私有云环境下基于虚拟集群的资源共享方法

发布时间：2018-12-13 17:02

【摘要】：随着云计算与大数据处理的飞速发展,越来越多的应用框架开始使用"云"的方式运行,这对企业私有云平台提出了一些重要挑战:如何让私有云环境有限的物理资源支持多个应用框架和多个用户,不仅能保证绝大多数现有的异构应用同时运行,还能为新应用提供扩展性与弹性支持。提出了一种新的服务模型ClaaS(cluster as a service),用一种简单而易于实现的方式为中小规模的数据中心解决了上述问题。ClaaS的主要思路是将虚拟化的单位扩展为集群,为分布式的应用框架直接提供虚拟的集群环境。这样,大部分的分布式应用就可以在无需任何修改的条件下直接运行在虚拟的集群环境中。为了验证ClaaS的可行性,基于现有的轻量化容器技术和软件定义网络技术等实现了一个开源轻量级私有云系统Docklet,能够支持用户按需共享资源,支持Spark、MPI等各种计算框架,及Python、R、Java等语言,并在真实教学科研环境中提供、部署服务。最后描述了系统的设计与实现,并进行了实验与评估。
[Abstract]:With the rapid development of cloud computing and big data processing, more and more application frameworks begin to run in the way of "cloud". This puts forward some important challenges to enterprise private cloud platform: how to make the limited physical resources of private cloud environment support multiple application frameworks and users, which can not only ensure that most existing heterogeneous applications run simultaneously. It can also provide extensibility and flexibility support for new applications. A new service model, ClaaS (cluster as a service), is proposed to solve the above problems for small and medium-sized data centers in a simple and easy way to implement. The main idea of ClaaS is to expand virtualized units into clusters. The virtual cluster environment is provided directly for distributed application framework. In this way, most distributed applications can run directly in a virtual cluster environment without any modification. In order to verify the feasibility of ClaaS, an open source lightweight private cloud system, Docklet, is implemented based on the existing lightweight container technology and software definition network technology. Docklet, can support users sharing resources on demand, Spark,MPI and other computing frameworks. And Python,R,Java and other languages, and in the real teaching and research environment to provide, deploy services. Finally, the design and implementation of the system are described, and the experiment and evaluation are carried out.
【作者单位】：北京大学高可信软件技术教育部重点实验室;北京大学(天津滨海)新一代信息技术研究院;
【基金】：国家自然科学基金Nos.61272154,61421091 国家重点研发计划No.2016YFB1000105~~
【分类号】：TP393.09

【相似文献】