Research On Scalability Of Software Defined Network

Software-Defined Network(SDN)is an emerging network architecture proposed by Stanford University.Software-Defined Network decouples the control plane from the data plane.The SDN controller logically manages the entire network,and generates a flow entry for the data flow based on special strategies.The SDN switch forwards network traffic according to the flow entries in the flow table.Now,Ternary Content Addressable Memory is widely used in flow table of the SDN physical.switch.But because of the level of technology and hardware costs,the capacity of Ternary Content Addressable Memory is limited.The physical SDN switch flow table’s capacity is difficult to meet the flow entities needed in the large-scale or high-load traffic SDN network,becoming one of the bottlenecks that limit the scale expansion of software-defined network.Paper studies how to improve the utilization of the physical switch flow table space in SDN data center network,which makes Software-Defined Network can be applied to the lager data center network.Based on the analysis of current researches which are related to the flow table capacity bottleneck problem,paper puts forward a scheme which models packet inter-arrival time and sets up the flow entity lifecycle value with the model parameters,in order to improve the SDN switch flow table utilization.The scheme is validated by experiment.Paper selects two experimental comparisons schemes,and measures the flow missed times and the flow dropped times in four experimental scenarios.Experimental results show that,in every experimental scenario,our scheme more efficiently uses the flow table capacity than experimental comparison schemes.Meanwhile,synthetically analyzing the four scenarios results can know our scheme can also be applied to the larger-scale or the higher-load traffic SDN data center network.