当前位置:主页 > 管理论文 > 移动网络论文 >

基于OpenStack开源云平台的网络模型研究

发布时间:2019-02-14 14:13
【摘要】:随着云时代的到来,各种基于IAAS层次的云管理平台应运而生例如亚马逊的EC2和S3的云基础架构等而围绕着开源的优势,OpenStack云平台得到越来越多的企业与个人开发者的支持.OpenStack其目的在于在于简化云平台的部署过程,其组件交互式的架构为其带来了良好的扩展性 云计算是基于虚拟化基础上的产物,对于计算ǐ存储ǐ网络都有着一切皆服务的趋势对于OpenStack云平台亦是如此,从网络的发展角度来看,早期的网络组件Nova_network结构过于简单,不能完全满足客户需求而且随着规模的扩大化,其性能要求也不能够满足遵循着网络即服务的理念,Quantum网络模型被提出来,其集中体现为SDN模型的一种演化型方式:云租户能够通过Quantum提供的API创建属于自己的网络拓扑,同时能够配置虚拟网络等 对于云计算网络来说,SDN(Software Defined Network)模型已然成为一种趋势,Quantum体现了一种SDN实现方式,支持了多租户云,它将虚拟的网络世界从服务器端扩展到用户端但是Quantum网络也表现出很大的缺点,其一是它所创建的虚拟网络位于网络层之上,其连接的终端是虚拟机层次,而不是更高层次的物理设备,其二,其支持的Vlan数量有限,并且对于网络数据流的负载均衡并不支持 针对Quantum网络的这些现象以及目前网络发展的现状:由于TCP/IP网络体系结构已经根深蒂固,面对其设计之初的缺陷,通过更多的协议来修补该体系结构无法从根本上解决问题,反而随着更多协议的运行,网络设备负载更重,导致会出现很多问题另外的一种SDN模型——OpenFlow模型从网络控制层面抽象出数据流的概念解决了这些问题与以往的网络架构不同,OpenFlow用控制层与转发层分离的架构来完成数据流的转发工作在其控制层面上,我们完全可以通过组件开发验证创新的网络体系和协议 本文结合自己搭建OpenSatck云部署过程,,详细分析了OpenStack早期网络模式,然后讨论Quantum网络的优点与缺点,并结合OpenFlow模型的发展趋势,将其以插件的形式融合在该平台上OpenStack网络在该SDN模型的基础之上,考虑到OpenFlow协议的缺点,进行路由实现 首先针对现有的网络体系结构分析数据包传输过程,并找出OpenFlow模型不能路由的原因 然后对OpenFlow协议的动作进行处理,合理的完成ARP报文和ICMP报文的处理,并且在OpenFlow控制器端实现链路发现组件与路由组件 最后在OpenFlow模拟平台Mininet与Nox控制器进行实验测试与验证,并对当前的研究做了总结与展望,希望后来的研究人员在该基础上进一步的深入
[Abstract]:With the advent of the cloud age, a variety of cloud management platforms based on the IAAS hierarchy emerge as the times require, such as Amazon's EC2 and S3 cloud infrastructure, and around the advantages of open source. The OpenStack cloud platform is supported by more and more enterprise and individual developers. OpenStack aims to simplify the deployment process of the cloud platform. Its component interactive architecture brings it a good scalability cloud computing is a product based on virtualization, and there is a trend for computing storage and storage networks to be all services, as is the case with the OpenStack cloud platform. From the perspective of network development, the early network component Nova_network structure is too simple to fully meet customer needs, and with the expansion of the scale, its performance requirements can not meet the concept of network as a service. Quantum network model is put forward, which is embodied as an evolutionary way of SDN model: cloud tenants can create their own network topology through API provided by Quantum, and can configure virtual network for cloud computing network. SDN (Software Defined Network) model has become a trend, Quantum embodies a SDN implementation method, supports multi-tenant cloud, it extends the virtual network world from server to client, but Quantum network also shows great shortcomings. One is that the virtual network it creates is located on top of the network layer, and the terminals it connects to are virtual machines rather than higher-level physical devices, and it supports a limited number of Vlan. And the load balancing of network data flow does not support the phenomenon of Quantum network and the current status of network development: because the TCP/IP network architecture has been deeply rooted, facing the defects of its initial design, Fixing the architecture with more protocols does not fundamentally solve the problem. Instead, as more protocols run, the network equipment becomes more heavily loaded. There will be a lot of problems. Another SDN model, the OpenFlow model, abstracts the concept of data flow from the network control level and solves these problems differently from the previous network architecture. OpenFlow uses the architecture of separate control layer and forwarding layer to complete the data stream forwarding. On its control level, we can fully verify the innovative network architecture and protocols through component development. In this paper, we can build a OpenSatck cloud deployment process. This paper analyzes the early network mode of OpenStack in detail, and then discusses the advantages and disadvantages of Quantum network. Combining with the development trend of OpenFlow model, the OpenStack network is fused in the form of plug-in on the platform on the basis of the SDN model. Considering the shortcomings of OpenFlow protocol, the routing implementation first analyzes the process of packet transmission according to the existing network architecture, and then finds out the reasons why the OpenFlow model can not be routed, and then processes the actions of OpenFlow protocol. The processing of ARP and ICMP packets is completed reasonably, and the link discovery components and routing components are implemented in the OpenFlow controller. Finally, experimental tests and verification are carried out on the OpenFlow simulation platform Mininet and Nox controller. It also summarizes and prospects the current research, and hopes that the later researchers will further deepen the research on this basis.
【学位授予单位】:吉林大学
【学位级别】:硕士
【学位授予年份】:2014
【分类号】:TP393.01

【参考文献】

相关期刊论文 前5条

1 侯长逸;;OpenFlow网络软件路由研究[J];兰州大学学报(自然科学版);2013年02期

2 苗秀;俞俊生;刘绍华;陈晓东;;基于云计算平台的移动IPTV系统设计及负载均衡技术研究[J];软件;2011年01期

3 周烨;李勇;王芳;杨旭;苏厉;金德鹏;曾烈光;;基于OpenFlow的网络实验平台技术[J];清华大学学报(自然科学版);2012年11期

4 左青云;陈鸣;赵广松;邢长友;张国敏;蒋培成;;基于OpenFlow的SDN技术研究[J];软件学报;2013年05期

5 李英壮;汪楠;李先毅;;基于OpenStack开发云平台的设计与实现[J];武汉大学学报(理学版);2012年S1期



本文编号:2422278

资料下载
论文发表

本文链接:https://www.wllwen.com/guanlilunwen/ydhl/2422278.html


Copyright(c)文论论文网All Rights Reserved | 网站地图 |

版权申明:资料由用户1e587***提供,本站仅收录摘要或目录,作者需要删除请E-mail邮箱bigeng88@qq.com