I published on article on evaluating response time measurements for the September 2011 issue of “Javamagazin”. It is in German and available in print only (for reference, please follow the link http://it-republik.de/jaxenter/java-magazin-ausgaben/Enterprise-Javascript-000463.html).
Often we start performance improvements much too late in projects. This article explains how important it is to measure the response times of your software – already throughout development, and how you can create meaningful information out of the data gathered. In particular this means:
- Establishing clusters of services/functions with excellent, very good, good and bad performance
- Recognizing trends of potentially bad performance, i.e. services/functions that are performing well but show signs of performance issues
- I packed the entire approach into a six repeatable steps, which can – most importantly – also get automated if you wish – for example – to embed them into your continuous build and deployment process
Copyright © 2011 Michael Pichler
Good article in the JM. Now it is clearer for me how to interpret JMeter-results.
Thank you for the feedback, Ben! It is encouraging to know that it was of help to you. Regards, Michael