With the constant development of Software-Defined Networking (SDN),the demands and services on SDN network will be more rised. As the whole network control and management center,SDN controller has lots of business pressure which can easily suffer from paralysis and failure,service interruptions and other disadvantages. At the same time,the limitations of a single management domains are also increased. In order to solve the serious consequences of SDN controller paralysis and failure and overcome the limitations of a sin-gle management domain,a feasible solution for controller failure recovery based on the monitoring results of SDN controller failure in the multi-management domains scenario is proposed. Specifically,the method of SDN controller failure monitoring based on OpenFlow mes-sages is proposed,the issues of multi-controller deployment in control plane are studied,and the synchronization mechanism of multiple controllers are discussed. The master and slave controllers model is built to implement switch migration,and different failure recovery strategies are adopted according to the different application scenarios within and between different domains. The minimum cost principle of the switch migration is used to select the standby controller to achieve controller failure recovery. The experiment indicates that the use of different recovery strategies for the failure controller can realize the failure recovery of SDN controller in the multi-management do-mains scenario.%随着软件定义联网(Software Defined Networking,SDN)的不断发展,SDN网络的需求和服务不断扩大.SDN控制器作为全网的控制和管理中心,承受着巨大的业务需求压力,极易造成SDN控制器的瘫痪和故障,出现服务中断等不利影响.同时,单个管理域所呈现的局限性也越来越大.为了解决SDN控制器瘫痪和故障造成的严重后果,克服单个管理域的局限性,在多管理域场景下,根据监测到的SDN控制器故障结果提出具体可行的控制器故障恢复的解决方案.为此,提出了基于OpenFlow消息进行SDN控制器故障监测的方法,研究了控制平面的多控制器部署问题,讨论了多控制器的同步机制,搭建主备控制器模型用于实现交换机迁移,根据域内和域间不同的应用场景,采用了不同的故障恢复策略.采用交换机迁移的最小代价原则选取备用控制器,实现故障恢复.测试结果表明,多管理域场景下对故障控制器执行不同的故障恢复策略可以实现对SDN控制器的故障恢复.
展开▼
