Data Mining Research And Application In IP Network Fault Management Database

With the IP network management system development, it becomes an important part of ISPs, enterprises and campus netwotk. Meanwhile, the information stored by the IP network management system has increased greatly with the expansion of the network, it is a new research area that how to analyse these large quantity of data in decision support for network management. The data mining technology are studied and introduced to the fault management of IP network management in order to provide the data mining for the network administrator, with the help of the technology in the fault association, the administrator can work easily and more efficiency.The characteristic of fault data in IP network management system is studied and analyzed. Based on the mission of the fault association, several technologies, which may suitable for the mining in the fault data, are analyzed. According to different purposes, we choose the Association analysis for the relation of alarms, Apriori algorithm, Eclat algorithm and MTAEM algorithm are compared in this paper, Apriori algorithm and MTAEM algorithm are choosed after the comparasion. On anther hand, we choose the Time Sequence analysis for the time relation between alarms. After analyzing the characteristic of several time sequence technologies, DTWMH algorithm.is selected.Based on analysis before, a data mining prototype about fau sequence technologies, DTWMH algorithm.is selected.Based on analysis before, a data mining prototype about fault data is designed and implemented. The system includes a complete data mining process which are data preprocess, data mining and the result presentation. The test results show that the Association analysis and the Time Sequence analysis are effective for alarm association analysis in IP Network management system; it can reduce the administrator's work intensity, and make them work more efficiency.