The increasing importance of computer networks in today's society demands h
igh levels of network availability and reliability. This is particularly ap
parent for the enterprises that perform their transactions depending highly
on e-commerce. However, a trivial network fault could cause costly damages
to the enterprises because of fault propagation. In this paper, we offer a
fault diagnosis mechanism for effective and automated network fault isolat
ion. This mechanism uses finite state machine (FSM) to model our classified
and refined fault propagation behavior. A concept called fault propagation
duration is also introduced into the mechanism and combined with the FSM-b
ased fault propagation knowledge to complete the mechanism's implementation
. Based on the diagnosis mechanism, an automated fault diagnosis system cal
led alarm correlation view (or ACView) for isolating network faults is prop
osed. This diagnosis system not only provides the process of automated alar
m collection, alarm correlation, and fault isolation, but also functions to
recognize the propagations of all the faults with their corresponding faul
t severities. (C) 2001 Elsevier Science B.V. All rights reserved.