Research On Failure Detection And Recovery Technology Based On SDN Network

Software defined Network(SDN)as a new new architecture,aims to give traditional network the configurable and programmable characteristics and enhance the controllability and plasticity of the network.With the dilatation of network user scale and service scope,the data plane link failure in SDN network occurs frequently,which will lead to the network service discontinuity and cut down network service quality.Therefore,considering the inevitability and severity of link failure,research on link failure detection and recovery solutions for the data plane of SDN scheme is worth discussing.In this thesis,in view of the problems and challenges existing in the current failure recovery methods,mathematical modeling is carried out and a comprehensive optimization objective is constructed for the recovery problem.The relevant link fault detection and recovery scheme is proposed for the optimization objective.First of all,to address the problem of failure detection in the current SDN network,this thesis proposes a dynamic link failure detection mechanism for SDN based on per-link of BFD,the method of each link detection is adopted to achieve a smaller failure detection time.At the same time,considering that the network situation is constantly changing,the BFD detection timer is dynamically adjusted based on the current network link congestion status periodically monitored by the controller to make the detection mechanism better adapt to the network changes,so as to avoid additional impact on network services.Combining with the fast detection scheme proposed in this thesis,an active failure recovery mechanism based on flow aggregation and priority scheduling is proposed to solve the problem of failure recovery.Firstly,combined with Fast Failover group table mechanism and using flexible data flow marking method to alleviate the problem of balancing the recovery time and storage space consumption.Secondly,considering the possibility of congestion on the backup path after failure recovery and the existence of various business data flows with different network service quality requirements in the network,a priority-based queue management and scheduling strategy is adopted to provide differentiated Qo S services for business applications with different priorities in the network,improve the link congestion and ensure failure recovery effect to a certain extent.In this thesis,the proposed fault detection and recovery means is experimented,verified and analyzed from multiple aspects based on simulation platform,and compared with other related research,which certificates the failure detection and recovery scheme proposed in this thesis can effectively increase the capability of failure recovery.