Research On Energy Efficiency Optimization Traffic Scheduling Algorithms For SDN Data Center Network

With the rapid upgrading of computer network and communication technology,the Internet industry has ushered in a vigorous development.Clients have more demands and restrictions on network flows,and the size of data centers has also been explosive growth.As the data center networks mostly adopt the redundant Fat-Tree network structure,the problem of high energy consumption is urgent to be solved.Improving the energy efficiency of the data center and effectively realizing the energy conservation of the topology has become a hot and important research topic in the recent years.The traditional data center network has problems such as poor scalability,lack of a global perspective,and inability to centralize management.The software defined network technology effectively solves these problems by decoupling the forwarding function and control function of the switch,and achieve centralized traffic scheduling and routing management from a global perspective.Therefore,this thesis proposes the data center network energy saving traffic scheduling mechanism based on SDN,which can ensure the basic quality of service requirements of the network flow and achieve the energy saving effect of improving energy efficiency.The main work of this thesis is as follows:(1)Study and classify the SDN based data center network energy saving methods that have been proposed in recent years,compare and analyze the energy saving effects of different energy saving schemes,compare their advantages and disadvantages,and summarize the feasibility and application prospects of equipment-level energy saving technology based on sleep strategy.(2)An energy-saving traffic scheduling mechanism IEMV-MLF is proposed,which integrates multi-layer virtual topology and the path priority method with the highest average link utilization.A mathematical model is established with the goal of minimizing energy consumption and flow conservation constraints.The original multi-layer virtual topology method is improved,and parameters and logic are optimized.Based on the entire layer control,the path with higher average link utilization is selected as the forwarding path for network flow within the layer to achieve fine-grained energy-saving routing management.This mechanism further saves energy while effectively avoiding frequent changes in the virtual hierarchy of network topology.(3)An energy-saving traffic scheduling algorithm PSO-SDN based on particle swarm optimization is proposed.To solve the problem that the energy saving control mode of the IEMV-MLF algorithm is not flexible enough,this thesis uses particle swarm optimization instead of hierarchical control method.The PSO-SDN algorithm firstly selects some paths in the shortest path set for the network flow as the initial particle population,and then updates and iterates the particles by redefining the fitness function,thus obtaining the global optimal path.At the same time,this method sets an overload link set,which can prevent link congestion.(4)The simulation experiment environment is built by using Ryu controller,Mininet platform and s Flow monitoring software.The energy efficiency of the two energy-saving flow scheduling methods proposed in this thesis is measured by three experimental indicators,namely,the energy consumption ratio,the maximum link utilization rate and the average flow completion time.The energy saving effect of the proposed methods is verified by comparing with other existing energy saving methods,and further energy saving is achieved.