Reliability-Aware Controller Placements in Software Defined Networks

Software Defined Networking (SDN) makes network management simple and flexible by decoupling control plane from data plane. Careful controller placements can achieve good performance in terms of switch-to-controller latency which is an important concern in SDNs. Most existing research on controller placements in SDN assumes a failure free network. However, a link failure may break down the communication between switches and controllers. The backup path may substantially deteriorate the switch-to-controller latency upon link failure without careful controller placements. In this paper, we investigate the controller placement problem by jointly taking into account both the communication reliability and the communication latency between the switches and the controllers if any link in the network fails. We first introduce a latency metric that considers the switch-to-controller communication latency in both the cases of no link failure and single link failure. We then formulate a novel SDN controller placement problem with the aim to minimize the accumulated latency of the primary and the backup paths, for which we propose an efficient Reliability-aware Controller Placement (RCP) algorithm. The algorithm partitions the network into multiple subnetworks, and places a controller in each subnetwork. The algorithm also performs local search to determine controller locations and switch-to-controller mapping relationship. The simulation results illustrate that the proposed algorithm can effectively reduce the backup path latency and the accumulated latency of the primary and the backup paths.