Deployed web applications typically run on top of stacks of configurable technology (e.g. web servers, interface modules, software load balancers, databases). Each of these components often has dozens of tuneable parameters. How many times are those values typically tweaked before a final set of tuned parameters are settled on? What criteria are typically used to determine the optimal set?

This session presents a new method of automatically tuning a common LAMP stack for optimal performance. We explore a solution using some common Python-based automated deployment and load testing tools and dive into scientific computing with SciPy.