Improved L7-Filter’s Pattern Matching Algorithm Based On Multi-core Processors

Along with computer network technology in the political, military, commercial and otherfields used widely, and rapid growth of the network traffic, which makes the network securityproblem increasingly prominent. As the core of the network security, packet’s classificationtechnology is particularly important. L7-Filter is a packet analysis tool based on packet’sapplication layer data on linux system platform, which ultimately determines the classification ofpacket. However, as the network of large-scale deployment of multi-core processors, L7-Filterdoen’t show a significant performance on multi-core environment.To solve those problems, this article first introduces the overview of multi-core computingsystem structure, analyse L7-Filter architecture and its performance bottleneck on multi-coreprocessors. Finally, this article puts forward a kind of classifications method for the classfyingtarget rule chains, which according to the network data’s time statistical properties to dynamicoptimize rules chain for improving data packet matching algorithms. And through experimentand the improved algorithm of the results were compared, found that the improved algorithmfully utilize the same connections within a particular time, make alterations of data packetsarithmetic performance had about30percent of the improved. Experimental results show that,with the constantly increasing, packets flow after making the learning the rules of reachingstability, chain of packet processing speed also shows certain growth and become stable.