NODEX: A KNOWLEDGE BASED EXPERT NODE MANAGER FOR HP9000 LAN

Abstract

Current day computer and telecommunication networks are very complex with diverse equipments being used in them, and varied network services being offered. The software need- ed to create, manage, and operate these networks is becoming more complex. Even the best human network managers wield a type of sorcery, mixing folklore, experience and incomplete data, to somehow operate networks successfully. Knowledge based expert systems can handle this complexity with ease. The programmed rules of expert systems can be effectively used to solve many of network manager's problems. Expert systems, in fact, are changing the programming landscape in network management, with currently operational systems demonstrating the ability to anticipate trouble and learn from experience. Encouraged by the positive impact of expert systems in easing network management, we are now, striving to develop a network management system for the HP9000 LAN. This dissertation presents an effort to develop NODEX, a knowledge based expert node manager for HP9000 LAN. NODEX has been coded in GCLISP. A forward-chaining problem-solver forms the heart of NODEX. NODEX can be used for 1)Upkeeping a node (HP9000 series 200 or 300 computer) on the HP9000 LAN 2)Diagnosing and troubleshooting network problems on the node computer.