Adaptive Consistency Strategy of Multiple Controllers in SDN.

In large-scale software-defined networking (SDN), the logically centralized controller usually uses multiple distributed controllers to manage and operate the network in a cooperative manner from the perspective of global network view. One of the major challenges with respect to deploying multiple controllers in SDN is how to synchronize the state among controllers to maintain information consistency. Aiming at solving this problem, we propose a periodically adaptive synchronization strategy of controllers. At first, the consistency level of controllers in the network is quantified based on the characteristics of SDN. Then, an adaptive synchronization strategy is proposed. By the use of this strategy, controllers are divided into three kinds of roles, and the synchronization period is adjusted dynamically by an adaptive function according to the current network state to attain a certain consistency level. The simulation results show that the proposed strategy has merits of reducing communication overhead of controllers as well as improving network availability compared with the other non-adaptive strategies. However, it lacks consideration of load balance among controllers, which will be our future research direction.