Link failure aware capacitated controller placement in software defined networks

The latency experienced by the switches in software defined network depends on the number and location of the controllers in the network. The switch to controller latency, in the event of a link failure, increases significantly if the failure of links are not taken into account while determining the placement of controllers in the network. We propose a capacitated controller placement strategy that recovers from a single link failure without significant increase in the worst case switch to controller latency. The goal is to deploy controllers so as minimize the worst case switch to controller latency in the event of a single link failure while satisfying capacity constraints. Moreover, we also propose a model that considers both the worst case latency with and without link failures in the objective function. We evaluated the performance of the proposed models using networks from Internet Topology Zoo and Internet 2 OS3E. Solving our models show that controller placement that plans for a single link failure not only lowers the worst case switch to controller latency with failure but also reduces the maximum and average inter controller latencies.