Research On Energy-Efficient Workflow Scheduling Algorithms In Heterogeneous Cloud System

Cloud computing consists of various development platforms,infrastructure,and software,usually delivered to customers for a fee over the Internet.Users can access resources anytime,anywhere with the rise of cloud computing models.As a possible solution,cloud computing provides a flexible on-demand computing infrastructure for many applications for optimal resource utilization.Scheduling the resource allocation of workflow applications in the cloud can be challenging.Workflow applications refer to applications that need to perform various subtasks in a specific way to complete the entire task.These tasks have a parent-child relationship,and the parent task needs to be executed before its child tasks.The workflow scheduling algorithm should retain the dependency constraints implied by its nature and structure.By considering these constraints,resources are allocated to each subtask of the original task.The main goals of workflow scheduling algorithm are to improve service performance and quality,maintain task efficiency and reduce costs.This paper mainly proposed energy-efficient workflow scheduling algorithms in a heterogeneous cloud system in response to different needs from users and cloud service providers.The main research contents of this article are as follows:1.Considering that the current algorithm is mainly to minimize the energy consumed by processing applications by choosing virtual machines(VMs)to shut down from all opened VMs,and such VM merging through an "on-to-close" approaches usually bring high computational complexity,an energy-efficient VM opening(EEVO)algorithm is proposed,which can choose VMs to turn on from all closed VMs while satisfying the real-time constraint of applications,thus reducing the energy consumed by executing the application.2.After using the EEVO algorithm,there are slacks that can be reduced or eliminated between adjacent scheduling tasks.Therefore,a dynamic scaling energyefficient VM opening(DEEVO)algorithm is further proposed.Based on the EEVO algorithm,DEEVO adopts dynamic voltage and frequency scaling(DVFS)technology to scale down the frequency of VMs executing each task to further reduce the energy consumed by the entire workflow.Experimental results demonstrate that,with the above-mentioned improvements,DEEVO achieves lower energy consumption for real-time workflows than state-of-the-art algorithms do.In addition,DEEVO outperforms state-of-the-art algorithms in the computational efficiency of accomplishing task scheduling.3.Since some cloud users do not know the scope of execution time in advance,it is difficult to define a deadline for the constrained optimization model.In Chapter 4,a workflow scheduling algorithm based on multi-objective game theory is proposed,which aims to optimize both the execution time and total energy consumption at the same time in workflow scheduling.The proposed algorithm is mainly divided into two layers.First,the balance between the total execution time of the workflow and the total energy consumption is minimized by using the dynamic game theory method based on complete information in the pre-optimization layer,and the sub-game perfect Nash equilibrium solution is obtained by the reverse induction method.Finally,in the dynamic adjustment layer,DVFS technology is used to further reduce energy consumption by setting an appropriate deadline for the task.