Abstract
— This paper addresses virtual machine (VM) resource allocation in cloud data centres, focusing on scalability and network performance. With increasing cloud demands, efficient VM deployment is essential for high system performance. The paper discusses NetDEO, an algorithm using particle swarm intelligence to place virtual machines based on real-time demand. Evaluating NetDEO across FatTree, BCube, and Tree topologies shows improvements in load balancing, traffic stress reduction, and workload adaptability. Simulation results demonstrate better resource utilization and less congestion compared to legacy methods. FatTree and BCube topologies performed well, while Tree topology showed limitations under high workloads. Key metrics validate NetDEO's effectiveness in improving cloud performance, and the study suggests exploring hybrid AI models for further enhancing dynamic resource management and optimizing large-scale cloud networks. Index Terms – Virtual Machine, swarm intelligence, FatTree, cloud computing and NetDEO.