Research And Design Of Distributed Network Management System Based On Multi-Agent

With the rapid development of computer network technology, the network is expanding. Then the structure of network becomes more complex. The importance of network management is more significant because of the increase of network service data. Traditionally, network management is based on centralized SNMP. It can implement easily and its expansibility is fine. But with the network's expanding,the defects are more serious. Once the administrator have some problems,the whole management system will collapse.In order to meet the requirements of the new network, network management using Agent bacomes a new research recently. The Agent have a feature of moving between heterogeneous nodes, and partically completes network management with the other agents through communications. According to the task assigend by network management center, Agent get and process data in their jurisdiction and transmit strcture to management center actively. Then the load and cost of network bandwidth are greatly reduced and the efficiency of network management is improved.Based on the recent research of Agent Technology in network management, this paper focuses on the research of network flow load and the Agent's applications in network management. The balancing of network load uses ant colony, and network simulation is achieved. Network management makes use of Agent technology and distributed approach of regional management. A management model based on multi-Agent is proposed. The network management based on multi-Agent use distributed and autonomous architecture, balance network load and relieve network congestion to some extent, then improve the effiency of network management. In network management, Agent can monitor and manage network equipments, and play a great role of saving resource of manager and network bandwidth,then solve the bottleneck of network managers effectively. Agent also can acquisite, analysize and process network data real-time and improve timeliness of network management.