GLBSM: Greedy-Based Load Balancing by Reducing Switch Migrations in Software-Defined Networks
Upendra Prajapati, Bijoy Chand Chatterjee, Amit Banerjee · 2022
Software-defined networks (SDNs) is an emerging technology to improve the scalability and network performance of data-intensive applications running with a high traffic volume. The load balancing among the controllers with a high number of switch migrations is a major obstacle in a multi-controller SDN, which degrades the overall system’s performance. To resolve the above issue, this paper proposes GLBSM, a Greedy-based Switch Migration approach, for load balancing among the controllers. GLBSM starts the switch migration process from an overloaded controller to an underloaded controller if the mean-square deviation of a load of each controller from the average load of all controllers is greater than a pre-defined threshold. A controller is overloaded if its load is greater than the average load of all controllers and vice-versa. We introduce an algorithm for load balancing and it achieves load balancing by reducing the number of switch migrations. Next, we analyze the time and space complexities of the algorithm. Furthermore, we evaluate GLBSM using a simulation study. Numerical results indicate that GLBSM outperforms SMCLBRT and CAMD in terms of the number of switch migrations by 31% and 12%, respectively.