EAWA: Energy-Aware Workload Assignment in Data Centers

One of the challenges that today's cloud computing infrastructures, and more specifically data centers, are struggling with is related to their energy consumption. Information technology (IT) equipment and cooling infrastructure are key parts of the total energy expenditure in a data center. A considerable amount of power is wasted due to workload management inefficiencies and the lack of coordination between cooling units and IT equipment. In this paper, server differences in terms of their cooling requirements and power consumption are taken into account for workload distribution. An optimal workload assignment problem that takes both server power consumption and thermal models into account is formulated. A simple low complexity algorithm is proposed. The algorithm not only assigns workload but it also adjusts the cooling unit set-point accordingly. Results show that the proposed algorithm can significantly reduce the total power consumed in a data center, in particular when compared to the uniform workload distribution algorithm.