Network aware job scheduling in green data centers

In the literature, there is a growing body of research that focuses only on server energy consumption while proposing energy-efficient solutions for data centers. However, energy consumption of a typical data center network constitutes 10-30% of total energy consumption. Therefore, it is paramount to consider network energy consumption for facilitating green data centers. In this paper, we analyze the impact of integrating network energy consumption in addition to server energy consumption to scheduling in data centers. We investigate potential energy savings that can be attained via putting idle servers and network elements in sleep mode, while satisfying the computation requirements of the incoming jobs. First, we formulate the problem as an optimization problem and, show that it is not in APX, i.e., there is no polynomial time constant-factor approximation scheme. Therefore, we propose two heuristics which take into account the energy consumption of both data center networking and computing elements for scheduling and routing decisions for the jobs. Our analyses show that considering the energy consumption of network elements in addition to the energy consumption of servers for job scheduling while allowing sleep mode for idle servers and network elements results in substantial energy savings. We observe that almost half of the servers and network elements can be put into sleep mode to save energy compared to always on approach, in certain cases.