For two independent restorable items A and B that together perform a function, with dangerous failures detectable, the asymptotic hazardous functional failure rate may be approximated as the product of the two failure rates multiplied by the sum of their safe down times (for identical items twice the square of the failure rate times the safe down time), valid only if the rates are constant, every fault that could be hazardous in combination is detected and negated, common-cause failures that are hazardous may be ignored and the product of failure rate and safe down time is much less than one; the assumptions are justified for each application. Two items of MTBF 10,000 hours with a safe down time of one hour give 2 times 10 to the minus 8 per hour; with detection by maintenance at 1,000 hours only 10 to the minus 5, and the approximation itself fails.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.