Research On SDN Control Plane Load Balancing Based On Migration Optimization

With the increasing scale of network users and the scale of data,today's traditional networks are already overwhelmed.The rapidly growing network equipment can no longer be completed with simple patching,which not only aggravates the complexity of network equipment and data redundancy,but also limits the growth of the network.The new SDN network architecture separates the data plane from the control plane,breaking the limitations of network patching when the network size growth,but at the same time a new load balancing problem have emerged.For the new load balancing problem,the common solution is to use the OpenFlow protocol to change the role of the master-slave controller of the switch.When the controller is overloaded,the next part of the overload controller is changed by master-slave control.The way the role is migrated to other light-load controllers to achieve load balancing.However,if multiple controllers are simultaneously migrated to one controller,the target controller may continue to be overloaded.Especially when the network traffic is large and multiple controllers are overloaded at the same time,it will be difficult to achieve load balancing quickly and efficiently.Therefore,how to adopt better migration mode and migration algorithm to improve migration efficiency has become a problem that must be solved in current load balancing of SDN switches,and it is also one of the important research directions of SDN control plane load balancing.In view of the above problems,the research of the thesis mainly has the following two aspects:1)Through analysis and research on the existing switch migration environment,the current RYU controller does not have a unified multi-controller exchange protocol,so the three-layers architecture multi-controller control system built during the switch migration process realizes multi-controller distribution through Zookeeper to management,and implement switch migration selection algorithm in Zookeeper,optimize the switch migration process under multiple controllers,improve the overall stability of multi-controller systems,and provide reliable simulation experiments for performance testing and simulation of subsequent switch migration algorithms.2)Through the analysis and research of the existing switch migration algorithm,an improved switch migration algorithm is proposed.The advantages of the ant colony algorithm and the genetic algorithm are combined,and the fusion algorithm is optimized in the switch migration process.Improve the efficiency of the algorithm,and finally apply to switch migration under multiple controllers,improve the efficiency and stability of switch migration under multiple controllers,and use RYU controller to compare and analyze its performance and other algorithms.