Software Defined Network: Load Balancing Algorithm Design and Analysis

Senthil Prabakaran¹ and Ramalakshmi Ramar²

¹Department of Electronics and Communication Engineering, Kalasalingam Academy of Research and Education, India

 ²Department of Computer Science and Engineering, Kalasalingam Academy of Research and Education, India

Abstract: Software Defined Network (SDN) cut down the monopolies of producing network devices and their applications. It allows the use of an omniscient controller that manages the overall network and promises for simplifying the configuration and management burden of the traditional Internet Protocol (IP) network. The use of hardware load balancer is a critical issue in conventional IP networks that creates many negative impacts such as the cost affordability, features customization, and availability. Also, the existing load balancing algorithm does not consider the flow size generated by the client nodes. Further, flows are not classified based on the threshold value of the dynamic flow size. The paper proposes to compare the performance of two load balancing algorithms such as flow-based load balancing algorithm and traffic pattern-based load balancing algorithm with distributed controllers' architecture. The result shows that the flow-based load balancing algorithm minimizes response time by 94%, enhances transaction rate by 14% and Traffic pattern-based load balancing algorithm has improved availability by 2.69%.               

Keywords: SDN, distributed controller, flow-based algorithm, traffic pattern based algorithm, failover.

                                                                                  Received January 10, 2020; accepted November 24, 2020