Open Flow based Dynamic Traffic Distribution among Servers in Software Defined Networks

Abstract

Software defined Networking (SDN) has been a major focus of research works, a solution for handling today’s growing network architecture.For handling large number of web requests, network entities like Routers, Switches, Servers, Firewalls, Load Balancers,etc are deployed at different locations of globe by Web Service providers.To address issues such as network congestion and overloading,service providers use multiple replicas in the server cluster to provide the same services where network virtualization and effective load balancing is very important.This work proposes the implementation of real-time traffic management strategy based on Fuzzy Logic.The Fuzzy membership characteristic that affects performance parameters of server load has been analyzed and load state of virtual servers in real-time is evaluated through Fuzzy Logic.With three major parameters of web server (CPU,RAM and Bandwidth Utilization) as an input of fuzzy system,the clients requests are forwarded to the server having minimum load at real time.The proposed load balance algorithm written in Python is Simulated on Mininet and Performance of each web servers are compared with existing Round Robbin(RR) and Least Connections(LC)load balancing schemes.The results demonstrate that the proposed scheme has improved response time and higher throughput as compared to the other load balance solutions.