This presentation gives you first-hand information how JMeter was used for testing web applications and web services at a large-scale data center consisting of altogether 300 applications servers and multiple database clusters. At the beginning JMeter and the testing approach is introduced before tackling the hard problems of setting up a scalable performance test infrastructure consisting of JMeter, Ant, Hudson and Git. During the course of the project a new reporting backend for JMeter was developed overcoming the limitations of the current XSLT approach because a SLA (Service Level Agreement) performance report was required based on huge JMeter result files (e.g. exceeding 2 gigabytes).