基于应用特征的PaaS弹性资源管理机制

发布时间：2018-05-18 23:23

本文选题：云计算 + PaaS　；参考：《计算机学报》2016年02期

【摘要】：如何在保障服务质量的前提下,尽可能地节省服务器资源,是云计算环境中弹性资源管理面临的主要问题之一.目前关于云计算环境中弹性机制的研究多数集中在IaaS层,且较少考虑不同应用的具体特征.文中提出了一种基于应用特征的PaaS弹性资源管理机制AFERM(Application Feature based Elastic Resource Management mechanism),主要贡献包括:(1)定义了包含资源开销信息与请求率变化信息的应用特征;(2)设计了一种基于应用执行单元的请求分配机制,在请求数量较大时,将它们划分给多个资源开销相对固定的应用执行单元分别处理;(3)基于应用特征将不同类型的应用搭配部署.作者在自主开发的PaaS平台PAE(Peking university Application Engine)上验证了该机制的可行性和有效性.实验证明,AFERM能够在保证服务质量的前提下,有效地减少服务器的使用:与对比策略相比,AFERM平均可以减少28.3%的虚拟机占用,最多可以减少50%.
[Abstract]:How to save server resources as much as possible on the premise of guaranteeing service quality is one of the main problems in elastic resource management in cloud computing environment. At present, most of the researches on elastic mechanism in cloud computing environment focus on the IaaS layer, and seldom consider the specific characteristics of different applications. In this paper, an application feature based PaaS elastic resource management mechanism, AFERM(Application Feature based Elastic Resource Management mechanism, is proposed. The main contributions include: 1) define the application feature including resource overhead information and request rate change information. With the request allocation mechanism of the execution unit, When the number of requests is large, they are divided into several application execution units with relatively fixed resource cost to deal with each other. Based on the application characteristics, different types of applications are deployed together. The feasibility and effectiveness of the mechanism are verified on the PaaS platform PAE(Peking university Application Engine) developed by the author. Experimental results show that AFERM can reduce server usage effectively on the premise of guaranteeing QoS: compared with the comparison strategy, AFERM can reduce the average virtual machine usage by 28.3% and reduce 50% at most.
【作者单位】：北京大学信息科学技术学院软件所高可信软件教育部重点实验室;
【基金】：国家“八六三”高技术研究发展计划项目基金(2013AA01A213) 国家“九七三”重点基础研究发展规划项目基金(2011CB302604) 国家自然科学基金(61421091)资助~~
【分类号】：TP393.07

【相似文献】