Work out mtbf and mttr instantly with clear inputs, formula shown and shareable results.
Availability is MTBF divided by MTBF plus MTTR, so it improves either by failing less often or by recovering faster — and recovery is usually the cheaper lever. Parallel redundancy compounds the result: two units each at 99.9 percent give 99.9999 percent provided their failures are genuinely independent.
Availability from MTBF and MTTR
MTBF = operating hours / failures; MTTR = repair hours / failures; A = MTBF / (MTBF + MTTR); with N units A = 1 - (1 - A)^N
MTTR, almost always. Hardware reliability is largely fixed by the vendor, while spares on site, documented procedures and automated failover can cut repair time by an order of magnitude.
Because failures correlate. Shared power, shared firmware bugs and a common cooling failure defeat the independence assumption entirely.