發(fā)布時間：2018-01-23 14:34

  本文關(guān)鍵詞： 云計算 Docker 容器 性能監(jiān)控 日志　出處：《浙江大學(xué)》2016年碩士論文　論文類型：學(xué)位論文

【摘要】：作為新興的云平臺,Docker打破了傳統(tǒng)以虛擬機(jī)為基礎(chǔ)的云計算形式,以容器為基本單位為開發(fā)人員提供構(gòu)建、發(fā)布和運(yùn)行分布式應(yīng)用所需的平臺。目前,在性能監(jiān)控和日志服務(wù)方面,Docker平臺僅僅支持對單個容器信息的命令查詢,功能相對欠缺,無法從整體上為分布式應(yīng)用提供完善的性能監(jiān)控方案,這也限制Docker平臺上應(yīng)用的發(fā)展前景。首先,本文基于Docker平臺現(xiàn)有功能,對Docker容器的性能監(jiān)控服務(wù)進(jìn)行了研究與完善,設(shè)計實(shí)現(xiàn)了分布式容器應(yīng)用的性能監(jiān)控方案,并設(shè)計實(shí)現(xiàn)了警報功能用于輔助監(jiān)控服務(wù),用戶通過本系統(tǒng)即可直觀地監(jiān)控其部署在Docker平臺上的應(yīng)用狀態(tài)及資源使用情況,解決了當(dāng)前無法對Docker容器集群進(jìn)行整體監(jiān)控的問題,使部署在Docker云平臺上的應(yīng)用能更穩(wěn)定、健壯地運(yùn)行,并為應(yīng)用的性能調(diào)優(yōu)提供了基礎(chǔ)。此外,本文分析了當(dāng)前Docker容器日志管理上的不足,利用Fluentd和MongoDB組件進(jìn)行封裝再設(shè)計,使用Etcd技術(shù)記錄日志數(shù)據(jù)的復(fù)雜關(guān)系,設(shè)計實(shí)現(xiàn)消息隊列機(jī)制以提高日志傳輸?shù)男阅?設(shè)計實(shí)現(xiàn)數(shù)據(jù)庫分片副本機(jī)制以提供高可用的數(shù)據(jù)庫集群,彌補(bǔ)了當(dāng)前Docker云平臺上的應(yīng)用日志缺乏整體性管理的弊端,實(shí)現(xiàn)了對應(yīng)用在多節(jié)點(diǎn)、多容器中所產(chǎn)生的復(fù)雜日志的收集傳輸、匯總存儲及快速檢索,為Docker容器應(yīng)用的運(yùn)維調(diào)優(yōu)及排錯提供了基礎(chǔ)。最后,本文對系統(tǒng)的功能進(jìn)行了測試檢驗(yàn)。本文中的內(nèi)容對Docker平臺服務(wù)的集成和完善有一定的參考意義。
[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.
【學(xué)位授予單位】：浙江大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：TP393.09

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 杜黨黨;賈曉亮;張宇;;航空發(fā)動機(jī)性能監(jiān)控系統(tǒng)設(shè)計與實(shí)現(xiàn)[J];航空制造技術(shù);2014年07期

2 諸昕;王瑜;李s��，

本文編號：1457771