Look ahead before you leap: SDN switch migration scheduling for load balancing

Abstract Although software-defined networking (SDN) promotes a centralized control plane for better resource management, it suffers from reliability and scalability issues. Consequently, the distributed software-defined networking (SDN) architecture proposes logically centralized but physically distributed SDN controllers. However, in the distributed architecture, the load among the controllers may not be balanced, which makes the system less scalable. In this paper, our goal is to balance the load of the controllers. We perform a detailed analysis to establish the ideal load of a controller. Our aim is to minimize the difference between a controller's ideal and instantaneous load. We use short-term traffic forecasts to predict the load of a controller in the next epoch. Thereupon the load balance is affected by migrating the switches. The prediction-based load balancing approach not only balances the load but also minimizes the switch reassignment by preventing switch oscillations. Extensive experiments and comparative performance analyses are performed to validate the framework.