Łukasz, Great work! How does that compare to (in order get an idea of baseline):
1. Serving the same file by nginx only? 2. Let nginx balance between nodes (nginx->nodes)? Thanks. Andriy Kornatskyy ________________________________ > Date: Tue, 29 Jan 2013 00:46:50 +0100 > From: [email protected] > To: [email protected]; [email protected] > Subject: [uWSGI] FastRouter (basic) benchmarks - 1.4.4 vs 1.5 > > As promised I've did a simple FastRouter performance test with 1.4.4 > and current 1.5 dev branch. > This is not very scientific but numbers are stable and repeatable. > > My cluster consist of 4 backend nodes and 1 FastRouter with nginx in > front (nginx->fastrouter->nodes). All nodes are running on HP DL 360 G6 > (2x Xeon [email protected] and 32GB of ram). > > I've tested small static files performance to stress the FastRouter > itself as much as possible, not the app I'm running. > All tests were run using apache benchmark with this command line: > > $ ab -c <256-1000> -n 100000 -H "Host: my.app.com<http://my.app.com>" > http://10.0.0.99:8080/static/css/fullcalendar.css > > fullcalendar.css has size of 10699 bytes > > Results are available at https://gist.github.com/4660375 (email will > most likely eat table formatting) > Detailed logs are attached. > > tl;dr version - 1.5 improved FastRouter stability and performance, > there are no failed requests with 1.5 under high concurrency, max > response times are also lower. It's just got better > > Great work Roberto, 1.5 is looking really good, it received a lot of > small fixes and enhancements, i'm getting tempted to start using it on > production nodes ;) > > > -- > Łukasz Mierzwa > > _______________________________________________ uWSGI mailing list > [email protected] > http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi _______________________________________________ uWSGI mailing list [email protected] http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
