| As a new centralized control network architecture,Software Defined Network solves the problems of high coupling and rigid design of traditional network architecture.However,with the continuous expansion of network scale and the increase of traffic demand,the single controller is faced with scalability and reliability problems.Therefore,the industry has proposed a logically centralized and physically distributed multi controller network architecture.However,the unreasonable deployment location of multiple controllers and the fluctuation of traffic during network operation will lead to the load imbalance between controllers,which will reduce the network performance and affect the user experience.This paper focuses on the static deployment and dynamic load balancing of SDN multi controllers.The research contents are as follows:Aiming at the static deployment of SDN multi controllers,this paper comprehensively considers the control plane delay and the load balance between controllers,and proposes a multi controller deployment based on delay and load strategy.The strategy first introduces the shortest propagation delay matrix into the clustering by Fast Search and Find of Density Peaks and K-means algorithm to select the initial cluster center and classify the nodes;Secondly,after each clustering iteration,outliers are detected,and the nearest neighbor method is used to redistribute the outliers,and then simulated annealing algorithm is used to optimize the results;Finally,the optimal director deployment location and the mapping relationship between switches and directors are obtained.The simulation results show that,compared with the related multi controller deployment strategy in this paper,the average delay between switches and controllers is reduced by 26.7%,the worst delay is reduced by22.2%,and the average delay between controllers is reduced by 16.2% while ensuring a more balanced load among controllers.The dynamic change of traffic in the network will also cause the load imbalance among controllers.On the basis of static deployment of multiple controllers,continue to study the dynamic load balance among multiple controllers.By optimizing the switch migration scheme,a dynamic load balancing based process optimization strategy based on process optimization is proposed.In the load monitoring module,through the collection of network load information,determine the migration out and migration into the controller;In the decision-making module,first select the probability model by designing the ideal outbound traffic and migration switch,select the switch to be migrated,and then comprehensively consider the impact of delay and migration cost on network performance.Aiming at the disadvantage of slow convergence speed and easy to fall into local optimization in ACO,optimize the heuristic function,The nonlinear decreasing volatilization coefficient is introduced to improve the pheromone updating formula,and the best migration controller is selected for the migration switch;In the switch migration module,the switches are migrated in the descending order of the optimization objective function,so as to avoid the overload of the migrating controller during the migration process and realize the coordinated migration of the switches.The simulation results show that,compared with the existing load balancing strategies,the migration cost is reduced by 10.3%,the flow setup time is reduced by 34.3%,and the delay is reduced by 13.6% on average while ensuring that the load of each controller is more balanced.Figure [24] Table [3] Reference [69]... |
PDF Full Text Request