Fair switch selection for large scale software defined networks in next generation internet of things

Abstract Software Defined Networking has been pivotal in enabling on-demand resource utilization and is poised to have an incredible impact on the next phase of the Internet of Things. Its ability to furnish a versatile and expandable network framework is instrumental in accommodating the overwhelming surge of IoT devices and applications. The combination of static mapping and the dynamic flow of traffic over time and space creates an uneven distribution of loads across SDN controllers. Dynamic migration is a solution aimed at rectifying this imbalance by redistributing the load between SDN controllers. Communication for control between switches and controllers becomes burdensome when the matching rules are absent from the table. Our prior research has addressed this issue by employing burst aggregation focused on consolidating similar destinations to reduce the control overhead. In this study, our focus is on ensuring fairness during migration and selecting the appropriate switch. We model a fair switch selection (FSS) algorithm tailored for large-scale software-defined networks. Unlike traditional methods using packets as a basis, FSS utilizes bursts as its input. This model prioritizes bursts considering both their distance and destination, ensuring that switches select bursts with the highest priority to maintain quality of service. Our research delves into evaluating the performance of the proposed algorithm in comparison to four baseline algorithms: round robin, exhaustive search, multi-protocol TCP (MPTCP), and random search. Through extensive simulations, we analyze experimental results based on cost, performance, packet loss, average throughput, and execution time. Experimental results demonstrated a reduction in packet loss by 30% with an average 25% throughput improvement.