Performance evaluation of software-defined networking architectures using Network Calculus

Software-defined networking (SDN) are the rising technology in near future to build programmable networks as a way to simplify the evolution of networks. It is a new networking paradigm with the main aim is to decouple forwarding hardware from control decisions. Several SDN architectures have been recently proposed and evaluated. Performance evaluation of these architectures, to study their scalability, was mainly performed using either simulations or experiments. Generally, experimenting with a real system is ex-pensive or impossible. Simulations are extremely slow especially for large systems and provide little insight on how different design parameters affect the actual performance. Therefore, modeling is more effective for evaluating and optimizing SDN architectures prior their implementation. In this paper, we used Network Calculus to study two typical SDN control plane structures, including centralized, hierarchical with 2level. The model describes the behavior of switches and controllers and allows extracting some performance metrics, mainly the delay and backlog.