RAID Rebuild Time Chart by Drive Capacity
A RAID rebuild takes roughly one to three days on a modern large drive. A 16 TB drive rebuilding at a realistic 80 MB/s takes about 55 hours, and about 111 hours on an array still serving files. Rebuild time scales with the capacity of the single replaced drive, not the total size of the array.
Rebuild time is the risk figure nobody quotes when selling you a NAS, and it is the entire reason single parity stopped being a sensible default on large drives. While an array rebuilds it is degraded: on RAID 5 or RAIDZ1 there is no parity left, so a second failure during that window destroys everything.
Every number below is computed from capacity and a stated throughput rather than typed in, so you can see and disagree with the assumption instead of taking it on faith.
How long does a rebuild actually take?
| Drive | Busy array 40 MB/s | Typical home 80 MB/s | Idle array 150 MB/s |
|---|---|---|---|
| 4 TB | 1.2 days | 13.9 hr | 7.4 hr |
| 6 TB | 1.7 days | 20.8 hr | 11.1 hr |
| 8 TB | 2.3 days | 1.2 days | 14.8 hr |
| 12 TB | 3.5 days | 1.7 days | 22.2 hr |
| 14 TB | 4.1 days | 2.0 days | 1.1 days |
| 16 TB | 4.6 days | 2.3 days | 1.2 days |
| 18 TB | 5.2 days | 2.6 days | 1.4 days |
| 20 TB | 5.8 days | 2.9 days | 1.5 days |
| 22 TB | 6.4 days | 3.2 days | 1.7 days |
Rates are sustained rebuild throughput, not spec-sheet sequential numbers. A rebuild competes with everything else the array is doing, which is why the busy column is roughly half the idle one. Rebuilding while serving media and backups. Light use during the rebuild. Nothing else touching the disks.
Why rebuild time is a risk number, not a patience number
A degraded array is an array with its safety margin spent. On RAID 5 or RAIDZ1 there is no remaining parity during the rebuild, so the window in the table above is the length of time a single additional failure wipes out the pool.
Two things make that worse than it sounds. Drives in an array were usually bought together, ran identical hours, and sit in the same thermal environment, so their failure probabilities are correlated rather than independent. And a rebuild is the heaviest sustained read the array ever performs, which is precisely the workload most likely to surface a marginal drive.
The read error problem that ends RAID 5 on big disks
A RAID 5 rebuild must read every remaining bit of data to reconstruct the missing drive. Consumer drives are specified at roughly one unrecoverable read error per 10^14 bits. Across several large disks that is not a remote possibility.
| Drive | Array | Data read on rebuild | Consumer drive risk | NAS rated risk |
|---|---|---|---|---|
| 4 TB | 4 drives, RAID 5 | 12 TB | 62% | 9% |
| 4 TB | 8 drives, RAID 5 | 28 TB | 89% | 20% |
| 8 TB | 4 drives, RAID 5 | 24 TB | 85% | 17% |
| 8 TB | 8 drives, RAID 5 | 56 TB | 99% | 36% |
| 14 TB | 4 drives, RAID 5 | 42 TB | 97% | 29% |
| 14 TB | 8 drives, RAID 5 | 98 TB | 100% | 54% |
| 20 TB | 4 drives, RAID 5 | 60 TB | 99% | 38% |
| 20 TB | 8 drives, RAID 5 | 140 TB | 100% | 67% |
Read the last two columns together. Moving from a desktop drive at 10^14 to a NAS rated drive at 10^15 is a full order of magnitude, and it is most of why NAS drives are worth their premium. It is a far better argument for them than the vibration tolerance usually printed on the box.
These are probabilistic estimates from published specifications, not predictions about your hardware. Real drives frequently outperform their specified error rate, and a single read error does not always fail a rebuild on modern filesystems. Treat the direction as reliable and the exact percentage as indicative.
What this means for the layout you should pick
- Under about 8 TB per drive: single parity is defensible if you have a real backup. Rebuilds finish inside a day and the read exposure stays modest.
- 8 TB and above: use double parity, RAID 6 or RAIDZ2. The rebuild runs for days and single parity leaves you unprotected for all of it.
- 16 TB and above: double parity is not optional. A rebuild pushing three days on a busy array with no parity remaining is not a risk worth taking to recover one drive's worth of capacity.
Work out what each layout costs you in usable space with the RAID capacity calculator, which compares Synology SHR, ZFS RAIDZ levels and standard RAID side by side rather than locking you to one vendor's answer. The RAID levels guide covers the tradeoffs in full.
RAID is not a backup, and a rebuild is when you find out
Everything above describes how long you are exposed, not how protected you are. RAID survives hardware failure; it does nothing about deletion, corruption, ransomware, theft or fire, all of which replicate happily across a healthy array. The 3-2-1 backup rule is what actually protects the data, and a rebuild is the single worst moment to discover your backup was stale.
Related
Frequently asked questions
How long does a RAID rebuild take?
Roughly one to three days for a modern large drive, and longer on a busy array. A 16 TB drive rebuilding at a realistic 80 MB/s takes about 55 hours, and the same drive on an array that is still serving files can take twice that. Rebuild time scales with the capacity of the single replaced drive, not with the total size of the array, which is the detail most people get backwards.
Why is RAID 5 considered risky on large drives?
Because a RAID 5 rebuild must read every bit on every surviving drive, and it has no parity left to protect it while doing so. With consumer drives specified at one unrecoverable read error per 10^14 bits, a full read of several large disks has a meaningful chance of hitting one, and in classic RAID 5 that can fail the rebuild. RAID 6 and RAIDZ2 survive it because a second parity copy is still available.
Does a bigger array take longer to rebuild?
Not directly. Rebuild time is driven by the capacity of the one drive being rebuilt, so an eight bay array of 8 TB drives rebuilds a single 8 TB disk, not 64 TB. What a larger array does increase is exposure: more drives means more chances that one of them fails during the window, and more surviving data that must be read without error.
Do NAS rated drives actually reduce rebuild risk?
Yes, and this is the strongest argument for paying the premium. NAS rated drives are typically specified at one unrecoverable read error per 10^15 bits against 10^14 for desktop drives, a full order of magnitude better. Across the multi-terabyte read a rebuild performs, that difference moves the chance of a read error during rebuild from uncomfortable to negligible.
Can I use the array while it rebuilds?
You can, and it will roughly halve your rebuild speed while extending the window in which a second failure is fatal. If the data matters, let the rebuild run unimpeded and take the downtime. If it truly cannot go offline, plan for a rebuild that takes two to three times the idle estimate and make sure your backup is current before you start.
What should I do while a rebuild is running?
Verify your backup first, because a rebuild is the moment an array is least protected. Then reduce load, avoid starting large writes, and do not power cycle the unit. If a second drive throws errors mid-rebuild, stop and restore from backup rather than pushing on, because a failed rebuild on a degraded array often ends worse than a clean restore.
How we choose: we compare published manufacturer specifications, drive datasheets, published reliability statistics and verified owner reviews. We do not test hardware in person, and we are not tied to any NAS vendor. Capacity and power figures here are researched guidance, not a warranty. RAID protects against drive failure, not against deletion, ransomware, fire or theft, so keep verified backups regardless of what any calculator tells you.
Working out your own cost per usable terabyte? The Home Server Build Planner is the paid version of these pages: 8 printable worksheets you fill in with your own numbers, plus the full PDF, $29.