Joint Optimization of VM Placement and Rule Placement towards Energy Efficient Software-Defined Data Centers

Abstract

Energy consumption in data center has already become a widely concerned issue in cloud computing research. There are many studies on data center energy efficiency optimization. But most of them focus on the energy consumption of the servers and cooling equipment. Few of them considered data center network (DCN), which is also a significant energy consuming component. By applying newly emerged Software Defined Networking (SDN) into DCN, Software Defined DCN (SD-DCN) is proposed as a promising network management paradigm. As traffic flows between VMs occupy most of the data traffic in DCN, ingenious VM placement is not only significant to the server energy consumption, but also to the network energy consumption. In this paper, we are motivated to jointly consider VM placement and rule placement towards energy efficient SD-DCNs. Specially, the inherent TCAM size constraints of SDN are considered. The problem is first described using an Integer Linear Programming (ILP) model. To avoid the computing complexity of ILP, a two-phase heuristic algorithm is designed. Simulation results show that our algorithm significantly reduce the overall energy consumption.