Design And Implementation Of Traffic Classification System Based On Cavium Platform

Internet's appearance has brought the enormous convenience to the people. With the rapid development of the network, the internet has already gradually instead of the traditional television and telephone business, becoming the main tool of communication and the information acquisition channel. In recent years, along with the day-by-day renewal of network, the enlargement of the network bandwidth and the rapid growth of the network application, the supervision and control of the network content also face the huge challenge. And the increasing of application has caused the increasing difficulty of gaining agreement characteristic. The appearance of many kinds of information encryption technology also greatly reduced the accuracy of the traditional agreement recognition technology. At the same time, the suddenly increasing of the network bandwidth has a huge influence on the traffic classification ands efficiency of the protocol identification. Because the architecture's innate factor limit of the traffic classification system, which was based on traditional general processor's traffic classification, already unable to adapt to the request of the present high speed network processing. It urgently needs introducing the new hardware construction and the processing plan, to satisfy the changed network environment and quickly adapted the higher network speed request.This article is to aim at the whole solution of the network traffic classification system. It is considered from the hardware platform, system construction, parallel processing, and system optimization and so on. It has designed and realized the new network traffic classification system which is met the present condition and new request of the network.This article achieves the traffic classification system which is based on Cavium platform and the foundation of traditional analysis and research. It carries on the optimum composition based on each method's accurate degree and the efficiency. It adopts the different recognition way in view of the different protocol. The speed and the accuracy of the traffic classification system surpasses the sole traffic classification system approach. Simultaneously, using the Cavium hardware platform's characteristic to accelerate the network traffic classification system, including adopts polynuclear parallel technology, the assembly line technology, the hardware DFA technology and so on. It also using the special hardware platform to increase the efficiency of traffic analysis system and realize the network traffic classification under the great data quantity network environment.