Analysis and Prediction of Tasks′ Resource Usage in the Cloud
DENG Li（1,2）,REN Yu-lin（1,2）,ZHU Jin-can（1,2）,HE Heng（1,2）,LI Chao3
1(College of Computer Science and Technology,Wuhan University of Science and Technology,Wuhan 430065,China)2(Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System,Wuhan 430065,China)3(Department of Information Development and Management,Hubei University,Wuhan 430062,China)
摘要 预测任务的资源使用状况是提高云平台资源使用率的重要手段之一.然而云计算平台资源使用的动态性、不确定性和突变性使得预测效果有限.为了提高云平台任务的资源使用率预测性能,本文做了如下工作:1）详细地分析了当前主流云平台的资源使用情况,提炼了云平台任务的资源使用特征；2）根据云平台的特点设计了适合任务的资源使用预测性能评价函数PEFOT(Performance Evaluation Function fOr Tasks,PEFOT)；3）设计并实现了一种云平台任务的资源使用率预测方法REPO-TASK(REsource Prediction method fOr TASKs,REPO-TASK).使用Google云平台数据集进行了实验,结果表明,相对于目前已经提出的任务资源使用率预测模型BP和LSTM,REPO-TASK方法具有更好的预测性能,PEFOT值平均下降了3.2591.
Abstract：The key to make right management decisions in big data centers is to predict the future load of tasks.However,the dynamics,uncertainty and mutability of tasks in cloud computing make the prediction difficult.In order to solve the above problems,the following work is done in this paper:1)Resource usage of several main clouds is analyzed in detail,and the characteristics of resource usage are summarized;2)Performance evaluation function for the prediction of task resource utilization(PEFOT)is designed;3)A resource utilization prediction method for tasks(REPO-TASK)in cloud platform is designed and implemented.Then we test this method on Google Trace.The result shows that,our method′s PEFOT is reduced by an average of 3.2591 compared with proposed method.
邓莉（,）,任雨林（,）,朱金灿（,）,何亨（,）,李超. 云平台任务资源使用状态预测分析研究[J]. 小型微型计算机系统, 2020, 41(2): 381-386.
DENG Li（,）,REN Yu-lin（,）,ZHU Jin-can（,）,HE Heng（,）,LI Chao. Analysis and Prediction of Tasks′ Resource Usage in the Cloud. Journal of Chinese Computer Systems, 2020, 41(2): 381-386.
2020, Vol. 41 
