MEF:基于在线容错的云计算资源调度方法  

MEF:CLOUD RESOURCE SCHEDULING METHOD BASED ON ONLINE FAULT TOLERANCE

作　　者：都繁杰 李静[1] 王亮 夏天 Du Fanjie;Li Jing;Wang Liang;Xia Tian(College of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,Jiangsu,China;Information and Communication Company of State Grid Shanghai Electric Power Company,Shanghai 200000,China)

机构地区：[1]南京航空航天大学计算机科学与技术学院,江苏南京211106 [2]国网上海市电力公司信息通信公司,上海200000

出　　处：《计算机应用与软件》2024年第3期336-344,共9页Computer Applications and Software

基　　金：国家电网公司总部科技项目(SGSHXT00JFJS1900093)。

摘　　要：云计算系统效率的关键是提升资源利用率与性能,系统可靠性的关键是备份组件。针对云环境下任务完成时间最小化、容错成本优化问题,提出一种基于在线容错的云计算资源调度方法。在线容错具有静态、动态的属性。静态容错根据组件可靠性和完成过程可靠性改进LeaderRank算法备份关键组件;动态容错在故障时快速替换故障组件;通过MEF算法实现任务完成时间短、容错成本低的多目标优化。实验表明,与现有容错方法对比,MEF算法不仅降低了容错费用,而且增强了故障时的系统效率。The key to cloud computing system efficiency is to improve resource utilization and performance,and the key to system reliability is backup components.In order to minimize task completion time and optimize fault-tolerant cost in cloud environment,an online fault-tolerant resource scheduling method for cloud computing is proposed.Online fault tolerance has static and dynamic properties.The LeaderRank algorithm was improved to backup key components according to component reliability and completion process reliability.The dynamic fault tolerance could quickly replace the fault components when the fault occurs.The MEF algorithm was used to achieve the multi-objective optimization with short task completion time and low fault tolerance cost.Experimental results show that compared with the existing fault-tolerant methods,MEF algorithm not only reduces the cost of fault-tolerant,but also enhances the system efficiency.

关 键 词：云计算 在线容错 多目标优化 资源调度 云组件排名 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]