基于时间序列的Openstack云计算平台负载预测与弹性资源调度的研究  

Research of load forecasting and elastic resources scheduling of Openstack platform based on time series

作　　者：李达港 李磊[1] 金连文[1] 黄甘波 吴权[2] LI Dagang;WU Quan;LI Lei;JIN Lianwen;HUANG Ganbo(School of Electronic and Information Engineering, South China University of Technology, Guangzhou 510641, P. R. China;21CN, Guangzhou 510630 , P. R. China)

机构地区：[1]华南理工大学电子与信息学院,广东广州510641 [2]世纪龙信息网络有限责任公司,广东广州510630

出　　处：《重庆邮电大学学报（自然科学版）》2016年第4期560-566,共7页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家科技支撑计划(2013BAH65F01-2013BAH65F04);广东省工业高新技术领域科技计划项目(2013B010202004);广东省应用型科技研发专项(2015B010131004)~~

摘　　要：随着云计算技术的发展与运用,云计算在资源的效用比、按需服务等方面优势显著;相对于传统的计算资源构建,云计算凭借其安全性保障、高效的弹性计算资源分配能力、简易的硬件要求等特性,能实现面对不同需求时的计算资源快速弹性构建。基于云计算的弹性资源效用比为研究点,以经典的电信接入随机过程模型—泊松过程为基础,采用相关性时间序列模型对资源预测进行研究,并对相似的多类模型进行仿真与分析;最后以Openstack云平台为基础,结合实际需求对模型进行了工程化实现和初步测试。结果表明,该方法为云计算按需使用和资源弹性构建提供了一种可行的方式,在保证服务等级协议(service-level agreement,SLA)的同时,进一步降低云计算平台的运行损耗,提高资源的效用比。With the rapid development of cloud computing, cloud computing outperforms in aspects of providing high utility ratio and on-demand resources. Compared with traditional computing resource construction technolog}^, cloud computing,which provides better security levels, higher elastic resource allocation capability and lower hardware requirements, can satisfy the need of fast resource allocation. We focus on one of the researching hot spot of cloud computing, elastic resource effectiveness ratio, and use the time series correlation based on network access probability distributions, Poisson distribution,to analyze and simulate different related models. Finally,based on Openstack cloud computing platform, the model has been developed, which is suitable for the real environment. We also have tested the platform and got some useful results.The results show that the way we provided is a feasible way to improve the cloud computing resource utilization ratio and guarantee service-level agreement( S L A ).

关 键 词：时间相关 泊松分布 弹性资源分配 高效用比 OPENSTACK 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]