An Efficient Approach for Load balancing in Software-Defined Networks

Abstract

Software-Defined Networking (SDN) premise sepa-rates the data plane from the control plane. The centralized control plane brings innovation, programmability, and flexibility for managing the underlying networks. In this paper, we propose a hierarchical control plane-based SDN scheme to achieve load balancing in SDN. The hierarchical architecture is composed of a load balancing control plane and a distributed control plane based SDN architecture. To achieve load balancing, we add an analytical network process (ANP)-based module to rank the controllers according to their load status, for which ANP is applied as an enabling module to forward the flows according to the load distribution of the controllers. The ANP module ranks the controllers according to their weights. Hence, it does not send the request to the overloaded controllers in the network. Our proposed scheme surpasses the analytic hierarchy process (AHP), and Default SDN based load balancing schemes. We have demonstrated the results in three Internet topologies i.e. Abilene, Chinanet, and OS3E regrading delay, and Jain's fairness index in Mininet emulator.