Dynamic configuration of web server clusters with QoS control

To make data centers ideally sized, turning off servers and tuning speeds for each server are essential for energy-efficient service. We model the problem of selecting the servers that will be on and what are their speeds as mixed integer programming; we also show how to combine such solutions with control theory. For proof of concept, we im-plemented this dynamic configuration scheme in a real web server cluster with soft real-time requirements and QoS con-trol, that guarantees both energy-efficiency and good user experience. In this cluster, we show the good performance of our scheme, a comparison of centralized and a distributed approach for QoS control, and a comparison of two schemes for choosing speeds of servers. 1