新型网络环境下目标探测技术研究

发布时间：2018-09-16 19:53

【摘要】：随着互联网的迅速发展,新型网络服务及应用的出现使得网络规模逐渐发生了变化。伴随着网络规模的日益扩大,网络各方面的要求也越来越高,尤其对于网络管理方面所面临的难度越来越大。网络拓扑探测是网络管理的基础,为网络管理人员能够更好地了解网络的各方面情况,更好地管理优化整个网络起到决定性作用。作为一种近年来兴起的新型网络应用模式,云计算利用网络将存储设备和服务器相连接,创建一个资源池以实现资源共享。由于云计算网络环境的特点,外部网络无法实现对云环境内部的请求访问,同一计算节点虚拟机使用的网络协议相同,无法通过基于协议特征或探测数据包时间差的方法获取云计算内部虚拟机的分布,因此适用于传统网络环境的拓扑探测方法无法完成新型云计算网络环境的拓扑探测。本文对新型云计算网络环境下目标探测技术中拓扑探测进行了研究,完成工作主要包括:1.分析了传统网络环境下基于协议的拓扑探测方法和层析成像探测方法,介绍了云计算模式、云计算虚拟化技术,以及OpenStack相关知识;2.在分析研究已有探测方法的基础上,提出分层拓扑探测方法,将新型网络分为两层分别采用不同的技术进行探测,第一层包括云网络外部的路由器/交换机、云网络计算节点以及云网络外部路由,主要采用层析成像探测方法进行拓扑探测,第二层包括云计算网络内部计算节点上的各类虚拟机,主要采用基于TCP/IP协议的Ping探测包进行拓扑探测;3.基于上述所提出探测方法,设计一个原型系统,在系统第一层中,采用基于TCP协议和UDP协议的两种“三明治”包实现拓扑探测。在系统第二层中,创建的虚拟机之间发送不同大小的Ping探测包,采用MPI进行时延数据采集,并利用并查集算法进行数据聚类,实现拓扑探测,最终设计原型系统证明了所提拓扑探测方法的有效性。
[Abstract]:With the rapid development of the Internet, the emergence of new network services and applications has gradually changed the scale of the network. With the increasing expansion of the network scale, the requirements of all aspects of the network are becoming higher and higher, especially for the network management is more and more difficult. Network topology detection is the basis of network management, which plays a decisive role for network managers to better understand the various aspects of the network and better manage and optimize the entire network. As a new network application model, cloud computing uses the network to connect storage devices and servers to create a pool of resources to achieve resource sharing. Because of the characteristics of cloud computing network environment, the external network can not realize the request access to the cloud environment, and the same computing node virtual machine uses the same network protocol. The distribution of virtual machines in cloud computing can not be obtained by the method based on protocol characteristics or packet time difference, so the topology detection method suitable for the traditional network environment can not complete the topology detection of the new cloud computing network environment. In this paper, the topology detection of target detection technology in the new cloud computing network environment is studied. In this paper, the protocol based topology detection method and tomography detection method in traditional network environment are analyzed. Cloud computing model, cloud computing virtualization technology and OpenStack related knowledge are introduced. Based on the analysis and study of the existing detection methods, a hierarchical topology detection method is proposed, which divides the new network into two layers and uses different techniques. The first layer includes the router / switch outside the cloud network. Cloud network computing nodes and cloud network external routing, mainly using tomography detection method for topology detection, the second layer includes all kinds of virtual machines on the cloud computing nodes. This paper mainly adopts Ping probe packet based on TCP/IP protocol for topology detection. Based on the above detection method, a prototype system is designed. In the first layer of the system, two "sandwich" packets based on TCP protocol and UDP protocol are used to realize topology detection. In the second layer of the system, different sizes of Ping packets are sent between the virtual machines created, the delay data is collected by using MPI, and the data clustering algorithm is used to realize topology detection. Finally, the prototype system is designed to prove the effectiveness of the proposed topology detection method.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.02

【相似文献】