Load-balancing In Software-Defined Networking: An Investigation On Influential System Parameters

Abstract

Software-Defined Networking (SDN) is a recently-developed networking paradigm that achieves flexible network management through separation of control plane and data plane. In an SDN-enabled network, the deployment of a multi-controller architecture as a centralised control plane along with the implementation of a load-balancing strategy to share the load among controllers in the control plane achieve higher degree of availability and scalability. To efficiently balance the load between SDN-controllers, two tasks seem to be of utmost importance, namely: identifying the system parameters whose values' variations initiate the load-balancing operation, and the amount of time taken to complete the load-balancing operation between controllers. In this paper, an SDN-enabled network topology has been emulated to identify a subset of influential system parameters that trigger load-balancing operation between SDN-controllers. Additionally, this paper determines the impact of all possible combinations of such system parameters on the amount of time taken to complete the load-balancing operation in a cluster of OpenDay Light controllers.