Title :
Off-line diagnosis of parallel systems
Author :
Benkahla, Oum-El-Kheir ; Robach, Chantal
Author_Institution :
LCIS/ESISAR-INPG, Valence, France
Abstract :
This paper presents an off-line diagnosis strategy for parallel message-passing systems. This strategy, called host-diagnosis, allows an external observer, i.e. the host system, to perform centralized diagnosis of the system state, given results of distributed tests performed among the system processors. Three algorithms that use the host-diagnosis strategy are proposed. The performance of the three algorithms are evaluated and compared to those of a classic distributed self-diagnosis algorithm. The obtained results show an interesting behaviour of the host-diagnosis algorithms in comparison with the self-diagnosis one
Keywords :
distributed algorithms; message passing; program diagnostics; software performance evaluation; centralized diagnosis; distributed self-diagnosis algorithm; distributed tests; host system; host-diagnosis; parallel message passing systems; parallel systems offline diagnosis; performance evaluation; system processors; Assembly systems; Automatic testing; Electronic mail; Fabrication; Fault detection; Maintenance; Manufacturing; Performance evaluation; System testing; Time to market;
Conference_Titel :
Reliable Distributed Systems, 1998. Proceedings. Seventeenth IEEE Symposium on
Conference_Location :
West Lafayette, IN
Print_ISBN :
0-8186-9218-9
DOI :
10.1109/RELDIS.1998.740524