Load balancing and quality of experience in software-defined networks

Abstract

Software-defined networks (SDN) aim to eliminate the disadvantages of traditional network environments by separating the control plane and the data plane. As a result, the controllability of the network is increased. This study proposes to examine the relationship between load balancing and Quality of Experience (QoE) in SDNs in terms of Structural Similarity Index Method (SSIM), Round Trip Time (RTT), delay and hop count measurements. In addition, the data loss parameter is also examined. A comparison is made between clients and servers in a video streaming test. Five scenarios, namely no-load balancing, Round Robin, average RTT, delay-based and the proposed load balancing method, are tested. The average SSIM value of 0.9678 is obtained with the proposed load balancing method. When the data loss is examined, the average data loss value is 0.25 Mbytes higher than the Round Robin method, but better results are obtained than the other scenarios. High SSIM value and reduction in data loss indicate a better QoE. It is observed that load balancing operations performed by considering the load status of the servers and the statistical data of the network give better results.