Friday, July 30, 2010

Web Platform Benchmarks

SkyHi @ Friday, July 30, 2010
The following graph is in no way scientific. One day the engineers at UsenetBinaries.com just decided to author a "Hello, World" program on a variety of web development platforms to illustrate general relative performance. There are *many* more techniques for analyzing the performance of a web development platform then the one we used, and performance certainly isnt the only criteria when selecting a development platform either.

Obviously UsenetBinaries.com moves a tremendous amount of dynamic content daily, so we were interested in seeing how the different platforms stood up to what we were currently using. We posted our results here as a public service.

The benchmarks included mod_perl2, C Apache Modules, Static HTML, C, Perl, Python, Ruby, and Ruby On Rails CGI/FastCGI, and PHP.

All of the tests were run standalone on an unloaded webserver consisting of a 2.4 GHz Tyan Opteron 150,1 GB DDR 400 ECC RAM, running Linux 2.6.8.1-20mdk, 64-bit.

In all cases, we followed the most straightforward and canonical install instructions we could find, compiled 64-bit from source with gcc -O2 (3.4.1), and always implemented the 'Hello, World' as dynamic text.

Webserver was Apache 2.2.

The test simply consisted of running ab -c 1000 http://localhost/etc.html on the same server. We would do this several times to come up with a good representational number.

We did also play with multiple requests but the overall *relative* performance was similar.

You may also be interested in our new GCC vs ICC Compiler Benchmarks - Perl Executable Performance









REFERENCES
http://www.usenetbinaries.com/doc/Web_Platform_Benchmarks.html