RAID is unavoidable when you run a NAS: it determines your usable capacity, read/write performance, and the boundary of your data safety. This guide starts from zero — which level to pick, how to calculate capacity, and which “common knowledge” is actually a myth — with methods you can verify yourself using the 17NAS RAID calculator.
1. What problem does RAID actually solve?
RAID (redundant array of independent disks) combines several drives intomore capacity, more speed, or more fault tolerance. But remember one iron rule first:
RAID is not a backup. It protects you against “one drive dies and your data is still there”. It does not protect you against accidental deletion, ransomware, or a lightning strike that takes out the whole machine. For genuinely important data, follow the 3-2-1 rule (3 copies, 2 media types, 1 offsite).
2. The common RAID levels in one table
Taking 4 × 12TB drives as the example (you can verify the usable capacity yourself with the RAID calculator):
RAID 5: the home sweet spot
Four 12TB drives in RAID 5 give you 36TB usable, and the array survives any single drive failure — the default answer for a home NAS with three or more bays. The trade-off: during a rebuild (which can run 1–2 days with large drives) there is a risk of a second failure, so avoid heavy I/O while it runs.
RAID 6 / RAID 10: many bays, or performance
With eight or more drives, or 16TB+ large drives, RAID 6 (which tolerates two failures) is the steadier choice. For random-I/O-heavy workloads such as databases and virtual machines, RAID 10 is faster.
SHR: the Synology user's easy answer
SHR is essentially an automated RAID 5/6. Its biggest advantages are thatdrives of different capacities still use their space efficiently and that swapping drives later to expand is flexible. On Synology hardware you can pick it without thinking; on non-Synology systems, use OMV's “linear + redundancy” or standard RAID.
3. Capacity maths: don't be fooled by “12TB”
There are three layers of shrinkage you should know about up front:
- Vendor TB vs system TiB: a 12TB drive shows as about 10.9TiB in the OS (a base-2 vs base-10 difference, not short weight)
- RAID redundancy overhead: RAID 5 costs you one drive's capacity, RAID 6 costs two
- Filesystem and snapshot reserve: with Btrfs and snapshots, leave another 10% or so
Example: 4×12TB RAID 5 ≈ 36TB vendor rating ≈ 32.7TiB actually shown ≈ plan for about 29TiB of usable data. Pick a level and drives in the RAID calculator and you get the real usable capacity, the RAID 10 grouping, and mixed-drive layouts in one click.
4. Choosing by bay count (decision tree)
- 2 bays: RAID 1 (mirror) or two independent drives (one data, one backup). One piece of advice: with 2 bays, “independent drives + regular cross-backup” is more flexible than a mirror
- 3–5 bays: RAID 5 / SHR, the balance point between capacity and safety
- 6–8 bays: RAID 6 or SHR-2 — big arrays take a long time to rebuild, so dual redundancy is reassuring
- 8+ bays / enterprise: RAID 6/10 plus a hot spare, together with scheduled snapshots
Once the array level is settled, don't forget the supporting work:
- Capacity and expansion planning → RAID calculator
- Whole-system power draw and electricity cost (running 24/7) → power calculator
- Offsite/remote disaster recovery → remote access port forwarding, IPv6 direct-connection options
5. Three things to do before building an array with new drives
- Run a bad-sector / health check first: put each new drive through smartctl before it joins the array, so a factory-defective drive doesn't get in
- The same-batch trap: drives bought in the same batch are theoretically “born together, die together”, so mix batches where you can (keeping capacities identical)
- Prioritise a hot spare: when you have spare bays, configure a global hot spare — steadier than swapping manually after a failure
6. Common misconceptions
- “RAID 5 lost a drive, I'll keep writing to it” — continuing to write in a degraded state is high-risk; replace the drive and rebuild as soon as you can
- “During a rebuild I'll just pull the drive and plug it back in” — any bay change during a rebuild can destroy the array; back up critical data before you touch anything
- “RAID 5 on SSDs as cache isn't worth it” — for most home setups the read/write cache gains are limited, and the money is better spent on RAM
- “With RAID I don't need backups” — one more time: RAID is not a backup. Accidental deletion, malware, and whole-machine failure are all beyond RAID's reach
FAQ
Q: How much can 4×12TB RAID 5 actually hold?
About 36TB by vendor rating, roughly 32.7TiB as reported by the OS, and plan for about 29TiB of usable data.RAID calculatorhandles exact calculations for any combination of drive count and size.
Q: SHR or RAID 5?
On Synology hardware choose SHR (flexible expansion, friendly to mixed capacities); other systems don't have SHR, so standard RAID 5 is fine. Performance and redundancy are broadly equivalent.
Q: How long does it take to rebuild a 16TB drive in RAID 5?
It depends on drive speed and load, typically 15–30 hours. During the rebuild the array has no redundancy, so avoid power loss (a UPS matters) and heavy load.
Q: Can I start with 2 drives in RAID 1 and convert to RAID 5 later?
Most NAS systems support online RAID 1→5 conversion (Synology and QNAP both do), and SHR grows naturally from just two drives. Back up first, though — expansion failures are rare but irreversible.
Q: Are HDDs plus an SSD cache worth it?
They pay off on 10GbE networks or with lots of small-file random reads; on a gigabit home network the gain is small. Upgrading RAM and networking first is usually the better spend.
