Improving the Reliability of Software-Defined Networks with Distributed Controllers Through Leader Election Algorithm and Colored Petri-Net

Abstract

Software-Defined Networks (SDNs) are developed to compensate the complicated function of the controlling parts of the given network elements and making the scalability easier. In SDNs, the controlling operations are implemented by a logically centralized controller, where the occurrence of failure resulting in separation of the control and data plane is inevitable. Since the reliability of centralized controllers is low due to their being of the single point of failure, the focus of this study is to improve the reliability of the SDNs with distributed controllers. In this article, the wide networks are partitioned into smaller subnetworks, where each one is being controlled by a controller in order to reduce the failure effect(s). In each subnetwork, its reliability is calculated by considering the number and degree of nodes and the loss rate of the links and then is transmitted among the controllers through the Leader Election and Dijkstra Algorithms. Afterward, the controller with the highest reliability rate is considered as the coordinator through the newly proposed Coordinator Finder Algorithm. In practice, when a controller fails, the coordinator would choose the appropriate controller for its subnetwork in a transitory manner, where the fault tolerance and accuracy would be improved and the latency would reduce. A newly designed Colored Petri-Net is applied to verify this proposed method.