Low overhead multiprocessor allocation strategies exploiting system spare capacity for fault detection and location

摘要: Several schemes for detecting faults at the processor level in a multiprocessor system have been discussed past. One such scheme (A. Dahbura et al., 1989) works by running secondary versions of jobs on unused or spare processors and uses comparison approach (J. Maeng M. Malek, 1981) to detect faults. We build upon this propose three new allocation strategies that run variable number per job. These permit online detection and, many cases, location faulty with nominal degradation its delay/throughput performance; these delays are limited chiefly associated job preemptions. Two metrics, fault capability (FDC) (FLC), introduced evaluate schemes. Extensive simulation results performed obtain performance figures various Stochastic Petri net models also developed approximate results. The show utilize capacity more efficiently, thereby improving capabilities system. >