Load balancing is an efficient technique used to maximize throughput, optimal resource utilization, minimized response time and avoiding congestion. This can be achieved by distributing the workload evenly across two or more network stations, nodes or buffers, links, central processing units, hard drives, or other resources. In this paper, we have modeled and developed a load balancing approach in a discrete-time domain to analyze and evaluate the system of finite network buffers using an early arrival system. Our approach of modeling such a system consists of two steps. The first step is the determination of all system-state stages and their corresponding transition probabilities. Next, we compute various performance measures by utilizing the system state transition probabilities for its steady-state behavior.