Budget-constrained Controller Placement in Software-defined Network

Software-defined network (SDN) provides an emerging platform for future Internet as it allows flexible reconfiguration of the network by separating the control programs (controllers) from the forwarding devices (switches, routers). The placement of controllers, and assigning switches to them is an important problem in SDN, since it depends on various constraints such as reduction of latency between nodes, placement cost, load of the network, and re-assignment of switches in case of failure. In this paper, we have formally formulated the controller placement problem in SDN, where our objective is to find a subset of locations for placing controllers, and assigning switches to them such that the maximum latency of the network is minimized within a specified budget. The problem is first proved to be NP-complete. Then we propose a genetic algorithm-based meta-heuristic approach to solve the problem. Our scheme is then extended to tackle fault tolerance, where the switches of a failed controller are reassigned to other active controllers. If reassignment causes an overload to the SDN, then a scheme is proposed to place a new controller in the SDN, while satisfying the load and budget constraints. The proposed algorithms are evaluated on various networks from the Internet Topology Zoo. The simulation results show that our approach can be effectively implemented in real-life SDN scenarios, such as data center networks, enterprise networks, and cloud-based networks.