RAID for Home NAS: Which Level, How Many Drives

1. What RAID Buys You (and What It Doesn't)

RAID keeps your system running when a drive dies — that's the whole product. Data stays online, you swap the dead drive, the array rebuilds, life continues. What RAID does not do: protect against accidental deletion, ransomware, file-system corruption, a failed controller, fire, or theft — every one of those cheerfully replicates across all your redundant drives. The mantra 'RAID is not a backup' survives because people keep learning it the expensive way. Plan RAID for availability and a separate backup (another machine, cloud, or offline drive) for survival, and size your drive budget for both from day one.

2. The Levels That Matter at Home

Four configurations cover essentially all home use. RAID 1 (mirror, 2 drives): simplest, loses half your capacity, survives one drive failure — right answer for 2-bay NAS units. RAID 5 / RAIDZ1 (3+ drives): one drive's worth of parity, survives one failure — the classic 4-bay choice, with a caveat covered below. RAID 6 / RAIDZ2 (4+ drives): two drives of parity, survives two simultaneous failures — the community default for arrays of large drives. And honorable mention: SnapRAID or mergerfs-style setups for media libraries that grow drive by drive. RAID 0 has no redundancy and no place holding data you'd miss.

3. The Big-Drive Rebuild Problem

Here is the argument that moved the community from RAID 5 to RAID 6 as drives grew: when a drive fails, the rebuild must read every sector of every surviving drive — on 16TB+ drives, that's a day or more of maximum sustained load on aging disks that just lost a sibling from the same batch. If a second drive fails, or even hits an unrecoverable read error mid-rebuild, a RAID 5 array is done. RAID 6's second parity drive exists precisely to survive that window. Practical rule: RAID 5/Z1 is defensible up to roughly 4–8TB drives with backups in place; at 10TB and above, spend the extra bay on RAID 6/Z2. This is also the strongest argument against SMR drives in arrays — rebuilds are sustained-write hell.

4. The Capacity Math (What You Actually Get)

Usable capacity = (number of drives − parity drives) × drive size, and all drives contribute at the size of the smallest member. Four 12TB drives in RAID 6 yield 24TB usable, not 48TB. Then subtract file-system overhead and the fact that a decimal-TB drive formats to less than its label — plan on ~90% of the RAID-level number being really usable, and keep pools under ~80% full for performance. When comparing purchase options on our price table, run the math on usable-TB-per-dollar for your chosen level: sometimes five mid-size drives in Z2 beat four large ones on both cost and safety, and sometimes the reverse — the answer moves with market prices, which is the entire reason to track them.

5. Buying Strategy: Start Right, Grow Sanely

Expansion is where RAID choices bite later. Traditional RAID and ZFS historically wanted all drives upfront (ZFS RAIDZ expansion now exists but grows capacity one drive at a time with caveats), while Synology's SHR and unRAID tolerate mixed sizes and gradual growth — a genuine reason home users pick them. Strategy that serves most people: buy all array drives at once when a good per-TB price appears, ideally from mixed batches or vendors so a bad manufacturing lot can't take out multiple members, size the array for three years of growth, and keep one tested cold spare. Then set up SMART monitoring and scheduled scrubs — an array nobody watches is just a slow-motion data loss.

Quick Decision Guide

Frequently Asked Questions