Optimization Of String Match Algorithm In Large Flow Network

String match has been a classic research direction in the field of computerscience. During the IPv6protocol replace IPv4protocol process, due to theexpansion of the address space, more devices can be connected, so that there willproduce more data on the Internet. To the information security system, expandingthe patterns set, network traffic increasing, they all request the string matchingalgorithm put forward a higher performance.This article first explains the background of the research, studies a variety ofclassical string match algorithm theory, and gives their strengths and weaknesses,as well as for the occasion. Then it introduces the current situation of the increasi ngpopularity of multi-core processors, due to the traditional string match algorithmdoes not migration to multi-core processors flexibly., this article presents a parallelstring match algorithm based on DFA, PSMBD, it avoids some defects of commonparallel methods, through classification on the patterns set by the first character, itallows the processor to handle all the core to process the text on the same time, andwill not be conflicted, improving the speed. Since using the automatic machinestructure, the common compression methods of memory for automatic machine canbe transplanted in the optimization. Subsequently, in accordance with the views putforward before: It should optimize string match algorithm according to thecharacteristics of the patterns set in different application, it analyzes thecharacteristics of the patterns set and the text for matching of the informationsecurity system, and the it gives a string match algorithm based on the thoughts ofclassification, AWXF, which combines the properties of AC algorithm and WMalgorithm, dividing the pattern set in accordance with the length of each pattern,making the time-consuming and memory cost on the balance. Especially for WMalgorithm part, it proposes an optimized method for the large-scale patterns set,playing a key role on increasing the speed.Finally, this article carries the performance tests for the two algorithms above.Through experimental data, it shows that both in the case of low or high hit rate, thePSMBD algorithm has a good performance. In the testing process of AWXFalgorithm, first through the offline tests on different patterns set with the previousgeneration algorithm on comparison of initialization time, memory usage andmatching time and then through the online test, with the previous generationalgorithm on CPU utilization compared, through experimental data, it shows thatAWXF algorithm is better than the previous generation algorithms.