Agreed - the assumption is that the MTBF of a single network device is a relevant statistic is undoubtedly incorrect in anywhere but a hobbyist's network.