Web Benchmark -- Perform a web benchmark

SYNOPSIS bin/wb.rb [options] http://www.yourdomain.com/

OPTIONS -h : Print this screen and exit -v : Be verbose -f : Perform a full test -a : Do not fetch assets -r requests : The number of requests each visitor should perform on average (default 3) -c visitors_count : The number of visitors to launch (default 5) -C seconds : The number of seconds to cool down in a full test (default 10) -s seconds : Sleep less then X seconds between requests. Specify times 100 (default 2000 [= 20 seconds]). Specify 0 if you don't want the visitors to sleep between requests.

VISITORS The Web Benchmark simulates visitors to your online resource. It creates the number of specified visitors and they click random links on your page.

Each visitor has a pre determined number of pages to visit around the number you have specified, where N = n [-|+] rand(n) [n = specified nr]

Each visitor has a seperate cache for assets and doesn't download an asset twice.

FULL TEST Wen performing a full test, the benchmark is conducted 3 times. First with only one visitor, to establish a base line.

Then with half the visitors, and finally with all the visitors.

You can set a cool down period between the tests. Between tests visitors and caches are destroyed

SETUP/INSTALL Run bundle install.

LICENSE This software is released under the MIT License. See LICENSE.txt for more details

BUGS/CAVEATS/TODO

• This DOES NOT keep an eye on the load of the server you are benchmarking. You should!

• Measured times are not fully accurate, althoug measurements are taken as close as possible to the requests, there is still code I don't control or own in there, but I think it can be neglected

• When you use S3 (or any other CDN) for images, they to are collected. Still need to decide if that is a good or a bad thing

• All the other caveats and need-to-knows for benchmarking websites apply

• TODO : Post form data - would be rather nice for testing applications...