Configuration detection as a problem of knowledge discovery in computer networks

Presents a type of knowledge discovery that is slightly different from data mining in database systems-configuration detection in computer networks. In computer networks, one needs to know which objects exist and where they are located. We call the acquisition of this knowledge "configuration detection". We are interested in hosts and routers on an internetwork within an organization. As internetworking has become popular, the growth in the number of network objects has been substantial and dynamic. In such an environment, the maintenance of knowledge of the current configuration of the network has become a practical problem. The scale problem has been solved either by facilitating dynamic learning or by dividing the management domain into subdomains of a manageable size. In either solution, we face the problems of inconsistency and invalidity. We suggest the use of a procedure for the authorisation of network addresses to solve some of the serious network-level threats using a security technique. Finally, we apply the solution to address resolution.