SHR vs Classic RAID, Explained
SHR (Synology Hybrid RAID) is Synology's automated layering of standard Linux RAID (mdadm) and LVM that lets you mix drives of different sizes without wasting the extra capacity, and grow the array by swapping in bigger drives one at a time. Redundancy is the same as classic RAID: SHR-1 survives one drive failure, SHR-2 survives two. With identical drives, SHR and classic RAID (RAID 1/5/6) are effectively the same in capacity and performance — SHR's advantages only appear with mixed sizes or incremental upgrades. The tradeoff: SHR is Synology-only, so your disks aren't portable to other NAS brands. And whichever you pick, RAID is not a backup.
RAID is not a backup — and neither is SHR. Both protect against a drive dying so the array stays online. Neither protects against accidental deletion, ransomware, corruption, theft, fire, or a power surge — those hit the live array and every parity/mirror copy at once. Keep a separate, tested backup. Full explainer →
What do the classic RAID levels actually give you?
RAID combines multiple drives into one array, trading raw capacity for redundancy (except RAID 0, which trades safety for nothing you should want in a NAS). Here's the fast decode, with usable-capacity math for a typical 4×4TB build:
| Level | Mechanism | Drive-failure tolerance | Usable capacity (example) | Who it's for |
|---|---|---|---|---|
| RAID 0 | Stripe — data split across drives, no redundancy | Zero. One drive dies, everything is gone | 2×4TB → 8TB (100%) | Nobody storing data they care about. Don't use it in a NAS. |
| RAID 1 | Mirror — every drive holds a full copy | 1 drive (on a 2-drive mirror) | 2×4TB → 4TB (50%) | 2-bay NAS owners; the simplest safe setup |
| RAID 5 | Striping + single distributed parity | 1 drive | 4×4TB → ~12TB (one drive's worth to parity) | 3+ bay arrays balancing capacity and redundancy |
| RAID 6 | Striping + dual distributed parity | 2 drives simultaneously | 4×4TB → ~8TB (two drives' worth to parity) | Larger arrays with big drives, where a second failure during a long rebuild is a real risk |
| RAID 10 | Mirrored pairs, striped together | 1 drive per mirror pair (up to 1 per pair) | 4×4TB → 8TB (50%) | Users prioritizing rebuild speed and write performance over capacity efficiency |
The general arithmetic: RAID 5 usable space ≈ (N − 1) × drive size; RAID 6 ≈ (N − 2) × drive size; RAID 1 and RAID 10 ≈ half the raw total. So 6×8TB in RAID 6 gives roughly 32TB usable from 48TB raw. (Real formatted capacity runs a bit lower after filesystem overhead and TB/TiB accounting.)
What's the classic-RAID problem SHR solves?
Classic RAID has one rigid constraint: every drive in the array is treated as if it were the size of the smallest drive. Mix a 4TB into an array of 8TB drives and each 8TB drive contributes only 4TB — in a 4-drive RAID 5 of 4+8+8+8TB, you get about 3×4TB = 12TB usable and 12TB of the larger drives' capacity sits unusable.
That also makes upgrades all-or-nothing: on most classic-RAID systems, replacing drives with bigger ones only yields extra space after every drive in the array has been upgraded. For a home user who buys one bigger drive per sale cycle, that's years of paid-for capacity you can't touch.
How does SHR actually work?
SHR isn't a new RAID algorithm. It's automation on top of standard Linux storage layers: Synology's DSM partitions each drive into segments, builds multiple standard mdadm RAID arrays across those segments, and joins them into one volume with LVM. Concretely, with 4+8+8+8TB drives, SHR creates one RAID array across the 4TB slice of all four drives, and a second array across the remaining 4TB slices of the three 8TB drives — then presents the combined space as a single volume. Instead of ~12TB usable (classic RAID 5), you get roughly 12 + 8 = ~20TB usable with the same single-drive redundancy.
- SHR-1 ≈ single-drive fault tolerance (RAID 1/5-class constructs underneath). Minimum 2 drives — and works from 1 drive with no redundancy.
- SHR-2 ≈ dual-drive fault tolerance (RAID 6-class constructs underneath). Requires at least 4 drives.
Because the underlying pieces are standard md/LVM, redundancy math is the same as classic RAID — SHR doesn't conjure extra safety, it just wastes less space when sizes differ.
What are the real benefits of SHR?
- Mixed-size efficiency. Different-sized drives contribute most of their capacity instead of being clipped to the smallest drive. (Synology publishes a RAID/SHR capacity calculator to preview exact figures for any drive mix.)
- One-at-a-time upgrades. Swap a single drive for a bigger one, let it rebuild, repeat. In SHR-1, new capacity becomes usable once at least two drives are the larger size — you don't have to replace the whole array before seeing any benefit.
- Zero decisions for beginners. SHR picks appropriate RAID constructs automatically as you add or replace drives, which is why it's Synology's default suggestion.
What are the honest limits of SHR?
- It's Synology-only. No other NAS brand runs SHR. If your Synology box dies, you can't drop the disks into a QNAP or UGREEN and keep going. Recovery on a plain Linux machine is possible — the layers are standard mdadm and LVM, and Synology documents the procedure — but it's a command-line rescue job, not plug-and-play. That's a soft lock-in, and it compounds with Synology's drive-compatibility tightening: the more Synology-specific your setup, the more your next box is probably also a Synology.
- No performance win. With matched drives, SHR performs like the equivalent classic RAID level. With mixed drives, some data lives on arrays spanning fewer disks, so throughput can be modestly uneven across the volume. SHR is a capacity-flexibility feature, not a speed feature.
- Same rebuild risk as any RAID. Every one-at-a-time upgrade is a full rebuild that stresses the remaining drives. On big arrays, that's the argument for SHR-2 (dual tolerance) — and always for a real backup first.
- Competitors are catching up, partially. TerraMaster's TRAID similarly targets mixed-size flexibility with single- or dual-drive redundancy (verify current details in TerraMaster's documentation). QNAP and UGREEN stick to classic RAID levels. Unraid and ZFS-based DIY systems handle mixed drives in their own, quite different ways.
When is classic RAID just as good — or better?
- Uniform drives from day one, no upgrade plans: SHR and classic RAID produce the same capacity and the same redundancy. There is no advantage to convert or agonize over.
- You value a standard on-disk layout: plain RAID 1/5/6 (still Linux md on a Synology, but a simpler single-array arrangement) is the most conventional, widely understood structure.
- You're not on Synology: the question answers itself — pick the classic level that matches your redundancy need (RAID 1 for 2 bays; RAID 5 or, on larger arrays, RAID 6 for 3+), or the vendor's own flexible mode if it has one.
Rule of thumb: on a Synology box, SHR-1 (or SHR-2 on larger arrays) is the sensible default because it costs nothing today and buys flexibility tomorrow. Everywhere else, classic RAID plus the 3-2-1 backup rule is the whole game.
Related reading
Frequently Asked Questions
For practical home use, neither. SHR is built from the same standard Linux mdadm RAID and LVM layers that classic RAID uses on a Synology box — SHR-1 uses RAID 1/5 constructs and SHR-2 uses RAID 6 constructs under the hood. With identical drives, SHR and the equivalent classic RAID level perform essentially the same. Choose SHR for flexibility, not speed; choose classic RAID if you value portability of a standard, well-understood layout.
SHR-1 tolerates one drive failure (comparable to RAID 1 on two drives or RAID 5 on three or more). SHR-2 tolerates two simultaneous drive failures (comparable to RAID 6, and requires at least four drives). As with any RAID, redundancy keeps the array online through a drive failure — it is not a backup and does nothing against deletion, ransomware, fire, or theft.
Yes, but it's non-trivial. SHR volumes are built from standard Linux md (mdadm) RAID and LVM, so a Linux PC with enough drive ports can usually assemble the array and mount the volume for data recovery — Synology documents this recovery path. But it is a command-line rescue procedure, not plug-and-play, and you cannot simply move SHR disks into a QNAP, TerraMaster, or UGREEN NAS and keep running. Treat SHR as Synology-specific in practice.
TerraMaster has TRAID, its own flexible-array system that similarly aims to use mixed-size drives efficiently and support single- or dual-drive redundancy (TRAID/TRAID+) — check TerraMaster's current documentation for exact behavior. QNAP's QTS and QuTS hero use classic RAID levels (with static-volume/pool flexibility but not SHR-style mixed-size pooling). UGREEN's UGOS Pro, as a young OS, offers standard RAID levels. Outside turnkey boxes, Unraid and ZFS-based systems solve mixed-drive flexibility in their own, different ways.
With uniform drives, SHR-1 produces the same layout, capacity, and redundancy as RAID 5 (or RAID 1 on two drives), and SHR-2 matches RAID 6 — there is no capacity or performance advantage either way. Many people still pick SHR on a Synology box for the future option of mixing sizes when they upgrade one drive at a time. Pick classic RAID if you want the most standard, portable on-disk arrangement.
Yes — this is SHR's second big advantage. You can replace drives one at a time with larger ones (letting the array rebuild after each swap), and once enough drives have been upgraded, SHR expands the volume to use the new space. In SHR-1 the extra capacity of larger drives becomes usable once at least two drives are the larger size. Classic RAID on most systems only grows after every drive in the array has been replaced. Note that each swap triggers a full rebuild, which stresses the remaining drives — keep a current backup before you start.