Abstract: A novel scheme of network traffic data collection and storage is proposed in detail, which resolves the difficult problems of traffic monitoring in large-scale networks. Traffic prediction is applied in the network management system so as to implement the network traffic prediction, which is a change of the traditional network management method. Firstly, the SNMP-based method of obtaining traffic data was adopted so as to improve the computing accuracy. Secondly, the comparison and analysis were performed in different schemes of traffic monitoring. Thirdly, the ARMA-based prediction was implemented and analyzed. Finally, the definition of extended TRAP types was also presented to handle the alerts that are sent because of breaking the threshold. As shown in our practice, the system can meet the requirements of traffic monitoring in large-scale telecom data networks with good efficiency.

Key words: computer network, network management, traffic monitoring, traffic prediction, SNMP