Optimal Destination Node Selection in Live Virtual Machine Migration

Abstract

Abstract Virtual Machine Placement (VMP) is a critical challenge in cloud computing, directly affecting resource utilization, energy efficiency, system scalability, and operational costs. Effective placement strategies are essential to ensure optimal use of resources while maintaining service-level objectives and minimizing energy consumption. However, existing heuristic-based approaches often struggle to balance placement quality with computational efficiency, particularly in large-scale, heterogeneous, and dynamic cloud environments. This research introduces ACO-VMP, a Virtual Machine Placement algorithm based on Ant Colony Optimization (ACO), aimed at minimizing resource wastage across RAM, CPU, storage, bandwidth, and power. Inspired by the decentralized foraging behavior of ants, ACO-VMP employs pheromone-guided probabilistic decision-making to explore potential placements while adaptively reinforcing resource-efficient mappings. Unlike brute-force approaches, which are computationally infeasible at scale, ACO-VMP achieves near-optimal placement performance with significantly lower execution time, offering a practical alternative for real-time and large-scale deployments. To further enhance algorithmic performance, we perform hyperparameter optimization on the influence of pheromone trails (α) and heuristic visibility (β), analyzing their effects on convergence speed, solution quality, and energy savings. ACO-VMP is implemented and evaluated using the CloudSim Plus framework across diverse configurations of virtual machines (VMs) and physical machines (PMs), under both synthetic and realworld workload scenarios. Experimental results demonstrate that ACO-VMP consistently outperforms traditional heuristics such as First Fit (FF), Round Robin (RRB), and Power-aware Best Fit Decreasing (PBFD), closely approaching the optimality achieved by exhaustive Brute Force (BFR) methods, but with significantly reduced computational overhead. These findings establish ACO-VMP as a scalable, adaptive, and energy-aware solution for intelligent VM placement, contributing towards more sustainable and efficient cloud data center operations.