Throughput Evaluation of Load Balancing Algorithms (Round Robin, Number of Connection, Server Load) using Identical Servers and Multiple-Application Environments

This paper compares various load balancing algorithms, including round robin, least connections, through simulations implemented with OPNET® Modeller, we analyse the throughput and severs load variation of these algorithms when connected to 4, 8, and 12 servers serving 10000 clients under a 10,000-client mixed workload consisting of HTTP, database, and mail applications. The results illustrate that Round Robin consistently achieves the highest throughput and exhibits stable performance across all cluster sizes, Number of Connections performs poorly in small clusters but demonstrates substantial improvement as server count increases. Variance analysis further confirms stability of Round Robin and Number of Connections in large clusters, while Server Load remains the least stable algorithm.