Research And Implementation On Technology Of IP-level Network Topology Discovery

With the rapid development of Internet technology and application, today’sInternet has become increasingly large and complex. For the infrastructure ofInternet is constructed and operated by a large number of ISPs, there is a lack ofknowledge and understanding of the overall Internet topology. It helps to find thebottlenecks of the network、optimize network configuration、improve networkmanagement level by conducting measurement and analysing the IP-level networktopology; At the same time, it helps to find and understand the basic characteristicsof the network topology、improve the accuracy of network simulation and decision-making of network system by analysing data on network infrastructure modeling.Firstly, this thesis describes the development status and the relevantbackground of network topology discovery. It also describes the most commonlyused network topology measurement techniques: SNMP-based and traceroute-basednetwork topology measurement techniques. It is found that the former belongs tocollaborative measurement method, so the measurement accuracy is relatively high.However, it need the collaboration of network managers to carry out themeasurement, and the range is limited. The latter obtains the connection informationby throwing a large number of probe packets into the network. It needs no networkaccess authorization and is suitable for Internet topology measurement, but it canonly get a snapshot of the network during a short time and has limited measurementaccuracy.Secondly, to further improve the integrity of multi-point network topologymeasurement and reduces the affection of the number and deployment location ofmeasurement points, this thesis presents and implements a service-based third-partyIP-level network topology discovery system–thirdTrace. It can effectively utilizethe world wide third-party sources for topology discovery. Then it makes a detaileddescription of the four modules of the thirdTrace-measuring point collectionmodule、result extraction module、results finishing module、topology generationand analysis module. it point out that this design of modular makes the goodscalability of thirdTrace.At last, this thesis compare the topology discovery capabilities of thethirdTrace with that of iplane and analyze the basis characteristics of chinese network topology measured by both. It is found that the result of thirdTrace containsa richer network core links and edge topology information. It shows moreobservable dissortativity, weaker clustering, shorter distance and moredisequilibrium of node traffic distribution. Meanwhile, we compare them with thechinese topology founded in2004and find the changes of scale、clustering、nodetraffic distribution and other characteristics of chinese network.