FractionalLB: Controller Load Balancing Using Fractional Switch Migration in Software-Defined Networks

Abstract

In the context of multi-controller software-defined networks (SDNs), the efficacy of load-balancing can be improved through fractional switch migration. This letter proposes a controller load balancing approach using fractional switch migration, named FractionalLB. FractionalLB aims to reduce the difference between the load of each controller and its corresponding predefined threshold, which is used to identify overloaded and underloaded controllers. The load balancing is achieved by redistributing the load fractionally from the heavily loaded controller to other controllers. We formulate FractionalLB as an optimization problem using a mixed-integer linear program (MILP). When MILP is not tractable, we introduce a heuristic approach based on MILP. Numerical results indicate that FractionalLB outperforms a conventional scheme that follows time-sharing switch migration in terms of synchronization cost and the min-max controller’s load ratio.