Path Selection in Software Defined Network Data Plane using Least Loaded Path

Abstract

Centralized control with a global network view, Software-Defined Network (SDN) provides excellent chances for enhanced network performance. However, SDN may suffer from congestion traffic, high latencies, and low transfer rate data due to limited occupancy and unbalanced traffic allocations. A path selection mechanism has been researched to determine which path to be chosen for transferring packet data. Furthermore, in this paper, we propose a path selection mechanism within SDN's data plane to prevent congestion or avoid link overload and achieve load balancing in the data plane. Fat-tree topology has been used to assess the load balancer outcome. This study's results indicate a decrease in latency, while an increase in data transfer rates. The succeeding of the Least Loaded Path (LLP) mechanism in sending data packets through the best path that has been chosen represents the load balancing achieved, where this is not done on the default integrated Load Balancing (LB) on a controller.