Efficient Resource Allocation for Containerized Applications in Virtualized Environments Using Particle Swarm Optimization

Abstract

Efficient resource allocation remains a crucial challenge in cloud and containerized environments, where workloads fluctuate dynamically. Although virtual machines provide strong isolation, virtual machines impose significantly higher computational and resource overhead compared to lightweight Docker containers. To overcome this, Particle Swarm Optimization (PSO) algorithms were applied for intelligent and adaptive resource management. Five PSO variants were evaluated under normal and stress workloads to optimize CPU and memory utilization across containers. Experimental results demonstrate that Adaptive PSO, Standard PSO, and Constriction PSO significantly outperformed the traditional spread-based scheduling strategy. Adaptive PSO achieved the best performance, reducing average latency by 20.5% and CPU usage by 17%, while improving throughput by 19% compared to the baseline. Standard PSO followed closely, maintaining high stability and precision under load. These findings confirm that PSO-based optimization enhances container efficiency, scalability, and responsiveness, offering a lightweight yet powerful solution for modern cloud infrastructures.