Real-Time Network Traffic Classification Based on CDH Pattern Matching

Abstract

In recent years, with the rapid development of the Internet, the data scale of application behavior and application traffic have exploded. How to classify the real-time traffic of network becomes a big challenge. How to balance the accuracy and real-time of traffic classification is a difficult problem in technology. Therefore, this paper proposes a pattern matching real-time traffic classification method named PM, which first uses jpcap to accept network traffic data in real time, and then uses pattern matching to perform real-time matching traffic characteristics to achieve traffic classification. Among them, the use of the distributed message system kafka and the parallel computing framework Spark significantly improve the execution efficiency of the program. The experimental results show that PM has good performance in terms of accuracy.