Design And Implementation Network Real-Time Analysis Platform

With the rapid expansion of network, network management is facing unprecedented challenges. The traditional network management software is poor scalable, it’s network data analysis mode is fixed, that users can’t customize data analysis mode and so on; To solve these problems, this paper proposes a real-time network analysis system and method of supporting big data, to compensate for deficiencies in this area.For defects of existing network analysis software, this paper presents a network analysis system designed like "assembly line", combined with P2P technology and event handling mechanism. Similar to the "assembly line", each data analysis process consists of several relatively independent data analysis nodes to complete, and each data analysis node communicates on the unified event mechanism with each other. And also, we adds logic management module, which users can use to extend troubleshooting needs easily by the generic template interface.To achieve the above design, the paper uses two kinds of data nodes to complete the system functions:data analysis node and data adapter node. Data adapter node mainly completes the adaptation between network data and internal events, and we can use data analysis node’s scalability and custom logic to complete the "assembly line" construction, to complete task to the monitoring data analysis and other operations. In this paper, we use class diagrams and flowcharts to elaborate achieve. Finally we use the test of checking timeout packets to complete functional verification and performance test.The system features as this paper describe:First, According to user’s own requirements, they can custom data analysis logic through the interface we provide. And only inject it into the container to take it effect; second, the system can improve performance by dynamically expanding nodes. With the support of pipeline working way and event routing mechanism, we only need to add the server like building block, the data adapter node will dynamically reload the data to the new server. So, we achieve the requirements of improving the performance by dynamic expanding nodes.