Docker容器的性能监控和日志服务的设计与实现

发布时间：2018-01-23 14:34

本文关键词： 云计算 Docker 容器性能监控日志　出处：《浙江大学》2016年硕士论文　论文类型：学位论文

【摘要】：作为新兴的云平台,Docker打破了传统以虚拟机为基础的云计算形式,以容器为基本单位为开发人员提供构建、发布和运行分布式应用所需的平台。目前,在性能监控和日志服务方面,Docker平台仅仅支持对单个容器信息的命令查询,功能相对欠缺,无法从整体上为分布式应用提供完善的性能监控方案,这也限制Docker平台上应用的发展前景。首先,本文基于Docker平台现有功能,对Docker容器的性能监控服务进行了研究与完善,设计实现了分布式容器应用的性能监控方案,并设计实现了警报功能用于辅助监控服务,用户通过本系统即可直观地监控其部署在Docker平台上的应用状态及资源使用情况,解决了当前无法对Docker容器集群进行整体监控的问题,使部署在Docker云平台上的应用能更稳定、健壮地运行,并为应用的性能调优提供了基础。此外,本文分析了当前Docker容器日志管理上的不足,利用Fluentd和MongoDB组件进行封装再设计,使用Etcd技术记录日志数据的复杂关系,设计实现消息队列机制以提高日志传输的性能,设计实现数据库分片副本机制以提供高可用的数据库集群,弥补了当前Docker云平台上的应用日志缺乏整体性管理的弊端,实现了对应用在多节点、多容器中所产生的复杂日志的收集传输、汇总存储及快速检索,为Docker容器应用的运维调优及排错提供了基础。最后,本文对系统的功能进行了测试检验。本文中的内容对Docker平台服务的集成和完善有一定的参考意义。
[Abstract]:As a new cloud platform, Docker breaks the traditional cloud computing form based on virtual machine and provides build for developers with container as the basic unit. The platform needed to publish and run distributed applications. Currently, the Docker platform only supports command queries for single container information in terms of performance monitoring and logging services, which is relatively lacking. Can not provide a complete performance monitoring scheme for distributed applications, which also limits the development prospects of applications on the Docker platform. Firstly, this paper based on the existing functions of Docker platform. The performance monitoring service of Docker container is researched and perfected, the performance monitoring scheme of distributed container application is designed and implemented, and the alarm function is designed and implemented to assist the monitoring service. Through this system, users can directly monitor the application status and resource usage of the Docker container cluster deployed on the Docker platform, and solve the problem that the cluster of Docker containers can not be monitored as a whole at present. Make the application deployed on the Docker cloud platform more stable and robust, and provide a basis for application performance tuning. In addition, this article analyzes the current Docker container log management deficiencies. Fluentd and MongoDB components are used to encapsulate and redesign, Etcd technology is used to record the complex relationship of log data, and message queue mechanism is designed to improve the performance of log transmission. Design and implementation of database fragmentation replica mechanism to provide a highly available database cluster, to make up for the current Docker cloud platform application log lack of integrity management shortcomings, the implementation of the application in multi-node. The collection, transmission, summary storage and fast retrieval of the complex logs generated in multiple containers provide the basis for the operation and maintenance tuning and debugging of Docker container applications. In this paper, the functions of the system are tested and tested. The contents of this paper have a certain reference significance for the integration and improvement of Docker platform services.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09

【相似文献】