A Mechanism for Controller Failover in Distributed Software-Defined Networks

Abstract

In Software Defined Networks (SDN), reliability plays a major role in maintaining a consistent end-to-end connectivity and dealing with controller failures. A traditional SDN implementation relies on a central controller that runs the control plane. However, in a large-scale Wide Area Network (WAN) deployment, centralized approach has several limitations related to performance and scalability. To address these issues, distributed SDN where multiple controllers that work cooperatively to control a network are deployed. One key limitation of such a distributed network is failure of one or more controllers may cause severe issues like network outage and reduced Quality of Service (QoS). In this paper, we address the problem of controller failure in distributed SDN by proposing two controller failover techniques, namely (a) RandomWeightLoadBalancing and (b) ProgressiveAssignmentLoadBalancing. Our work focuses on load balancing the orphan switches, connected to a failed controller, among available controllers rather than overloading a single controller with all orphan switches. Simulation results show that our proposed solutions perform a successful reassignment of orphan switches and ensure that the load of each controller is well balanced.