Research On Load Balancing Algorithm Based On SDN Multi-controller Optimization Selection

Software-defined network(SDN)is an emerging network technology in the modern Internet.By decoupling the control plane from the data plane,the network is easily managed and evolved.The intelligence of SDN is reflected in the fact that a logically centralized controller manages the switch by providing the switch with rules that can determine its packet processing behavior.Due to the expansion of network scale,the scalability of centralized controllers has become a major problem for SDN.The deployment of distribution controllers is a promising way to deal with this problem.each controller manages some of the switches in the network.However,due to the static mapping between the controller and the switchs,this distributed architecture can cause uneven load distribution.Network load balancing affects the performance of the network,Therefore,it is particularly important to solve the problem of controller load balancing in distributed software-defined networks.The main research work done in this article is as follows:(1)Introduces the research background and significance of software-defined networking,describes the research status of SDN multi-controller and control plane load balancing in detail,elaborates the definition and architecture of SDN,introduces the architecture of controller and several commonly used controller,Briefly introduced the load balancing technology and its classification.(2)To solve the problem of uneven load distribution of controllers caused by the deployment of SDN multi-controllers,we proposes a game theory-based load balancing(GTLB)mechanism.The mechanism first builds a game domain model based on game theory,and then determines the target controller based on the time delay between the controller and the switch and the switch migration cost collection particle swarm optimization algorithm,and selects the migration switch based on the node distance and the flow request volume.The migration timer realizes the orderly and seamless migration of the switch.Experimental results show that GTLB reduces the total communication overhead of the network,reduces the flow establishment time,and the controller load is more balanced.(3)In the process of SDN multi-controller load balancing,it is prone to dull selection of migration objects and migration conflicts.In response to the above problems,we proposes a migration-optimized SDN load balancing algorithm(migration-optimized SDN load balancing,MOLB).The algorithm is divided into two stages.First,the objective function is set according to the controller load,and the migration domain and the migration domain of the switch are selected based on the particle swarm algorithm;then,on this basis,the subdomain migration degree and the switch validity period and period are calculated Manage the migration process of the switch in order to select the switch with a high number of requests to complete the dynamic migration of the switch.Experimental consequences show that the MOLB algorithm shortens the flow table establishment delay,reduces the switch migration communication overhead,improves the resource utilization by 14.4%,and enhances the network stability.