DOFSON - a distributed fault management system for open system networks

This paper presents a distributed fault management system called DOFSON for open system networks. The proposed model overcomes the inadequacies of existing methodologies for detecting errors and facilitates on-line fault isolation in an operational network. Moreover this distributed approach to fault management results in reduced network management compared to centralized models. DOFSON is modelled as a set of fault managers and agents residing in all the open systems in the network. Each fault manager is an application process responsible for managing faults in the network. Control is distributed among all the fault managers in the network. Management agents present in all the open systems provide the services required by each fault manager for performing its tasks. Software monitors detect and report errors to local fault manager. Faults are diagnosed and isolated by performing a series of diagnostic tests to check the functionality of the suspected components. A model case study discussing the detection of cascaded faults is presented to demonstrate the applicability of DOFSON.