PaaS云资源调度技术研究  

Research on resource scheduling of PaaS

作　　者：徐鹏[1] 张岩江[1] 苏森[1] 

机构地区：[1]北京邮电大学网络与交换技术国家重点实验室,北京100876

出　　处：《华中科技大学学报（自然科学版）》2013年第S2期52-56,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：新一代宽带无线移动通信网科技重大专项资助项目(2011ZX03002-002-01)

摘　　要：重点研究了平台即服务(PaaS)云资源调度方面的相关技术.首先对PaaS云平台的通用架构和关键技术加以说明,而后分析了PaaS云资源调度的特殊性,提出了将机器学习引入到PaaS云资源调度的方法,在对多种机器学习算法进行了分析和比较之后,进一步提出将决策树算法引入到PaaS云资源调度,并给出了具体的算法模型和运行流程.结论表明:将机器学习引入到PaaS云资源调度可以有效地解决传统PaaS资源调度算法遇到的扩张、收缩和保持时机的问题.The resource scheduling of platform as a service(PaaS)was studied.Firstly,the generic architecture of PaaS and the related key technologies were illustrated.Secondly,the diversity of the resource scheduling of PaaS was stated.Then the design in which machine learning was adopted to optimize the resource scheduling of PaaS.After the analysis of a variety of machine learning algorithms, decision tree algorithm was used in the resource scheduling of PaaS.In the end,the specific algorithm model and the possessing procedure were brought forward.It is shown that to introduce machine learning into resource scheduling in PaaS could effectively solve the problem to decide the time of scaling operations in PaaS.

关 键 词：平台即服务 基础架构即服务 资源调度 机器学习 决策树 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]