Research On Key Technologies In Large-scale Network Traffic Measurement And Realization

The rapid development of IP networks makes that it is gradually becoming people's main communications infrastructure.With the rapid expansion of the scale of IP network, the increasing of transmission rate, increasingly complex of network structure, so that we need take some measurement for performance and behavior of IP network, and make an accurate assessment of the state of IP network. Network measurement is to provide reference for improving network performance, optimizing the network structure and design planning of network through the measurement of network traffic and network performance. Due to the rapid development of high-speed Internet technology, we need a more efficient approach to measure the current network traffic , thereby manage the network better.In this thesis, in the background of the Internet, the theory and technology of network traffic measurement were discussed, and have take some detailed study about the history of the development of network traffic measurement and research achievement of every period.In this thesis, we deeply studied the technology of internet traffic measurement, and analyze indicator system and standards of evaluation of network traffic measurement, and discussed the characteristics of passive and active measurements,then studied the architecture of network traffic measurements.From point of view of large-scale network measurement of traffic and performance, carried out the research a number of important indicators for flow measurement such as packet loss rate, transmission delay, data sampling technique, finaly network packet capture and some related technologies, including NDIS and WinPcap were studied in detail.In this thesis, Based on those studies and analysis of traffic measurement technology, flow measurement architecture and technology of network packet capture, then design a passive and real-time measurement model of network traffic.and summrized up two key technologies for a highly efficient and large-scale flow measurement system, namely high-speed packet capture technology and flow classification algorithm. Then mainly focused on two key technologies, and proposed a solution to these two key technologies. The solution is based on the technology of WinPcap to achieve high-speed network packet capture and Hicuts algorithm for the flow classification algorithm. After researching and resolving these key technologies, this thesis described designment and implementation of submodules and the overall architecture modules of the large-scale network traffic measurement system in detail, and took the test for the system, the test results show that the flow measurement system implemented packet capture for high-speed Network with lower packet loss rate, reaching the requirements of performance for high-speed network packet capture, and complete traffic classification functionality. Finally, we gave a summary of this thesis, and proposed further research that it is necessary to do.