Research On Energy Saving Traffic Scheduling Of Data Center Network Based On SDN

With the continuous development of Internet of things,cloud computing and big data technology,the scale of data center is expanding,and its energy consumption is becoming larger and larger.The energy saving of data center network has become an important research field.Traditional data center network is complex,difficult to expand,and poor controllability.Based on SDN technology,this paper studies the energy-saving problem of SDN data center network and establishes a data center energy-saving scheme which integrates traffic prediction and multiple virtual topology networks.The contributions of this paper are as follows:1.The network traffic prediction algorithm of data center is established.K-means clustering algorithm is used to cluster the traffic data set,and then SVM is used to train the regression model to achieve the prediction of network traffic.2.Aiming at the problem of low energy utilization of "rich connection" data center network under low load,an energy=saving multi-layer virtual topology traffic scheduling algorithm(EMV-SDN)is proposed.Firstly,a mathematical model of integrated linear programming(ILP)for energy-saving traffic scheduling is established to minimize the network energy consumption under the premise of carrying all network loads.Then,an energy-saving multi-layer virtual topology traffic scheduling algorithm is proposed to solve the mathematical optimization model and get the energy-saving scheduling scheme of data flow.It can save energy and reduce network energy consumption by sleeping high-level virtual topology,switch and its port.Experimental results show that EMV-SDN is superior to ECMP(equal cost multi path routing)and Dijstra shortest path algorithm in terms of network energy consumption and average completion time of data flow.3.Integrate the traffic prediction algorithm and the multi-layer virtual topology traffic scheduling algorithm(EMV-SDN)to further improve the energy-saving effect and network performance.First,the data center network topology is divided into multi-layer virtual topology and the first layer virtual topology is opened;then,for the newly arrived data flow,according to the average link utilization of the current layer virtual topology and the predicted traffic bandwidth demand interval in the next period,whether to open or close the higher layer topology is decided.The experimental results show that EMV-SDN is better than ECMP(equal cost multi path)in terms of network energy consumption and average completion time of data flow Routing)and Dijstra shortest path algorithm can improve network stability without increasing network energy consumption compared with the multi-layer virtual topology traffic scheduling algorithm(EMV-SDN).