RAID data-loss risk calculator
How likely is your array to actually lose data? This uses the real annual failure rate (AFR) from Backblaze - not a vendor MTBF brochure number - plus your RAID level and how long a rebuild takes.
A large-drive resilver often takes 12-48h+ under load. Fleet-average AFR is 1.36%.
Estimated risk
chance of array data loss over 5 years
about 1 in 19734 arrays like this
Very low - the array itself is unlikely to be your point of failure.
What this does and does not tell you
The model treats an array as lost when one drive fails and enough more drives fail, inside the rebuild window, to exceed your parity. The big levers are the per-drive AFR, the parity (RAID 6 tolerates two failures, RAID 5 only one), and how long a rebuild leaves you exposed.
It deliberately understates real risk: it assumes drives fail independently and ignores two things that bite in practice - unrecoverable read errors during a rebuild (a real hazard on large drives) and correlated failures from a same-batch or same-environment set. Treat the number as an order-of-magnitude floor, not a guarantee.
Takeaways that hold up: prefer RAID 6 / double parity as drives get large, keep arrays from getting too wide, pick low-AFR drives (sort the reliability rankings), and remember RAID is not a backup. For usable capacity, use the RAID capacity calculator.