Dynamic Task Migration for Enhanced Load Balancing in Cloud Computing using K-means Clustering and Ant Colony Optimization

Abstract

Cloud computing efficiently allocates resources, and timely execution of user tasks is pivotal for ensuring seamless service delivery. Central to this endeavour is the dynamic orchestration of task scheduling and migration, which collectively contribute to load balancing within virtual machines (VMs). Load balancing is a cornerstone, empowering clouds to fulfill user requirements promptly. To facilitate the migration of tasks, we propose a novel method that exploits the synergistic potential of K-means clustering and Ant Colony Optimization (ACO). Our approach aims to maximize the cloud ecosystem by improving several critical factors, such as the system's make time, resource utilization efficiency, and workload imbalance mitigation. The core objective of our work revolves around the reduction of makespan, a metric directly tied to the overall system performance. By strategically employing K-means clustering, we effectively group tasks with similar attributes, enabling the identification of prime candidates for migration. Subsequently, the ACO algorithm takes the reins, orchestrating the migration process with an inherent focus on achieving global optimization. The multifaceted benefits of our approach are quantitatively assessed through comprehensive comparisons with established algorithms, namely Round Robin (RR), First-Come-First-Serve (FCFS), Shortest Job First (SJF), and a genetic load balancing algorithm. To facilitate this evaluation, we harness the capabilities of the CloudSim simulation tool, which provides a platform for realistic and accurate performance analysis. Our research enhances cloud computing paradigms by harmonizing task migration with innovative optimization techniques. The proposed approach demonstrates its prowess in harmonizing diverse goals: reducing makespan, elevating resource utilization efficiency, and attenuating the degree of workload imbalance. These outcomes collectively pave the way for a more responsive and dependable cloud infrastructure primed to cater to user needs with heightened efficacy. Our study delves into the intricate domain of cloud-based task scheduling and migration. By synergizing K-means clustering and ACO algorithms, we introduce a dynamic methodology that refines cloud resource management and bolsters the quintessential facet of load balancing. Through rigorous comparisons and meticulous analysis, we underscore the superior attributes of our approach, showcasing its potential to reshape the landscape of cloud computing optimization.