Research On SDN Controller Failure Recovery Deployment And Switch Flow Table Overflow Mitigation Mechanism

It is the rapid development and wide application of network technology that not only provides more convenient services for people's lives,but also greatly promotes the development of social economy.Software-Defined Network?SDN?realizes the separation of the control plane and the data plane,and it has the characteristics of centralized control and programmable.So it has become an effective way to deal with the current huge network scale and data requirements.However,network equipment is under attack,failure and so on under such a huge demand for network services,it will cause very serious consequences if it cannot be recovered in time.However,network equipment is under attack and failure,serious consequences can occur if it cannot be recovered in a timely manner.Therefore,it is beneficial to improve the survivability of SDN to maintain its key functions unaffected or restore network functions at a lower cost when SDN fails.This article studies the controller deployment problem of control node failure recovery in the SDN control layer and the flow table overflow problem of switches in the data layer.The main work is as follows:?1?Aiming at the problem of unrecoverable failure of the controller during SDN operation,it is designed that a deployment scheme considers the controller node failure recovery in the WAN from the perspective of controller deployment.The first stage of this solution adopts Canopy and K-means clustering algorithms to divide the network topology into many subnets,and then the improved particle swarm optimization algorithm is used to determine the controller nodes.In the second stage,the controller selection algorithm based on entropy weight and PROMETHEE-II is used to determine the ideal controller that needs to be changed to the master state in the fault domain when the master state controller has an unrecoverable failure during network operation,so as to recover the master state controller fault.?2?Aiming at the problem that the SDN switch is prone to occur the flow table overflow problem,it is designed that a software-defined network flow table overflow mitigation mechanism that is based on SM-PEPA.SM-PEPA is used to mathematically model the data packets processing process of the SDN to obtain the switch's flow table storage space threshold?₁ and flow table cache threshold?₁.Then the multi-objective decision based on entropy weight is used to determine the ideal neighbor switch for flow table overflow mitigation,and the flow table to be issued is set according to the decision result.?3?Aiming at the problem that the load of the controller connected to the switch where the flow table overflow occurs is high,a linear speed limit scheme based on the controller load is designed to make the load of the controller be at a reasonable level during mitigating the flow table overflow.?4?Aiming at the problem that the packet rate in the network constantly changes and causes the threshold to fluctuate,this will cause the related algorithms to be continuously turned on and off,thereby consuming a large amount of controller resources.Therefore,it is effective when it is required to be no less than the threshold three times in a row.Through comparative experiments on the above schemes,the experimental results show that the related schemes designed are better.The one is able to recover the network function of the fault domain with a lower delay and a higher load balancing rate after the master state controller fails.Another scheme designed takes longer time to maintain the network to perform normal tasks and the load of the controller increases slowly when SDN encounters the same large data flow.Therefore,the related scheme is designed to improve the survivability of SDN.