A load balanced congestion aware routing mechanism for Software Defined Networks

Abstract

Nowadays, by accelerated growth of number of devices connecting to the Internet and emergence of IoT, the internet traffic is rapidly increasing. In these environments the probability of congestion is constantly rising. In the traditional networks, the end devices are responsible for controlling the congestion, but since these devices do not have a global and comprehensive view of the network, they cannot predict the congestion to prevent the packet loss. Hence the overall network throughput is reduced. This paper proposes an efficient algorithm to avoid congestion in SDN environments by rerouting as few flows as possible to reduce the network overhead. In this algorithm, the network condition is monitored by calculating the link utilization periodically. When a new flow arrives, it is routed by considering the network condition to prevent the switch overutilization and congestion. If the average link utilization exceeds the rate threshold, the controller predicts the congestion on the link and calculates how much load must be shifted to proper backup paths. Then based on the flow parameters, flows are rerouted to proper backup path which will not be congested by adding these flows. This policy prevents the unbalanced load distribution after rerouting. The emulation on ONOS controller and Mininet emulator shows the efficiency of the proposed algorithm in increasing the network throughput and reducing the number of overutilized links and packet loss.