Temperature Aware Power Allocation: An Optimization Framework and Case Studies

In English In this paper we propose a general power model along with a versatile optimization methodology that can be applied to different applications for minimizing service delay while satisfying power budget in data centers. We test our methodology on two totally different applications from both cloud computing and enterprise scenarios. Our solution is novel in that it takes into account the dependencies of power consumption on temperature, attributed to temperature-induced changes in leakage current and fan speed. While this dependence is well-known, we are the first to consider it in the context of minimizing service delay. Accordingly, we implement our strategies with Hadoop, Tomcat and MySQL on a 40-node cluster. The experimental results show that our solution can not only limit the power consumption to the power budget but also achieves smaller delays against static solutions and temperature oblivious DVFS solutions.