RAID is one of those storage topics that sounds far more complicated than it is. Strip away the jargon and it is a single idea: use more than one drive at once, so that either your data survives a drive dying, or your files move faster, or both. The complication comes from the fact that there are half a dozen ways to arrange those drives, and most of the advice you will find online was written for data centres running twelve-bay servers, not for someone putting two hard drives in a Synology box under the stairs. This guide covers the RAID levels a home user will realistically meet, what each one actually costs you in usable capacity, and the one rule that matters more than any of it — RAID is not a backup. If you are still deciding whether you need a NAS at all, our storage buying guide for 2026 compares NAS against plain external drives before you spend anything.
A RAID array presents several physical drives to your computer as one logical volume. How it splits your data across those drives determines what you get out of it.
There are three mechanisms in play, and every RAID level is some combination of them:
That is the whole vocabulary. Everything below is those three ideas arranged differently.
Two or more drives combined into one volume with data striped across all of them. You get the full capacity of every drive and roughly the combined speed. You also get a volume that dies completely if any single drive fails, and your odds of that get worse with each drive you add.
RAID 0 has almost no place in a home setup in 2026. The performance argument evaporated when NVMe SSDs arrived — a single modern NVMe drive is faster than any array of hard drives you would build at home, and if speed is the goal you are better off buying one good drive. It is worth considering only for genuinely disposable scratch space, like a video editing cache you can rebuild from the source files.
Two drives, identical content, half the total capacity usable. One drive fails, the other keeps serving files while you replace it. Reads can be slightly faster because either drive can answer; writes are no quicker.
For a two-bay NAS — which is what most home users actually own — RAID 1 is the sensible default and effectively your only redundant option. It is also the easiest to recover from in a disaster, because each drive holds a complete, readable copy of the data rather than a fragment that only means something in the context of the array.
Three or more drives, with parity distributed across all of them. You lose the capacity of exactly one drive regardless of how many you have, so four 8TB drives give you 24TB usable instead of the 16TB that mirroring would leave you. Any one drive can fail without data loss.
This is where the maths gets uncomfortable. When a drive fails, the array must read every sector of every remaining drive to rebuild the replacement. With a four-bay array of 12TB drives, that means reading roughly 36TB without a single unrecoverable read error. Consumer drives are typically rated for one unrecoverable read error per 10^14 bits read, which works out to roughly one error per 12.5TB. You can see the problem: on large modern drives, the rebuild itself is statistically likely to hit an error, and on a single-parity array an error during a rebuild can cost you the pool.
The practical rule most of the industry has converged on: RAID 5 is still fine on drives up to around 4–6TB, questionable at 8TB, and a genuinely bad idea at 12TB and above. Since the cheapest cost per terabyte now sits firmly in the 16–22TB bracket, that rules RAID 5 out for most new builds. NAS-rated drives from the Red Plus, IronWolf and N300 families generally carry better error-rate ratings than desktop drives, which softens the maths a little, but it does not reverse it.
Same idea as RAID 5, but with two parity drives instead of one. Four drives minimum, and you sacrifice two drives' worth of capacity. Two drives can fail simultaneously — and, crucially, the array survives hitting a read error partway through rebuilding after a single failure.
If you are building a four-bay or larger array with drives of 8TB or more, RAID 6 is the level to use. Rebuild times are long and write performance takes a hit from calculating two sets of parity, but neither of those matters much on a home file server that spends most of its life idle.
Four drives arranged as two mirrored pairs, then striped together. Half your capacity, excellent performance, fast rebuilds because reconstructing a drive means copying its mirror rather than recalculating parity from everything. It survives two drive failures only if they happen in different pairs.
RAID 10 is what you use when the array is running databases or virtual machines. For a home NAS storing photos, media and backups, paying half your capacity for write performance you will never notice over a gigabit network is hard to justify.
Synology's Hybrid RAID, and similar schemes from other vendors, deserve a mention because they solve a real annoyance. Traditional RAID levels waste capacity when your drives are different sizes — a 4TB drive in an array of 8TB drives gets treated as 4TB, and everything above that is stranded. SHR partitions the drives so mixed sizes are used far more efficiently, which matters enormously if you plan to upgrade capacity one drive at a time.
Underneath, SHR-1 gives you RAID 5-equivalent protection and SHR-2 gives you RAID 6-equivalent protection, so the same capacity rules apply: SHR-1 for small drives in a two- or three-bay unit, SHR-2 once you are at five or more bays or drives of 12TB and up.
One thing to check before you buy: Synology restricted third-party drive support on its 2025 Plus-series units and then reversed that decision in DSM 7.3, restoring health monitoring and storage pool creation for standard WD and Seagate drives. That reversal covers 3.5-inch drives and 2.5-inch SATA SSDs but not M.2 cache drives, and the company has not clearly committed to the same position on its 2026 hardware. If you are buying a new unit, verify the current compatibility list for that exact model rather than assuming.
Two bays: RAID 1. Four bays with drives under 6TB: RAID 5 or SHR-1 is defensible. Four or more bays with drives of 8TB and up: RAID 6 or SHR-2. Any situation where you were tempted by RAID 0: buy a single faster drive instead. That covers close to every home scenario, and if your case is genuinely unusual, the deciding factor is almost always rebuild risk rather than performance.
RAID protects against exactly one thing: a drive failing. It does nothing about accidental deletion, ransomware, a corrupted file that gets faithfully mirrored to both drives, a failed power supply that takes several drives with it, theft, or a flood. All of those are more likely to cost a home user their data than a straightforward drive failure is.
You still need real backups, on separate media, with at least one copy somewhere else. A single external hard drive rotated offsite covers more realistic failure modes than adding a second parity drive ever will, and our 3-2-1 backup rule guide sets out the minimum sensible arrangement. Treat RAID as uptime insurance — it keeps the NAS serving files while you replace a dead drive — and never as your only copy.
Prices below are checked regularly and reflect current UK retailer pricing at the time of writing.
Browse the latest UK prices on EasyDriveCompare and find the best deal for your next storage upgrade.
Compare Hard Drive Prices →