That site you reference is meaningless. Neither Django or web.py are web servers but application frameworks.
That is the same sort of flawed benchmarking that Tornado used to claim their web server was so much better when they first announced it. That is, they compared a hello world program written with a full stack web framework to a very basic hello world program written directly to low level web server API. The fairer comparison would have been basic WSGI hello world program to basic Tornado hello world program. Do that and you will find the results are very very different. Anyway, when will people learn that these simplistic benchmarks are meaningless. Your site would never run at 100% capacity and you would never want it to. The overheads are also never usually going to be in the hosting mechanism unless you really screw up its configuration, which admittedly many do. Instead the real bottlenecks are in your application and database usage. Ultimately these benchmark comparisons are little more than a pissing competition and you are just wasting your time. Use whatever solution you find fits your ability to manage it. If you find one solution too much for your level of knowledge then don't use it. Just because you cant get a solution to work doesn't mean there is something wrong with that solution. This though doesn't seem though to stop people rubbishing particular solutions even though reality is that the problem is their own incompetence and inability to get it running properly. Graham On Wednesday, December 29, 2010 7:28:20 AM UTC+11, David Marko wrote: > > U just tested tornado using others 'example' pages and its much, much > faster than rocket. It event doesnt generate failed requests comparing to > rocket. Can tornado be better/faster solution then Apache wsgi? I mean some > apache/nginx+reverse proxy & tornado+web2py? > > > https://github.com/fiorix/cyclone/wiki/benchmarks > > David >
