Cloud Task Scheduling Algorithms using Teaching-Learning-Based Optimization and Jaya Algorithm

Abstract

Over the last few years, cloud computing has accelerated in the economic and scientific communities due to breakthroughs in virtualization technology. It is an emerging computing technology in which many users submit their requirements (i.e., compute, storage, network, etc.) in the form of tasks to process them through widely dispersed resources (i.e., virtual machines (VMs)) on a pay-as-you-go basis. However, it is pretty challenging to manage the submitted tasks and process them on the VMs, such that overall completion time (i.e., makespan) is minimized. Many researchers have proposed meta-heuristic algorithms to solve the above-discussed task scheduling problem. However, these algorithms are based on algorithm-specific parameters. This paper uses the concepts of well-known teaching-learning-based optimization (TLBO) and the Jaya algorithm, and model them to solve task scheduling problem individually. The rationality behind using these algorithms is that they are algorithm-specific parameter-less algorithms. We call the modeled algorithm as cloud-TLBO and cloud-Jaya algorithm. We model the candidate solutions as tasks and design variables as VMs, and consider the makespan as the objective function. We simulate both the cloud-TLBO and cloud-Jaya algorithm using five synthetic datasets and monitor their results over 50 iterations. Finally, we compare the results with the online benchmark algorithm, called minimum completion time (MCT), to show that the results of the proposed algorithms are near-optimal.