利用Docker容器技术构建大数据实验室

摘要

Nowadays,the cultivation of big data talents has been popularized both at domestic and foreign universities,but there are many problems in common method of building big data laboratory.In view of the feature of big data experimental environment,we propose a method of constructing Hadoop cluster on an ordinary single computer with Docker container.Moreover,the details of laboratory construction are also elaborated,such as Hadoop image production,Docker environment deployment,container operation,Hadoop cluster testing,Eclipse development environment configuration and image migration.Practice has shown that the new method can save lab funding without additional equipment and maintenance workload,and the reusability of the lab can be improved as well.Meanwhile,the new method has good effect on experiment teaching.By this way,a useful and worth spreading idea is provided for the construction of large data labs in colleges.%目前大数据人才培养在国内外高校开始普及,但大数据实验室的几种常用建设方法都存在着诸多问题.针对大数据实验环境的特点,提出一种利用Docker容器技术在单台普通计算机上构建Hadoop集群的方法,并详述了Hadoop镜像制作、Docker环境部署、容器运行、Hadoop集群测试、Eclipse开发环境配置、镜像迁移等实验室建设细节.实践表明,新方法无需额外购置设备,节省实验室资金,同时未增加机房维护工作量,提高机房可复用性,新方法实验教学效果良好,为高校的大数据实验室建设提供参考.