Research On Link And Controller Failure Recovery In SDN

Failure recovery is a key function in guaranteeing the normal transmission and system reliability.In Software Defined Network(SDN)based networks,the data plane and control plane both face the failure problem which cause a huge threat to the network reliability.Developing fault tolerance technologies and improving the network reliability have become a hotspot of research in SDNs.For link failure recovery in data plane,balancing the recovery time and resource usage is the main challenge.Inspired of reactive and proactive recovery strategy,we propose a cooperative link failure recovery scheme based on them.We formalize the recovery problem as a multi-objective optimization problem and devise a two-stage algorithm for it.Predicting the failure and preconfiguring the backup paths for the link happen in the first stage,which guarantees the connectivity connection.In the second stage,we adjustment the backup paths according to the real-time state for better resource usage.By aggregating the advantages of two recovery schemes,the cooperative recovery scheme can recover the failure efficiently.For the challenge that recovery time conflicts the controller load in recovering the controller failure in SDNs,we propose a controller load-aware recovery scheme.We formalize the recovery problem as a math model and design a heuristic algorithm to address the backup controller selection.We distinguish the uncontrolled switches into some sub-domains according to the recovery time between switches and backup controllers,then we adjustment the matching pairs with the aim to avoid load oscillation of the backup controllers.The heuristic algorithm guarantees the proper allocation of the loads in switches,which accomplishes the balance of recovery time and controller load during the recovery.We use the Mininet and Ryu controller to verify the performance of the proposed failure recovery scheme in recovery time,storage space,QoS guarantee and load proportion of the controllers in SDNs.Compared to other schemes,the proposed failure recovery scheme can satisfy carrier-grade recovery requirement,meet QoS requirements in terms of delay and bandwidth,and conserve communication cost in SDNs.