Shared Ring And Traffic Load Balancing-Based Failure Recovery Approach In SDN Data Plane

After more than ten years of development,information and communication technology has made great progress.However,the traditional network architecture has encountered enormous challenges in the face of emerging technologies,and the defects of the traditional network architecture have gradually emerged.For example,the underlying network is distributed over a large number of complex and isolated network protocols,thus the network maintenance cost is too high.Additionally,based on the traditional network architecture,the traffic transmission rules are difficult to configure in real time,so it is difficult to implement the network behavior required by the upper layer application in real time.In face of the defects and bottlenecks of traditional network architecture,the academic community proposed the concept of Software Defined Network(SDN)and conducted in-depth research.SDN separates the data forwarding function and control function of the traditional network node,and the control function is concentrated on the controller.Based on the controller,the operation and maintenance personnel can manage the data plane forwarding device through programming,and the SDN network can be programmed.In the process of SDN development,the SDN data plane failure is a problem that needs to be paid attention to and solved.Because the data plane plays the role of data forwarding,component failure in the data plane will cause a small range of network paralysis and directly affect user communication.In general,SDN data plane failure recovery schemes can be divided into two categories:reactive failure recovery schemes and proactive failure recovery schemes.Compared with the former,the latter has a superior performance in terms of recovery delay.Therefore the proactive recovery scheme is the mainstream in the field of current failure recovery area.However,currently most proactive recovery schemes have two drawbacks:(a)the proactive recovery scheme consumes a large amount of backup resources,in this paper the backup resource refers to the backup flow entries(b)the change of network traffic affect backup path performance,resulting in congestion of backup paths after data flow redirection.We propose two fault recovery schemes for these two defects:(1)In order to reduce the backup resource consumption on the basis of the guaranteed recovery delay,we propose a ring-based single link failure recovery approach in SDN data plane.The approach improves the utilization of the backup flow entries by planning the backup path and designing the corresponding flow table,thereby reducing backup resource consumption.(2)In order to avoid that the change of network traffic affects the backup path performance,the backup path does not become congested after the data flow is redirected,and the load of each link of the network is balanced.We propose a traffic load balancing-based single link failure recovery approach in SDN data plane.We combine traffic prediction and failure recovery.We predict traffic and use the predicted value for backup path selection.In addition,backup path is updated regularly based on predicted values to maintain the performance as much as possible.