基于Mesos和Docker构建大数据平台技术研究与应用  

Research and Application of Big Data Platform Based on Mesos and Docker

作　　者：李丽新[1] 

机构地区：[1]吉林建筑大学电气与计算机学院,长春130118

出　　处：《吉林建筑大学学报》2016年第5期81-84,共4页Journal of Jilin Jianzhu University

摘　　要：本文针对目前在集群中各种框架分别独立运行、框架间静态划分资源、集群中资源利用率低、吞吐量低等问题进行了研究,提出以Mesos分布式资源管理技术和Docker容器隔离技术,构建一个将不同的大数据框架整合在一起的基本大数据分析处理平台,允许不同的大数据分析框架之间在同一个集群内共享资源分配,跨框架管理集群资源,提高了集群内资源使用率和吞吐量,增加了大数据分析平台的灵活性和扩展性,降低了大数据平台的复杂性.In this paper, we do some researches aiming at respectively independent operations in different present frameworks of clusters, static partition of resources between frames, low ratio of resource utilization in cluster and throughput, etc. In this paper, we propose the technology of distributed resource management based on Mesos and Docker container isolation and construct a basic big data analysis and processing platform that can integrate different big data frames which allows different frameworks of big data analysis to share resource allocation in the same clus- ter and manage different resources in clusters to improve the ratio of resource utilization and throughput within the cluster increase the flexibility and scalability of big data analysis platform, and reduce the complexity of the big data platform.

关 键 词：大数据 分布式资源管理 容器隔离 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]