Research On Network Fault Management System Based On Event Mechanism

As the scope of the network continuously expands, the internal network faultfactors are more and more. Massive fault events will be produced and events storm willbe formed making the management and maintenance of the network quitely complicated.It is difficult to solve the network faults rapidly and accurately by the traditional toolsand methods. The available network fault management systems are poor with function,expansibility and flexibility. Therefore, it is of important and practical significance toconstruct a automatic and intelligent network fault management system in this paper.Firstly, the paper analyzes the architecture of Network Management System. Thetheories on Fault Management, Expert System, Data Mining are studied. Thearchitecture of the system is designed and each module of the system is managed basedon the events mechanism. Secondly, both the information collection model and the faultjudgment model are designed and implemented respectively to judge whether themonitored object produces faults. Based on the knowledge base and the reasoningmachine of Expert System, the models of the fault filtering and the fault diagnosis aredesigned to solve the problems about filtering and diagnosing the fault events. Thispaper focuses on the design of the model to mine the association rules among the faultevents and the fault event association tree. The events are filtered effectively based onthe fault event association tree. The case reasoning strategy of XNMS_CBR is furtherresearched to find the best similar case, with the solution of which to repair the fault.The design and implementation of the Agent which can operate and control the networkequips are detailed discussed. Finally, in the paper the prototype system of network faultmanagement is established and all the function modules are tested with the test casesrespectively. The result shows that the system is flexible and efficient and can quicklylocate and solve the network faults, ensuring the running of the network normally andeffectively.