An SDN-based approach using RYU controller for load balancing and performance evaluation in hybrid networks with machine learning algorithms

Abstract

Software-defined networking (SDN) is a technology that is transforming network efficiency, particularly in terms of balancing loads \cite{bhardwaj2023network}. In this work, an analysis of an SDN-based load balancing and performance analysis via machine learning approaches in the RYU controller is presented. In a simulation conducted in a thorough manner in Mininet, performance analysis of SDN in managing network traffic in a range of topologies including tree, star, linear and cluster networks is discussed. In our work, an analysis of the performance impact of SDN load balancing in terms of performance factors including throughput, latency, jitter, and packet loss is discussed. Heavy traffic as example media, voice, VoIP, etc was used for evaluating the performance. We observed a performance improvement via SDN when accompanied by smart techniques in balancing loads is noticed to have a significant impact in terms of dynamically distributing loads and minimizing congestion in networks using our load balancer which is based on Round Robin Scheduler. With such observations, SDN proves to be an effective and efficient mechanism for high-performance and high-scalability networks in current infrastructure requirements.