【摘要】：云計(jì)算的概念涵蓋了資源的抽象整合,動(dòng)態(tài)流通和高效利用等各個(gè)方面,旨在將資源從物理環(huán)境抽象出來(lái),進(jìn)行靈活的使用。在云平臺(tái)上,通過(guò)統(tǒng)一的接口可以進(jìn)行資源的管理、監(jiān)控、調(diào)度工作,整合了松散的功能,提供了大量的服務(wù)。傳統(tǒng)的云平臺(tái)使用虛擬機(jī)作為基本調(diào)度單位,而本文的PaaS平臺(tái)使用Docker容器來(lái)代替虛擬機(jī)。Docker是基于LXC(Linux Container)發(fā)展而來(lái)的容器技術(shù),和虛擬機(jī)相比,從資源規(guī)模上來(lái)看,容器的粒度更小。從靈活性上來(lái)看,容器的部署時(shí)間更短,可移植性更強(qiáng)。因此,使用Docker來(lái)改造云平臺(tái)更加的貼合未來(lái)的需求。本文使用Docker容器作為底層的虛擬化技術(shù)設(shè)計(jì)了一個(gè)PaaS平臺(tái)。分析了PaaS相關(guān)技術(shù),進(jìn)行了集群整體設(shè)計(jì),提出了具體設(shè)計(jì)方案,包括Mesos資源調(diào)度,Docker網(wǎng)絡(luò),NFS存儲(chǔ),容器資源限制等。分析了平臺(tái)的功能需求,設(shè)計(jì)實(shí)現(xiàn)了各個(gè)功能模塊。本平臺(tái)使用容器作為應(yīng)用節(jié)點(diǎn),使用Spark處理計(jì)算請(qǐng)求。實(shí)現(xiàn)了一個(gè)應(yīng)用發(fā)布平臺(tái),其中使用Docker容器作為統(tǒng)一開(kāi)發(fā)環(huán)境,SVN管理代碼,Jenkins完成集成和構(gòu)建,docker-compose和Marathon完成應(yīng)用的編排和部署。該平臺(tái)還具備了基于負(fù)載均衡和彈性伸縮的高可用性,負(fù)載均衡通過(guò)Nginx代理,彈性伸縮策略是自主設(shè)計(jì)的。使用CAdvisor進(jìn)行資源監(jiān)控,提供統(tǒng)一視圖,同時(shí)對(duì)平臺(tái)的日志系統(tǒng),用戶(hù)管理系統(tǒng)等進(jìn)行了設(shè)計(jì)。對(duì)彈性算法進(jìn)行改進(jìn),優(yōu)化了系統(tǒng)。最后,部署系統(tǒng),測(cè)試功能。測(cè)試的內(nèi)容包括彈性測(cè)試和應(yīng)用發(fā)布測(cè)試,分析了測(cè)試結(jié)果,驗(yàn)證了平臺(tái)的可用性。本平臺(tái)利用Docker容器技術(shù)的優(yōu)點(diǎn),在傳統(tǒng)PaaS平臺(tái)的基礎(chǔ)上進(jìn)行了一定的底層改造和功能補(bǔ)充,為云平臺(tái)的設(shè)計(jì)提供了新的思路。
[Abstract]:The concept of cloud computing covers the abstract integration of resources, dynamic circulation and efficient utilization, aiming at abstracting resources from the physical environment and using them flexibly. On the cloud platform, resources management, monitoring and scheduling can be carried out through a unified interface, which integrates loose functions and provides a large number of services. The traditional cloud platform uses virtual machine as the basic scheduling unit, while the PaaS platform in this paper uses Docker container instead of virtual machine. Docker is a container technology developed based on LXC (Linux Container). The container has a smaller granularity. In terms of flexibility, containers have shorter deployment times and greater portability. Therefore, the use of Docker to transform the cloud platform more in line with future needs. This paper uses Docker container as the underlying virtualization technology to design a PaaS platform. This paper analyzes the related technologies of PaaS, designs the whole cluster, and puts forward the specific design scheme, including Mesos resource scheduling, Docker network, NFS storage, container resource limitation and so on. The functional requirements of the platform are analyzed, and each functional module is designed and implemented. This platform uses container as application node and Spark to process computing request. An application publishing platform is implemented, in which Docker container is used as a unified development environment, SVN manages code, Jenkins is integrated and built, and docker-compose and Marathon are used to orchestrate and deploy applications. The platform also has high availability based on load balancing and elastic scaling. Load balancing is designed by Nginx agent. CAdvisor is used to monitor resources and provide unified view. At the same time, the logging system and user management system of the platform are designed. The elastic algorithm is improved and the system is optimized. Finally, deploy the system and test the function. The content of the test includes elastic test and application release test. The test results are analyzed and the usability of the platform is verified. This platform makes use of the advantages of Docker container technology, and on the basis of the traditional PaaS platform, it has carried on the certain bottom layer transformation and the function supplement, has provided the new thought for the cloud platform design.
【學(xué)位授予單位】：華北電力大學(xué)(北京)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP393.09

【參考文獻(xiàn)】

相關(guān)期刊論文 前8條

1 李麗新;;基于Mesos和Docker構(gòu)建大數(shù)據(jù)平臺(tái)技術(shù)研究與應(yīng)用[J];吉林建筑大學(xué)學(xué)報(bào);2016年05期

2 飛揚(yáng);;Docker遇上Mesos 輕量級(jí)PaaS的最佳演繹[J];互聯(lián)網(wǎng)周刊;2016年10期

3 柯尊旺;于炯;廖彬;;適應(yīng)異構(gòu)集群的Mesos多資源調(diào)度DRF增強(qiáng)算法[J];計(jì)算機(jī)應(yīng)用;2016年05期

4 蔡云龍;;基于Docker的私有云系統(tǒng)的設(shè)計(jì)[J];電子產(chǎn)品世界;2016年04期

5 伍陽(yáng);;基于Docker的虛擬化技術(shù)研究[J];信息技術(shù);2016年01期

6 劉思堯;李強(qiáng);李斌;;基于Docker技術(shù)的容器隔離性研究[J];軟件;2015年04期

7 陳清金;陳存香;張巖;;Docker技術(shù)實(shí)現(xiàn)分析[J];信息通信技術(shù);2015年02期

8 鄧維;劉方明;金海;李丹;;云計(jì)算數(shù)據(jù)中心的新能源應(yīng)用:研究現(xiàn)狀與趨勢(shì)[J];計(jì)算機(jī)學(xué)報(bào);2013年03期

，

本文編號(hào)：2373949