A Measurement-based Characterization of the Energy Consumption in Data Center Servers

Increasing the energy efficiency of cloud data centers has become one of the hottest research topics. However, in order to properly increase the energy efficiency of a data center we need to understand how energy is consumed. In this work we present an exhaustive empirical characterization of the power requirements of multiple components of data center servers. To do so, we devise different experiments to stress these components, taking into account the multiple available frequencies and the fact that we are working with multicore servers. In these experiments, we measure their energy consumption and identify their optimal operational points. Our study proves that the curve defining the minimal CPU power utilization, as a function of the load in Active Cycles Per Second, is neither concave nor purely convex. Moreover, it definitively has a superlinear dependence on the load. Similarly, we present results on how to improve the efficiency of network and disks. Finally, we validate the accuracy of the model derived from our characterization. For this, we compare the real energy consumed by two Hadoop applications, PageRank and WordCount, both in a stand-alone or a cluster fashion, with the estimation from our model, obtaining an error below 4.1% on average.