Research And Realization Of Key Technologies In High Performance Acquisition Software

With the continuous development of Internet technology, which has penetrated into lots of aspects of people's lives and subtle changes in people's lifestyles. Internet companies are springing up. The major companies in order to get a greater share of profits need to provide users with more quality services, all these lead to the birth of "Big Data" era.By analyzing the behavior of the user characteristics, Internet suppliers can be more targeted to provide services, but as the business continues to expand, the network link carrying traffic is on the growing. The data acquisition system needs to be able to process the fleeting data and generate the required field bills for big data platform analysis. With the increase speed of network link, network traffic monitoring meets a greater challenge. Network traffic monitoring, which based on data packet collection, can get more effective network planning management and maintenance with the help of excellent performance of the acquisition system.In this thesis, the topic of research is focus on the key technologies in the design of the high-performance acquisition software system. Firstly,the basic framework of the acquisition software system is introduced and lead to the key points in the system design. The performance testing tool Intel VTune will be introduced too. We design and implement three kinds of association algorithms for HTTP packages and flow packages. We also experimentally test the correlation performance of different correlation algorithms. In the performance optimization of the output module in the acquisition system, we mainly design different formats for different types of fields and test their performance. Regular expression is widely used owing to its outstanding string-matching filtering performance. In this thesis, we analyze the basic work procedure of the regular matching engine, and design experiments to test wellknown regular matching library. We implement Intel Hyperscan, which has a better performance,in the analysis of HTTP packets procedure. Finally, we summarize the research contents and prospect for future researches.