Dynamic Controller-Switch Mapping Assignment with Genetic Algorithm for Multi-controller SDN
Biao Han, Xiangrui Yang, Xiaoyan Wang · 2019
The decoupling of data plane and control plane, which is the key feature of Software-Defined Networking (SDN), introduces considerable communication overhead between controllers and switches. Multiple distributed controllers are introduced in the control plane to improve the scalability and reliability of SDN. Controller placement problem and controller-switch mapping assignment problem are two of most critical challenges in multi-controller SDN. In this paper, we study the dynamic controller-switch mapping assignment problem for SDN, so as to minimize the average event response time from switches. We build the model in a generic way that can be used in both static and dynamic policies. The controller-switch mapping assignment procedure is formulated as a nonlinear integer programming problem. Due to the high cost of brute force and the locality of other heuristic algorithms in solving the formulated problem, we develop a genetic algorithm to solve the problem. Under the assumption that each controller subjects an M/M/1 queuing system, we simulate the proposed genetic algorithm in a static example of the model. Simulation results reveal that the proposed algorithm minimizes the average response time of events while incurring low computational overhead.