摘要
云资源负载预测是云计算系统体系规划的一个重要组成部分,其预测效果直接影响到云计算系统的经济性和服务质量.为保证基础设施及服务(Iass)模式下资源有效分配和高效调度,实现精准有效地负载预测,本文提出了一种基于三支决策的云资源负载预测模型(DMASVR-3WD).借鉴三支决策的基本思想,根据负载特征变化进行三支划分,来设立云资源需求的三个不同时期,并有针对性的施加策略.对于平缓期和抖动期,采取立即决策的处理方式,直接进行负载预测处理,对于中间波动期,采取延迟决策的处理方式,依据代价最小化的原则,对其进行划分处理.实验结果表明,所提出的算法能够精确实现负载预测,有效保证用户的服务等级协议.
The prediction of resource usage in cloud computing is an important part of cloud computing system capacity planning,and its prediction effect directly affects the economics and service quality of cloud computing systems.To ensure the accurate allocation of resources and efficient scheduling in the infrastructure and service(Iaas)mode,and to achieve accurate and efficient load forecasting,this paper proposes a cloud platform load forecasting model based on three-way decisions(DMASVR-3WD).Based on the basic idea of three-way decision,we set up three different periods of cloud resource demand,namely,the gradual period,the jitter period,and the volatility period according to the load characteristics,and then targeted application strategy are used to process three different regions.For the gradual period and the jitter period,the immediate decision processing method is adopted,and the load prediction processing is directly performed;For the intermediate fluctuation period,the processing method of delay decision is adopted,and the processing is divided according to the principle of minimizing the cost.The experimental results show that the proposed algorithm can accurately implement load prediction and effectively guarantee the user’s service level agreement.
作者
杨阳
姜春茂
李志聪
YANG Yang;JIANG Chun-mao;LI Zhi-cong(School of Computer Science Technology and Information Engineering,Harbin Normal University,Harbin 150025,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2020年第7期1363-1370,共8页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61202458,61403109)资助
黑龙江省自然科学基金项目(F2017021)资助
哈尔滨师范大学硕士研究生创新科研项目(HSDSSCX2019-11)资助。
关键词
三支决策
三支划分
云资源
负载预测
代价评估
three-way decisions
three-way divisions
cloud resources
load forecasting
cost assessment
