2012/9/18 Andriy Kornatskyy <[email protected]>: > > You can always apply caching to results returned from database... so you get > back to your web framework / template engine, etc. The question is how much > particular framework adds to serve a simple thing...
Agree, but the connection to the database framework also often adds something. Specifically template systems. So if the framework (for example) A needs 2 ms to return a simple "hello world", and the framework B needs 1 ms. This is where test "hello world" comes to us that B is 2 times faster than A. If we write in both real web application, it turns out that one query to the database needs 100 ms, and then the previous test will have no meaning. I do not deny the whole test. I want to point out that in the real application it is of little importance. It can be understood as an introduction to a more serious test. > > The only thing I found nginx burst the application performance is by using > keep-alive... that I am missing in uWSGI (of cause static files from from > nginx / CDN is always a good idea). Test "hello world" is a test of serving short static string. So fitting to compare it to a solution which is dedicated to this (Nginx, Varnish). uWSGI + Python IMHO is designed to generate dynamic pages and it is in this respect was to do the tests. I do not know if I expressed well, so I write for the sake of it again. The test is ok, but in real life does not apply. Per week (after updating one of framworków) everything can change things, even as the bottleneck is elsewhere. Tests were done on the local machine, so CDN is a totally different story. > > Another benchmark related to template engines is here (trivial big table > example plus real world application): > http://mindref.blogspot.com/2012/07/python-fastest-template.html > > P.S. I am planning contrast uWSGI/CPython2.7 vs Gunicorn/PyPy (can not get > uWSGI running on PyPy) somewhere later. Never too much testing, but it was nice to see those that bring a lot of interesting information and conclusions. For example, why one is faster than the other? > > Thanks. > > Andriy Kornatskyy > > > ---------------------------------------- >> Date: Tue, 18 Sep 2012 14:41:37 +0200 >> From: [email protected] >> To: [email protected] >> Subject: Re: [uWSGI] Trivial Benchmark of Python Web Frameworks in uWSGI >> >> I understand that the test "hello world" is such a stencil base, the >> starting point. But is not it better to do a decent test, for example >> a query to a database and display 10 results. Then soon it turns out >> that the results are significantly different. >> >> >> For this test, I would add the served from nginx a static page with options: >> >> open_file_cache max = 1000 inactive = 50s; >> open_file_cache_valid 60s; >> >> As a model serving from RAM. >> >> >> >> >> >> >> 2012/9/18 Andriy Kornatskyy <[email protected]>: >> > >> > I have run recently a benchmark of a trivial 'hello world' application for >> > various python web frameworks (bottle, >> > django, flask, pyramid, web.py, wheezy.web) hosted in uWSGI... you might >> > find it interesting: >> > >> > http://mindref.blogspot.com/2012/09/python-fastest-web-framework.html >> > >> > Comments or suggestions are welcome. >> > >> > Thanks. >> > >> > Andriy Kornatskyy >> >> >> -- >> Łukasz Wróblewski >> http://www.nri.pl/ - Nowoczesne Rozwiązania Internetowe >> http://www.hostowisko.pl/ - Profesjonalny i tani hosting >> http://www.katalog-polskich-firm.pl/ - Najlepszy darmowy katalog firm -- Łukasz Wróblewski http://www.nri.pl/ - Nowoczesne Rozwiązania Internetowe http://www.hostowisko.pl/ - Profesjonalny i tani hosting http://www.katalog-polskich-firm.pl/ - Najlepszy darmowy katalog firm _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
