Large-Scale Network Traffic Bottleneck Analysis

With the expansion of network size and the increase of services provided, the rapid development of the Internet brings us a lot of convenience. But, this can increase the probability of various kinds of network congestion, and network monitor has to face more challenges. Network traffic bottleneck analysis is one of most important part of network management. It is of vital role in improving network availability, reliability and the network quality of service.Rapid, large, varied dimensions are the characteristic of the large-scale network. Presently, the network bottleneck measure and monitor only by analysis the delays and bandwidth. This thesis presents a new method to analyze network bottle-neck on the large-scale network traffic, implement network traffic monitor function, improving the capability of detecting network traffic bottle-neck.First, we introduced the basic conception of network traffic, including development background and current research of network traffic, then put more emphasis on illustrating collection technology and analysis technology, by comparing the relative technology to find out the characteristics and shortcomings.Second, the thesis presents two methods of traffic analysis, one is based on threshold and another is based on Primary Component Analysis (PCA). Based on the Abliene traffic data which Totem issued, we made a test, in which we choosing the key-link using two ways respectively, on this basis, make bottle-neck analysis, in the end, the two results are very similar, validated both of them are reasonable.Finally, we present a large-scale network traffic bottle-neck detection system frame model, which is made up of traffic data collection and pre-disposal model, key-link analysis model, bottle-neck analysis model, synthesis analysis and visualized model. Several practices have been made to prove that the model is effective and practical.