Rerearch On Consistency And Load Balancing On Software-defined Networking Control Plane

With the rise of new applications such as cloud computing and virtualization,the traditional TCP/IP network is difficult to adapt to the requirements of these applications on the network.Moreover,under the framework of the traditional network,it is more difficult to obtain innovation of the network.Under this background,the researchers proposed the concept of Software-Defined Networking(SDN).The core idea of SDN is to separate the control functions of the network from the traditional equipment and achieve centralized control.While SDN solves the problems of traditional networks,its centralized control will inevitably make the controller facing performance bottleneck of the system.However,the distributed control plane composed of multiple controllers to solve this problem will face the problem that how to effectively share the network view.In order to allow controllers to work more harmoniously in the SDN multicontroller control plane,it is necessary to ensure a high degree of consistency of the control plane.On the other hand,when the SDN multi-controller distributed cluster design is implemented,each controller may experience unbalanced load.In order to balance the load among the controller nodes and improve the load balancing degree and resource utilization of the system,the study of load balance of the SDN control plane is indispensable.This thesis focuses on SDN control plane consistency issues.The SDN control plane consistency related parameters are normalized by the number of events.It is found that if the SDN distributed control plane,the consistency parameters of each controller account for the proportion of the total consistency limit and its network When the square root of the event rate is equal to the sum of the square roots of the network event request rates of all controllers,the minimum communication overhead and maximum availability under the current system consistency limit are obtained.According to this relationship,this thesis designs a SDN control plane dynamic consistency algorithm based on the event rate.The algorithm determines the consistency parameters of the controller by the network event rate of each controller node.Each controller node is consistent.The ratio of the sex parameter to the consistency parameter of all controller nodes is equal to the ratio of the square root of the rate of occurrence of the network event of the controller to the sum of the square root of the rate of occurrence of the entire SDN network event.Through simulation analysis,compared with the existing algorithms,the proposed algorithm can effectively reduce the communication overhead of the control plane while ensuring the availability of the entire SDN control plane is as high as possible.The SDN control plane dynamic consistency algorithm effectively solves the problem of configuring the consistency parameters of the SDN control plane,and at the same time,it also faces the problem of unbalanced configuration of parameters caused by unbalanced load.For this problem,this thesis designs a SDN control plane based on the system balance degree.Multi-threshold load balancing algorithm.The algorithm first removes the repeated and useless parameters used by the controller to determine the load status,and then determines the number and value of the thresholds according to the actual situation,determines the controller load,and more accurately describes the various states of the controller,avoiding general The single-threshold,double-threshold decision method is too general about controller states.The algorithm also improves the selection of migrating switches and target controllers in the process of switch migration,and effectively improves the effect of switch migration.Simulation experiments show that compared with the existing algorithms,the proposed algorithm can effectively balance the load of each controller,improve the load balance of the control plane,and at the same time accelerate the overload recovery rate of the overload controller and improve the throughput recovery process.Average throughput.