云计算环境下基于主动预测的节点部署模型研究

发布时间：2018-02-21 00:51

本文关键词： 云计算预测资源部署 Holt-Winters模型整体最优　出处：《计算机科学》2015年09期 　论文类型：期刊论文

【摘要】：随着云计算的不断普及,越来越多的用户选择将自身的业务迁移至云计算系统。用户的使用习惯与社会日常的运行规律也伴随着大量用户涌入云计算系统,如每早8点集中地向云计算系统申请资源节点,这给系统带来了一种可预期的资源冲击。针对上述问题,提出了一种基于主动预测模式的资源部署模型。该模型首先根据预测模块中Holt-Winters季节指数平滑模型的算法周期长度来预测下一个时间周期的任务请求量,通过设计的主动预测算法判断是否应对当前的任务请求量做出响应并得出其具体数量、位置等参数指标,以实现对用户使用规律的主动应对。使用CloudSim进行仿真实验,系统地评判模型的性能。实验结果表明,AF-HW模型在应对可预测的海量并发任务请求时可有效地提升单点及整体的响应速率,使用户得到更好的体验。
[Abstract]:With the popularity of cloud computing, more and more users choose to migrate their business to cloud computing system. If a centralized application for a resource node is made to a cloud computing system every eight o'clock in the morning, this brings a predictable resource impact to the system. In this paper, a resource deployment model based on active prediction model is proposed, which first predicts the amount of task requests for the next time cycle according to the algorithm cycle length of the Holt-Winters seasonal exponential smoothing model in the prediction module. The designed active prediction algorithm is used to determine whether or not to respond to the current task request and to obtain the specific quantity and location of the target parameters, so as to realize the active response to the rule of user use. The simulation experiment is carried out by using CloudSim. The experimental results show that the AF-HW model can effectively improve the response rate of both the single point and the whole in response to a large number of predictable concurrent task requests and enable users to get a better experience.
【作者单位】：解放军信息工程大学三院;
【基金】：武器装备预研重点基金资助项目(9140A15060311JB5201)资助
【分类号】：TP302

【参考文献】