容器云平台的设计与实现

发布时间：2018-05-30 02:03

本文选题：云计算 + Kubernetes集群　；参考：《北京邮电大学》2017年硕士论文

【摘要】：以前,开发人员开发应用需要自己实现复杂的框架保证应用的高可用和故障替换。测试人员测试应用需要自己重新部署测试环境,需要繁琐的安装依赖库和配置参数。运维人员部署应用更是头疼,经常碰到异构的操作系统,对不同系统进行不同的部署方式。应用的容器化,可以更加便捷的创建分布式应用,加快应用的开发与上线,减低运维成本。基于容器技术可以很容易地把应用部署在异构的平台上,很轻松地面对应用的迁移。Docker作为容器技术事实上的标准,可以减少运维的成本,加快项目的开发。对传统的应用系统以容器化的方式部署,利用容器技术把应用打包成Docker镜像,可以屏蔽环境的不一致,让开发、测试、部署同步进行。Kubernetes谷歌开源的容器管理系统,可以高效的管理容器集群,并提供一系列的自动部署、服务监控、高可用、监控、故障替换、负载均衡等一系列技术,让用户部署的容器集群更加高效的利用集群资源。使用Kubernetes提供的解决方案,可以节省开发成本,同时让开发、运维人员将精力更加集中于业务本身,运维难度和成本大幅度降低。OpenStack提供强大的基础设施管理功能,共享计算资源、存储资源和网络资源,可以灵活的分配资源。OpenStack每个组件都是高可用部署,可以为上层提供稳定可靠的资源服务。本课题通过 OpenStack、Docker 和 Kubernetes 的研究,基于 Openstack 创建 Kubernetes集群,构建容器云平台。论文从用户的角度分析需求和设计架构,设计和实现容器云平台,方便用户部署和管理Docker应用。利用OpenStack为Docker提供弹性资源分配,为Kubernetes集群提供用户隔离和同用户不同集群的网络隔离。在Kubernetes集群上部署Docker应用,利用Kubernetes方便容器应用的管理。设计实现容器资源配置模块的调用接口,融合OpenStack和Kubernetes,通过操作Kubernetes集群配置容器资源,为容器的部署提供环境和资源。论文对构建的容器云平台进行了测试,符合设计目标。
[Abstract]:Previously, developers needed to implement their own complex frameworks to ensure high availability and failover. Testers test applications need to redeploy their own test environment, require cumbersome installation dependencies and configuration parameters. Operation and maintenance personnel deployment is a headache, often encountered heterogeneous operating systems, different systems to deploy different ways. The containerization of applications makes it easier to create distributed applications, speed up the application development and go online, and reduce the cost of operation and maintenance. Container based technology can easily deploy applications on heterogeneous platforms. It is easy to face the migration of applications. Docker is the de facto standard of container technology. It can reduce the cost of operation and maintenance, and accelerate the development of projects. Traditional applications are deployed in a containerized manner, packaging applications into Docker images using container technology, which can mask inconsistencies in the environment and allow the development, testing, and deployment of the container management system, which synchronizes the development, testing, and deployment of Google's .Kubernetes, into an open source container management system. It can efficiently manage container cluster and provide a series of technologies such as automatic deployment, service monitoring, high availability, monitoring, fault replacement, load balancing and so on, so that the container cluster deployed by users can make more efficient use of cluster resources. Using the solution provided by Kubernetes, you can save the development cost, at the same time, let the development, operation and maintenance personnel focus more on the business itself, and greatly reduce the difficulty and cost of operation and maintenance. OpenStack provides powerful infrastructure management functions, sharing computing resources, Storage resources and network resources, can be flexibly allocated resources. OpenStack each component is highly available deployment, can provide stable and reliable resource services for the upper layer. Through the research of Open Stackman Docker and Kubernetes, this paper creates Kubernetes cluster based on Openstack and constructs container cloud platform. This paper analyzes the requirements and design framework from the user's point of view, designs and implements the container cloud platform, which is convenient for users to deploy and manage Docker applications. OpenStack is used to provide flexible resource allocation for Docker, user isolation for Kubernetes cluster and network isolation for different user clusters. Deploy Docker application on Kubernetes cluster and use Kubernetes to facilitate container application management. The calling interface of container resource configuration module is designed and implemented, which combines OpenStack and Kubernets, and provides the environment and resources for container deployment by using Kubernetes cluster to configure container resources. In this paper, the container cloud platform is tested, and it accords with the design goal.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【参考文献】