Weighted Round Robin Load Balancer to Enhance Web Server Cluster in OpenFlow Networks

Abstract

Web server clusters require a reliable network management for increasing the quality of service (QoS). A load balancer system installed in a software-defined network (SDN) is one method that can improve the performance and availability of web server services. SDN is a dynamic and a programmable network management approach, and one protocol that supports it is OpenFlow. This research aims to design and analyse a model of a load balancer on OpenFlow networks, implementing a Weighted Round Robin (WRR) algorithm. The analysis process is conducted by measuring the value of a QoS web server performance parameters, such as response time, throughput, HTTP success, and loss connection. The results showed the WRR algorithm can be implemented for balancing a network system with dynamic resource allocation. The weight workload of each service can be obtained from the needs and existing network resources. The performance of a load balancer on an OpenFlow network is 57% better than in a traditional one for testing of response time conducted in a high connection. However, the throughput and HTTP success connection decreased by 2% and 10%, respectively, while HTTP loss connection increased by 49%.