Pro-active network management using data mining

Network providers are continually looking towards the future, perhaps the year 2002, when network management systems will possess the capability of proactively changing network configurations and traffic routing, in real-time, in order to optimise traffic throughput in the network, and hence maximise their profit margin. However, the major barrier to the production of truly real-time pro-active network management systems appears to be the technology, or technique, which will enable the vast amount of network performance data, which is routinely collected, to be analysed, and for pro-active system changes to be made, in real-time. The rapidly emerging field of data mining appears to offer a solution to this problem. We discuss the potential suitability of the four main data mining categories of data mining algorithms, to the task of analysing network performance data in true real-time. Hence we provide a discussion regarding the classes of data mining algorithm which will provide the basis for the provision of a real-time pro-active network management system by virtue of their ability to rapidly mine vast amounts of data.