基于QoS的云任务调度算法研究  

Research on Cloud Task Scheduling Algorithm Based on QoS

作　　者：房超 黄春梅[1] FANG Chao;HUANG Chunmei(School of Computer Science Technology and Information Engineering,Harbin Normal University,Harbin 150000,China)

机构地区：[1]哈尔滨师范大学计算机科学与信息工程学院

出　　处：《软件工程》2020年第3期22-27,共6页Software Engineering

基　　金：国家自然科学基金项目(61202458/61403109);黑龙江省自然科学基金项目(F2017021)

摘　　要：随着云计算的发展,越来越多的人开始使用“云”来处理他们的业务,这对公有云平台提出了一些重要挑战:如何让公有云平台在不断激增的云业务模式下,既能保证云用户的服务满意度,同时也能稳步提高云服务商(Cloud Service Providers)的收益。首先建立了任务调度算法以及QoS需求约束等相关模型,然后将QoS(Quality of Service)需求约束分别引入到三种传统任务调度算法(FCFS(RR)、MinMin和MaxMin算法)中对其进行改进,接着将改进后的算法与传统任务调度算法之间进行比较,通过选取在任务完成度、任务最终完成时间(MakeSpan)、任务平均执行时间(这些影响用户的服务满意度),以及云服务商总收益等方面的指标表现,最后确定了一个较好的改进MinMin任务调度算法(I-MinMin算法)。实验通过CloudSim进行模拟,并采用了现有的阿里云ECS云服务器中的虚拟机实例相关数据。结果表明:在任务量不断增加的情况下,I-MinMin算法在用户的服务满意度各方面,以及云服务商总收益等指标表现上要更优于其他算法,更好地实现了用户和云服务商的双重利益。With the development of cloud computing,more and more people begin to use"cloud"to manage their business,which poses some important challenges for public cloud platforms.For example,how to make public cloud platforms,under the ever-growing cloud business model,to ensure the service satisfaction of cloud users and steady improvement of the revenue of cloud service providers(Cloud Service Providers).Relevant models such as task scheduling algorithms and QoS requirement constraints were established first.Then the QoS(Quality of Service)requirement constraints were introduced into three traditional task scheduling algorithms(FCFS(RR),MinMin,and MaxMin algorithms)for improvement.Then this paper compares the improved algorithm with traditional task scheduling algorithm.The three improved algorithms are compared with the original algorithms in terms of task completion,task final completion time(MakeSpan),average task execution time(these affect user service satisfaction),and the total revenue of cloud service providers'performance.Lastly,a better improved MinMin task scheduling algorithm(I-MinMin algorithm)was determined.The experiment was simulated by CloudSim,and the relevant data of the virtual machine instance in the existing Alibaba Cloud ECS cloud server was used.The results show that the I-MinMin algorithm outperforms other algorithms in terms of user service satisfaction and the total revenue of cloud service providers in the context of increasing tasks,therefore better achieve the dual interest of both users and cloud service providers.

关 键 词：云计算 任务调度算法 QOS需求 用户服务满意度 云服务商总收益 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]