Generating Diagnositc Tools for Network Fault Management

Today's network management applications mainly collect and display information, while providing limited information processing and problem-solving capabilities. A number of diierent knowledge-based approaches have been proposed to correct this deeciency, evolving from rule-based systems through case-based systems, to more recent model-based systems. Part of this evolution has been the recognition of the importance of constraints in a management context. This makes possible the assimilation into network management of a mature, theoretically developed technology from artiicial intelligence, namely, the constraint satisfaction problem (CSP). In this paper we investigate the role of constraints in manipulating management data, and give an example of the use of the constraint satisfaction framework in diagnosing problems arising with Internet domain name service conngurations. We also present ADNET, a system for automatically constructing C++ diagnostic programs from a model written in a simple modeling language.