Performance Testing

May 16th, 2008 | by bryan |

I spent some time last night going over several of the underlying systems that run ReadPath. Determining if the Apache setup was optimal. Making sure that ReadPath is using the latest version of MySQL. Also doing some rough load testing with JMeter.

It’s always nice to see that the blast type testing that JMeter is good at right out of the box shows that ReadPath is quite scalable. This type of testing isn’t really indicative of the real world, but it does take some fears out.

One of the things that I was looking at was I had read some reports stating that tomcat directly was significantly faster than using Apache as a load-balancing proxy. I can’t make broad statements for all cases, but I tested the main performance aspect for ReadPath, which is a lot of small requests for JSON objects. I tested hitting the url which proxies the request through apache and compared that to hitting the backend server directly. For my use case there was no significant difference between the two and both were able to scale to quite ridiculous levels.

While I’m sure that the system will hit bottlenecks at some point, I’m feeling much more confident that the core Apache-Tomcat configuration is not going to be the problem.


Post a Comment