-in
cherrypy WSGI server.

ab -c 10 -t 60 http://localhost:8123/benchmarks

This is the controller action (applications/benchmark/controllers/
default.py)

session.forget()
def index():
return dict(message="hello world")

This is the view (applications/benchmark/views/default/index.html)

{{=message}}

This is the output:

Total transferred: 10801080 bytes
HTML transferred: 2500250 bytes
Requests per second: 1578.29 [#/sec] (mean)
Time per request: 6.336 [ms] (mean)
Time per request: 0.634 [ms] (mean, across all concurrent
requests)
Transfer rate: 332.93 [Kbytes/sec] received

[JorgeRpo]

this is good, but it doesnt tell much, unless it is compared to other
frameworks on your same machine.

Can you test django and others too?

On Oct 25, 1:40 pm, mdipierro <mdipie...@cs.depaul.edu> wrote:
> I run this test for web2py on a Ubuntu VM Machine running on the
> latest PoweMac (the VM machine use one core only). I used the built-in
> cherrypy WSGI server.
>

> ab -c 10 -t 60http://localhost:8123/benchmarks

[JorgeRpo]

I would be good haveing benchmarks of other frameworks from
php(codeigniter comes to mind) and ROR

that would be really interesting...

[yarko]

I think Massimo is spread thin, working on T2, T3, and conferencing
project, etc.

Is someone willing to step up to this? Plan, suggest. It would be
good to create a branch on launchpad in the web2py area for
profiling / performance tests, and add a Blueprint there with your
plans. If you want help registering a branch, I can help.

Regards,
Yarko

[mdipierro]

Yarko is right. Moreover I do not like to publish benchmarks since
everybody assumes I am biased. ;-)
I am willing to help with benchmarks.

Anyway by test was triggered by this: http://paul-m-jones.com/blog/?p=236

My test obviously run on a different machine but, nevertheless, it
gives an idea of how fast web2py is compared with php based mvc
frameworks.

Massimo

[achipa]

It's hard, if not impossible to do meaningful comparison of php and
python _framework_ speed since the simpler the app (and hello world is
as simple as it gets), the more the web server/interface tech will
influence the result. So, even with web2py on it's own, you will see
drastic differences depending on whether it's running as a CGI,
cherrypy, mod_wsgi, mod_python, can it be precompiled, etc. On the
other side, PHP uses technologies which you also would have to explore
to make a meaningful picture, like mod_php versus fastcgi, mmcache
style caching accelerators, etc. If you compared it to Django,
TurboGears et al, it would be a much more lifelike comparison since
they are 'the competition'. PHP frameworks might serve the same
purpose, but are technologically pretty far (thus if you compare PHP,
you might as well compare ASP, JSP and their kin).