Implementation Of A User Behavior Analysis And Web Push System Based On Multi-Core Processing And Dpi

With the rapid development of Internet, the application of it was constantly applied into People’s Daily life and brought great convenience to people. In addition to chat, games and other applications, network shopping and advertising are increasingly brought to the attention of the masses of users and service providers. In order to help users to quickly locate the needed information effectively, and help enterprises to advertise to different consumer groups, user behavior analysis and web push technologies start to emerge in endlessly.There are two common implementation scheme of user behavior analysis and web push from the angle of the operators. One of the solutions is based on x86architecture. It mainly through the software module implements the analysis of the whole network traffic and subsequent processing. Although the solution can meet different kind of operational requirements, but the general processor in the network processing can produce a large number of interrupt, seriously affect the performance of the system. Another solution achieves different functions by dedicated hardware integration. It guarantees the throughput of the system. But the function of the system has been solidified, when demand changes, its function has been unable to change.In this paper, after fully studied the OCTEON CN5860chip, we designed and implemented a user behavior analysis and web push system based on the network multi-core processing architecture. The system uses the Deep Packet Inspection technology. And through the Internet business recognition and keyword matching, we set up a weighted keywords frequency statistics method to build the behavior model. Then we implemented the web push function based on the analysis results. The system makes full use of the hardware module to achieve the efficient DPI. And it adopted flexible software design at the same time. The system provides configuration interface of the behavior analysis and web push function used to meet changing business needs.System in the design is divided into two parts. The part on the main communication channel achieves the user behavior analysis and web push function. The bypass system display network data collected from the main line in real-time and offers concise and friendly configuration interface to help the manager to change the web push strategy. According to the function and performance test, the whole system is running smoothly and stably. It achieved accurate behavior analysis and flexible web push under a high load. The implementation has reached the expected goal.