While updating the hardware reliability section of an asset-management document, you look up the vendor's specification for the hot-swap fan module installed in 50 identical rack servers. The spec lists a mean time between failure (MTBF) of 300 000 h for each fan. Assuming a constant failure rate and 24 × 7 operation, which number of fan-module failures is the most realistic to record as expected during the next three years of production?
For a constant failure rate, the failure rate λ of one component is the reciprocal of its MTBF:
λ = 1 / 300 000 h ≈ 3.33 × 10⁻⁶ failures/h.
Three years of continuous service equal 3 × 365 × 24 = 26 280 h. The expected failures for one fan in that period are
26 280 h × 3.33 × 10⁻⁶ ≈ 0.088 failure.
With 50 identical fans, the expectation scales linearly:
50 × 0.088 ≈ 4.4 failures.
Rounded to the nearest whole unit, documenting about four fan-module failures over three years is the most accurate forecast. The smaller figure underestimates demand, while the larger figures assume a failure probability several times higher than that implied by the MTBF.