Research On Load Balancing Of SDN Controller Based On Switch Migration

In recent years,with the adoption of SDN architecture by major network manufacturers,SDN has gradually become a hot research topic in the network field.Different from the traditional network,SDN separates the data forwarding and forwarding control in the network,and realizes centralized control of the network.Since the controller bears a large number of processing tasks in the SDN network,in a large-scale network,a multi-controller deployment mode is usually adopted to satisfy the processing request,but the multi-control deployment method also brings about a problem that the controller load is unbalanced.For the problem of controller load imbalance in SDN,the paper has done the following work:Firstly,a switch migration strategy based on improved fuzzy multi-objective particle swarm optimization(FMOPSO-DSMS)is proposed.By modeling the load balancing problem of SDN controller,it is transformed into the choice of switch migration strategy,and a multi-objective optimization model with controller cluster load balance,average transmission delay and switch migration overhead as optimization targets is constructed.For the multi-objective optimization model,the improved fuzzy multi-objective particle swarm optimization algorithm is used to solve the problem.In the process of solving,in order to improve the diversity of particle populations in external archives,the fuzzy multi-objective particle swarm optimization algorithm is improved,and an external file update strategy based on niche number is introduced.Secondly,a load balancing framework MCD based on three-layer structure of switch migration is designed.Load balancing of the entire controller cluster is managed by adding a controller management layer on top of the controller cluster.The load monitoring module is mainly responsible for collecting the load information of the controller cluster and determining whether to perform the switch migration.The decision module is responsible for calculating the optimal switch migration strategy by using the fuzzy multi-objective particle swarm algorithm,and the action module is responsible for executing the specific switch migration action..In order to avoid a large amount of communication overhead between the controllers during the switch migration process,a seamless migration mechanism for the load balancing framework is designed.Finally,through the SDN multi-controller network built by Mininet simulation platform and Floodlight controller,the improved FMOPSO,FMOPSO-DSMS and designed loadbalancing framework MCD are verified.The final experimental results show that the external file update strategy based on niche number It can better maintain the population diversity and uniform distribution performance of external files.Compared with DSMA and M-DSMA,the proposed FMOPSO-DSMS can effectively reduce the average number of transmission hops in the network while effectively implementing controller load balancing.The number of switches is reduced,the network transmission delay is reduced,and the cost of performing switch migration is reduced.The designed three-layer load balancing framework MCD can effectively implement controller cluster load balancing and improve network performance.