Cloud Load Balancing using Meta-Heuristics

Abstract

When provided as a service, Cloud Computing (CC) makes it possible to access dynamically scalable and regularly virtualized resources through the Internet. It is hoped that this research will contribute to the disciplines of load balancing and task scheduling by understanding the concept of load balancing in cloud computing through the use of a simulator known as "CloudSim." In this research, it was found that, one of the most significant challenges in cloud computing is load balancing. This prevents scenarios in which some nodes are heavily loaded while others are idle or performing limited work. Load balancing (LB) spreads the dynamic local workload evenly among all the nodes in the cloud. The most important decision to make is the sort of program to run on the available system in order to maximise resource utilisation. Scheduling is the process of allocating a job to the resources that can complete it in the Cloud Computing environment.A heuristic is a strategy that may be used to solve issues rapidly. It is vital to have efficient and effective scheduling procedures in place in order to make full use of the large array of cloud computing possibilities. As a consequence of this investigation, a novel scheduling algorithm is introduced that is derived from two old approaches, Min-min and Max-min, and is designed to capitalise on their advantages while eliminating their disadvantages. The parameters are then evaluated by the CloudSim simulator. There are high-level techniques that function as the driving force behind a problem-specific heuristic known as Meta-Heuristics. A New Meta-Heuristic methodology has been developed by investigating and analysing previously developed processes. The primary purpose is to avoid the drawbacks of iterative improvement, particularly descents, by allowing the local search to escape local optima while doing local searches. One technique to accomplish this is to provide more "intelligent" possibilities to the local search engine rather than random initial responses as a way to assist it. The proposed meta-heuristic model is evaluated using CloudSim and the advantages of two existing techniques, Particle Swarm Optimisation and Genetic Algorithm with the results of several simulations.