Fuzzy logic for distributed systems troubleshooting

Automated distributed systems management is important to trading floor support because it improves service quality while reducing support costs. Systems management includes two key functions: first, it monitors the network to identify deficient service, and second, it diagnoses the monitored data to ensure the timely and least costly resolution of an identified deficiency. These tasks are difficult to achieve concurrently because each is bound by an inherent conflicting constraint imposed by the network. The conflict was settled by developing a real-time automated alarm interpretation system that uses fuzzy message clustering and averaging techniques acting in concert with a network monitoring platform. In particular, an intelligent network management facility was implemented integrating both network monitoring and diagnostic tasks. The architecture of the intelligent diagnostic system is described. The fuzzy knowledge base design is presented. An operation example is discussed. The system reduces the operator's message management workload by half. The new and improved computer-generated advice replaces conventional systems monitoring alarms.<>