Enhancing Load Balancing in Cloud Computing through Adaptive Task Prioritization

Abstract

Cloud computing has become an increasingly popular platform for modern applications and daily life, and one of its greatest challenges is task scheduling and allocation. Numerous studies have shown that the performance of cloud computing systems relies heavily on arranging tasks in the execution stream on cloud hosts, which is managed by the cloud's load balancer. In this paper, we investigate task priority based on user behavior using request properties and propose an algorithm that utilizes machine learning techniques, namely k-NN and Regression, to classify task-based priorities of requests, facilitate proper allocation, and scheduling of tasks. We aim to enhance load balancing in the cloud by incorporating external factors of the load balancer. The proposed algorithm is experimentally tested on the CloudSim environment, demonstrating improved load balancer performance compared to other popular LB algorithms.