8 × 16 TB in RAID 5: the numbers
| Usable space | 112 TB (101.9 TB in Windows) |
|---|---|
| Raw space | 128 TB |
| Used for protection | 16 TB |
| Efficiency | 88% |
| Drives that can fail | Any 1 |
| Rebuild at 150 MB/s | 29 h |
| Rebuild at 50 MB/s (busy) | 3.7 days |
| Data read during rebuild | 112 TB |
| Read error risk, desktop drives | over 99% |
| Read error risk, NAS drives | 59% |
Rebuild
Replacing a failed drive means writing a full 16 TB drive. At 150 MB/s with no other load that takes about 29 h; on a busy array running at 50 MB/s, about 3.7 days. The rebuild reads every surviving drive: 112 TB in total.
Read errors
Reading 112 TB, the chance of at least one unreadable sector is over 99% with desktop drives (1 per 1014 bits) and 59% with NAS drives (1 per 1015). Nothing is left to correct it, so the affected file, or on some controllers the whole rebuild, fails. These are worst-case datasheet rates; real drives usually do better.
Our take
With 16 TB drives we would pick RAID 6. One more drive in RAID 6 gives the same 112 TB and survives two failures. Keeping 8 drives, RAID 6 gives 96 TB. RAID 10 with the same drives gives 64 TB, faster writes and shorter rebuilds.