Evaluation of Load Balancing Optimization with Resource-Based Algorithms on a P4-Based Programmable Data Plane

The rapid growth of the internet has led to an increasing number of users relying on servers for diverse services, especially large-scale such as data centers. However, the surge in requests can overwhelm servers, leading to overloads and system collapse. Load balancing has emerged as a solution to efficiently manage and distribute these requests. This research explores load balancing in the context of Software Defined Networks (SDN) using the Resource Based Algorithm and Programming Protocol-independent Packet Processors (P4) language. The experiment involves three methods, each with specific objectives: system testing, performance evaluation, and fault tolerance analysis. The experimental setup utilizes VirtualBox as the hypervisor and Ubuntu 16.04.06 Long-Term Support (LTS) as the operating system. The results demonstrated that 64 flows in a topology with four servers attained the optimal equilibrium between maintaining response time stability and optimizing load balancing efficiency. Furthermore, the performance evaluation test with four servers reached equilibrium with an average response time of 0.035 seconds. The fault tolerance test results showed that the algorithm initially balanced the load with three online servers up to batch 15, and the algorithm restarted and balanced the load up to batch 27. Overall, this research contributes to evaluate optimization of load balancing in SDN environments P4 based to create a high-quality system that can be used in the data center)