A Non-Cooperative Data Center Energy Consumption Optimization Strategy Based on SDN Structure

Abstract

Aiming at the current high energy consumption problem of the Internet of Things data center, and the static state of the traditional data center network architecture is not convenient to the design and deployment of energy consumption optimization strategies, this paper specifically proposes a new energy consumption optimization strategy under the Software Defined Network (SDN) architecture, the control plane is separated from the data plane. The control plane is divided into two layers, bottom layer and top layer. The bottom layer controllers are uniformly scheduled by the upper layer controller. In order to reduce the computational pressure of the upper-level controller and make full use of the computing power of the bottom level controller, the relationship between the bottom layer controllers in this article is non-cooperative. On this basis, two energy efficiency strategies are designed in this paper. The first one is Strategy For Nash Equilibrium Point (SFNEP). The second one is Strategy For Energy Optimal(SFEO). The target of SFNEP is to search for the Nash Equilibrium Point. The target of SFEO is to gain energy efficiency via running SFNEP. Through extensive simulations, we show that SFEO has a better energy consumption optimization effect than other benchmark strategies. Under ideal condition, energy saving ratio is able to reach 24%.