Design And Implementation Of Intelligent Network Management System

Along with the explosive growth of the application of computer networks, the internet and intranet have been developed rapidly. How to effectively ensure the network's normal and efficient operation of network and how to reduce the losses caused by various network faults are serious problems that we are currently facing with. Network Management System (NMS) is committed to solving these problems.The TCP/IP protocol suite is the most widely used computer network transmission protocol in the world. In order to manage the TCP/IP-based internet effectively, the SNMP protocol is introduced by IAB. At present, most of network devices support the SNMP protocol.In the market, most of the products of NMS are based on the traditional centralized network management, but their intelligent and automated degree is not sufficient to solve the problems of network management. The centralized network management has two inherent shortcomings: Firstly, NMS is responsible to gather and process all the information. However, this has a higher request on the configuration of the management station. With the rapid expansion of the scale of the network, this solution is becoming more and more unfeasible. Secondly, all the collected information and commands are sent or received by the same management station. If the collection frequency is too high, it will bring tremendous burden to the whole network.For all the problems mentioned, this paper proposes a structure of B/S-based hierarchical distributed system which not only successfully overcomes the shortcomings of the centralized NMS but also increases the scalability and flexibility of NMS considerably.Network topology is the infrastructure of the whole network management. If you want to manage the whole network and make the best use of network resources, you should have a comprehensive understanding to the whole network. Therefore, this paper proposes a SNMP-based automatic topology discovery algorithm.Additionally, Fault management is one of the most important parts of NMS. It is still a challenging problem in NMS domain to localize the fault as fast and accurately as possible and recover the system in time. Hence, this paper proposes an ECCA-based fault localization algorithm, which improves the accuracy and intelligence of the fault localization and management respectively.