A Dynamic Association Strategy for Controller Load Balancing in Software-Defined Networks
Shuangqing Li, Zhihao Li, Weiqi Zhang · 2022 23rd Asia-Pacific Network Operations and Management Symposium (APNOMS) · 2022
Software-defined networks have been widely used in information infrastructure deployment due to the programmability adapting to the flexible changing of network requirements. Distributed multi-controller architecture enhances the scalability and reliability of SDN. Static association between controllers and switches cannot adapt to the change of loads on controllers. In this paper, a dynamic association strategy is proposed, which takes controller load balancing and the request overhead of switch as the optimizing goals. An improved Hungarian algorithm is utilized to adjust the association. Then Nash equilibrium is used in further optimization. This strategy can keep a small volume of controllers and switches involved in adjustment, saving network link and time costs, while reaching the load re-balance of controllers. Experiments show that it achieves satisfactory load balancing between controllers in various topologies, and has good scalability and performance in large-scale networks.