This repository is created for showing that H2O's mruby handler is not the bottleneck of the whole system. I assumed that H2O and nginx works as a reverse-proxy which delegate requests to an upstream app (app/app.psgi), and what they only do is some (little-bit complicated) Access Control.
Before running the benchmark, I supposed that the performance of H2O's Access Control itself would be worse than nginx due to it's mruby overhead, but the result was not what I expected..
https://github.com/i110/tinytodo/blob/master/conf/h2o.conf
https://github.com/i110/tinytodo/blob/master/conf/nginx.conf
wrk -t36 -c180 -d10 -s wrk.lua http://$REVPROXY:5001/