基于Docker的ARM嵌入式平台性能评测与分析  

Performace measurement and analysis of ARM embedded platform using Docker container

作　　者：彭博 杨鹏[2] 马志程[2] 姚建国[1] 

机构地区：[1]上海交通大学软件学院,上海200240 [2]国网甘肃省电力公司信息通信公司,兰州730050

出　　处：《计算机应用》2017年第A01期325-330,共6页journal of Computer Applications

摘　　要：针对移动互联网时代背景下嵌入式平台硬件在物理信息融合系统中的复杂使用场景和对嵌入式硬件性能提升的迫切要求,结合目前armv7架构的嵌入式开发平台的虚拟化支持条件,研究了嵌入式平台上容器虚拟化解决方案的可行性,提出并设计了基于Docker容器引擎平台的ARM嵌入式平台虚拟化方案。具体介绍了ARM上支持容器虚拟化的Linux系统内核的制作过程,并对Docker容器平台的容器客户机操作系统进行了包括CPU计算、调度性能和文件IO等多方面的性能测试,对主要性能测试结果进行了分析和探究,总结出了系统文件IO性能是ARM嵌入式平台上容器运行的瓶颈,提供通过优化系统IO调度算法的相关解决思路。With the rapid development of the mobile Internet, the increasing demand of embedded hardware usage and the complex usage scenario of the embedded platform hardware in cyber-physical systems, a more efficient virtualization solution was demanded. With the analysis of traditional virtualization solutions and the armv7 architecture embedded platform hardware support for virtualization, a container-based virtualization on ARM with Docker support was deisnged. The implementation details of building Linux OS was introduced which support container-based virtualization. The multi-evaluation of the Guest OS in Docker containers including the computing and scheduling ability of CPU as well as file system IO test with multiple parameters was given. With the performance evaluation of Guest OS, we make sure the container-based virtualization shows advantage over traditional virtualization, and we find that IO performance is the bottleneck of the ARM embedded platform. We have analyzed the reason and we focus on the IO scheduler algorithm and data structure which make great effect on the IO performance. We give some optimization suggestions about IO scheduler as our future work.

关 键 词：容器虚拟化 ARM嵌入式平台 Docker容器引擎 虚拟化性能分析 优化调度 

分 类 号：TP316[自动化与计算机技术—计算机软件与理论]