A Hybrid Approach for Cloud Load Balancing Optimization

Abstract

In this research paper, a critical and novel approach is presented for cloud load balancing which delves into scheduling scientific workflows in cloud computing. These workflows are characterized by their complexity, demanding significant computational resources and sophisticated data processing capabilities. By leveraging a multi-objective genetic algorithm, this study strategically addresses the challenging task of efficiently distributing the workflows across cloud resources. This is particularly noteworthy as it involves a delicate balance of various conflicting parameters such as time, energy, cost, and adherence to quality of service (QoS) standards. The ingenuity of the presented approach is evident in the integration of an advanced ranking heuristic alongside the application of Bayesian methods for predicting the earliest finish time (PEFT). This dual strategy enhances the decision-making process in the allocation and migration of virtual machines (VMs), a cornerstone in cloud computing efficiency. This research goes beyond traditional methods by focusing on cost and time efficiency and integrating energy consumption considerations, an aspect increasingly relevant in today’s environmentally conscious technological landscape. The results of this research, indicating substantial reductions in both cost and time delays, underscore the effectiveness of the proposed algorithm. By achieving these reductions, this approach offers a more sustainable and economically viable solution for cloud computing environments. Furthermore, the demonstrated potential of multi-objective genetic algorithms in this context opens new avenues for future research and development in cloud resource management and workflow scheduling.