Performance Analysis of Centralized and Distributed SDN Controllers for Load Balancing Application

Software Defined Networking (SDN) decuples data and control planes to simplify the management procedure. SDN control plane functions are to manage the overall network through a global view of the network, define policies for the incoming flows, and enforce these policies by using centralized or distributed controller environment. In this paper, we compare the performance of centralized and distributed controllers by using load balancing application with three algorithms namely Random, Round Robin, Weighted Round Robin algorithm. The results show that using distributed controllers is better than centralized in terms of response time and a number of transactions. Furthermore distributed controllers are better for scalability and reduces the bottleneck as well as support failover mechanism. Additionally, Weighted Round Robin algorithm proves its ability to perform better than other mentioned algorithms using centralized and distributed controllers.