I will demonstrate how I use VirtualBox, vagrant, ansible and Apache Jmeter to run reproducable performance tests of different versions of OwnCloud. At SURFsara we use this tool to measure performance of a new OwnCloud version before upgrading. In a nutshell, it starts with a level playing field every time, since installation of sources and populating the database are part of the process. We never go have to go back to old results and compare them with new results, that would not give a clear picture. Rather, we can spin up the environment afresh, be it on a laptop or a high-end server, and re-run the test suite for selected versions, giving a fair comparison. I will explain the concepts and tools used, give a live demonstration of a short test run, and will present results of longer test, that actually took several days to complete and consisted of millions of requests.