Drive Failure Math — URE, MTBF & RAID Survival Probability
RAID protects against drive failure — but only if the rebuild succeeds. Hidden deep in every hard drive specification is a small number that can derail your entire array: the Unrecoverable Read Error (URE) rate. This guide unpacks the math behind UREs, explains why consumer drives become dangerous at 12 TB+, and shows you exactly how rebuild survival probability changes with drive size, count, and drive class.
What Is a URE (Unrecoverable Read Error)?
Every hard drive has a specification called the Unrecoverable Read Error rate. This is the manufacturer's statistical guarantee of how many bits you can read before encountering a sector that cannot be recovered, even after all built-in ECC and retry mechanisms are exhausted.
Consumer Drives (1014)
Rated for 1 unrecoverable error per 1014 bits read. In human terms, that is one error per 12.5 terabytes of data read. This is the standard for desktop and external drives.
Enterprise Drives (1015)
Rated for 1 unrecoverable error per 1015 bits read — 10× better. That is one error per 125 terabytes read. Found in NAS/enterprise models like WD Gold, Seagate Exos, and Toshiba MG-series.
Crucial insight: Modern drives are 12 TB, 16 TB, even 22 TB. Reading a single 12 TB drive from end to end consumes nearly your entire consumer URE budget. In a RAID rebuild, you read all remaining drives — and that is where the math gets scary.
Rebuild Survival Probability — The Math
When a drive fails in a RAID 5 array, the controller reads every byte on every surviving drive to reconstruct the missing data. The probability of surviving the rebuild without hitting a single URE is:
P(survival) = (1 − URE_rate)bits_read
Where bits_read = (number_of_surviving_drives) × (drive_capacity_in_bits)
Because URE rates are tiny and the number of bits read is enormous, even "small" arrays can have surprisingly high failure probabilities. The table below shows your chance of hitting at least one URE during a RAID 5 rebuild with consumer (1014) drives:
Drives in RAID 5
4 TB Each
8 TB Each
12 TB Each
16 TB Each
22 TB Each
3 drives (2 survive)
~6%
~12%
~17%
~22%
~29%
4 drives (3 survive)
~9%
~17%
~25%
~32%
~41%
5 drives (4 survive)
~12%
~22%
~32%
~41%
~52%
6 drives (5 survive)
~15%
~27%
~38%
~49%
~60%
Interpretation: With 4 consumer 12 TB drives in RAID 5, there is roughly a 25% chance that the rebuild will hit a URE and fail. With 6 drives at 22 TB, that rises to 60%. This is why RAID 5 with large consumer drives is widely considered risky.
Why Enterprise Drives Matter for Large Arrays
Switching from consumer (1014) to enterprise (1015) drives multiplies the amount of data you can read before hitting a URE by 10. The difference in rebuild survival probability is dramatic:
Configuration
Consumer (1014)
Enterprise (1015)
Improvement
4× 8 TB RAID 5
~17% URE risk
~1.9% URE risk
9× safer
4× 12 TB RAID 5
~25% URE risk
~2.8% URE risk
9× safer
6× 16 TB RAID 5
~49% URE risk
~6.7% URE risk
7× safer
8× 22 TB RAID 6
~48% URE risk*
~6.4% URE risk*
7.5× safer
* RAID 6 can survive a URE during rebuild thanks to the second parity drive, so the data loss risk is lower than the raw URE probability suggests.
Enterprise drives typically cost 20–40% more than their consumer counterparts, but the reliability benefit is transformative. For any array with drives 8 TB or larger, the math overwhelmingly favors enterprise drives — especially if you value your data more than the upfront savings.
MTBF vs. URE — Two Different Failure Modes
Drive reliability is specified with two independent metrics that are often confused:
MTBF (Mean Time Between Failures)
The average operational hours before a drive fails completely. Typical values: 500,000–2,000,000 hours. A 1,000,000-hour MTBF does not mean a drive lasts 114 years — it is a statistical mean across a large population. About 1% of drives per year fail at 876,000-hour MTBF. MTBF predicts total drive loss, not sector-level errors.
URE (Unrecoverable Read Error)
The rate at which a drive encounters a sector it cannot read. This is the dominant risk factor during RAID rebuilds, because a single lost sector = a stripe with failed parity = data loss for RAID 5. URE rates are not improved by buying drives with higher MTBF — they are a separate specification tied to areal density and error correction firmware.
Key takeaway: A drive with 2,000,000-hour MTBF and a 1014URE rate is still risky for a large RAID 5 rebuild. MTBF tells you how often a drive dies; URE tells you whether you can recover when one does.
Frequently Asked Questions
What is a URE (Unrecoverable Read Error)?
A URE happens when a drive's read head cannot recover a sector despite all built-in error correction. The drive reports the sector as lost. Consumer drives are rated 1-in-1014 bits (~12.5 TB), while enterprise drives are 10× better at 1-in-1015 (~125 TB). During a RAID rebuild, a single URE on any surviving drive can cause data loss in RAID 5.
Why is RAID 5 dangerous with large drives?
A RAID 5 rebuild reads every surviving drive fully. With 4× 12 TB consumer drives, you are reading 36 TB during the rebuild — nearly 3× the URE budget of 12.5 TB. The probability of hitting a URE is ~25%. For 6× 22 TB drives, the probability exceeds 60%. This is known as the "RAID 5 rebuild gap" — the point where drive capacity has outpaced URE reliability improvements.
Does RAID 6 eliminate the URE problem?
RAID 6 greatly reduces the risk. With two parity drives, it can survive a single URE during rebuild (the second parity reconstructs the affected stripe). However, RAID 6 does not eliminate the problem entirely: a second actual drive failure during the longer rebuild window, or multiple UREs across different stripes, can still cause data loss. For 6+ drive arrays, RAID 6 or RAID 10 is the safe choice.
Are enterprise drives worth the extra cost for a home NAS?
For arrays with 3+ drives at 8 TB or larger, yes. Enterprise drives (1015 BER) reduce URE risk by roughly 10× compared to consumer drives. The price premium of 20–40% translates into dramatically higher rebuild survival probability. For smaller arrays (2–4 TB drives) or RAID 10 configurations, the benefit is less pronounced and consumer drives may be sufficient.
What is the difference between MTBF and URE?
MTBF predicts how often a drive fails completely (total loss). URE predicts how often a drive encounters an unreadable sector. During a RAID rebuild, URE is the more important metric because a single lost sector can cause array failure. A drive can have excellent MTBF but poor URE rates, making it risky for large RAID 5 arrays. Always check both specifications.
Run the Numbers Yourself
Don't guess — calculate your actual rebuild survival probability based on your drives, RAID level, and array size. Our interactive tools show you the exact math: