A Stochastic Approximation Approach for Foresighted Task Scheduling in Cloud Computing

TL;DR

This paper introduces a reinforcement learning-based foresighted task scheduling method for cloud computing that improves long-term system performance by reducing response time and resource wastage.

Contribution

It presents a novel online reinforcement learning approach for long-term task scheduling in cloud environments, addressing the limitations of existing myopic solutions.

Findings

Reduces response time and makespan of tasks

Increases resource utilization efficiency

Enhances long-term system performance

Abstract

With the increasing and elastic demand for cloud resources, finding an optimal task scheduling mechanism become a challenge for cloud service providers. Due to the time-varying nature of resource demands in length and processing over time and dynamics and heterogeneity of cloud resources, existing myopic task scheduling solutions intended to maximize the performance of task scheduling are inefficient and sacrifice the long-time system performance in terms of resource utilization and response time. In this paper, we propose an optimal solution for performing foresighted task scheduling in a cloud environment. Since a-priori knowledge from the dynamics in queue length of virtual machines is not known in run time, an online reinforcement learning approach is proposed for foresighted task allocation. The evaluation results show that our method not only reduce the response time and makespan of submitted tasks, but also increase the resource efficiency. So in this thesis a scheduling method based on reinforcement learning is proposed. Adopting with environment conditions and responding to unsteady requests, reinforcement learning can cause a long-term increase in system's performance. The results show that this proposed method can not only reduce the response time and makespan but also increase resource efficiency as a minor goal.