RAID combines several drives into one volume so the array survives a disk failure, and often reads or writes faster. In a NAS it is the difference between a dead drive being a Tuesday and a dead drive being a disaster. This note explains the levels that matter, the one trap that catches people on large modern drives, and how it maps to a real decision: moving from a 4-bay Synology DS413j to a 7-bay UniFi UNAS Pro.
The one rule before anything else
RAID is not a backup. It protects against hardware failure of a disk, nothing else. It does not protect against deletion, ransomware, a corrupt filesystem, a power surge that takes the whole box, theft, or fire. A rm -rf replicates instantly across every drive in the array.
Follow 3-2-1: 3 copies of the data, on 2 different media, 1 of them off-site. RAID gives you uptime and one of the copies. It is never the plan on its own.
The RAID levels that matter
| Level | Min drives | Usable capacity | Survives | Speed | Use for |
|---|---|---|---|---|---|
| RAID 0 | 2 | 100% | nothing (any 1 drive kills it) | fastest R/W | scratch/cache only, never data you keep |
| RAID 1 | 2 | 50% | 1 drive | fast read | small 2-bay, simple mirror |
| RAID 5 | 3 | (n-1) drives | 1 drive | good read, slower write | small arrays of modest drives |
| RAID 6 | 4 | (n-2) drives | any 2 drives | good read, slower write | large drives, bulk storage (the safe default) |
| RAID 10 | 4 | 50% | 1 per mirror (up to n/2) | fastest safe R/W | VMs, databases, IOPS-heavy work |
| JBOD / Single | 1 | 100% | nothing | native | throwaway or app-managed disks |
| SHR (Synology) | 1-2 | flexible | 1 (SHR-2: 2) | like RAID 5/6 | mixed drive sizes, easy expansion |
Quick reads on each:
- RAID 0 stripes with zero redundancy. One drive dies, everything is gone. Pure speed, never for data you care about.
- RAID 1 mirrors two drives. Simple and safe for a 2-bay, but you pay 50% of capacity and it does not scale.
- RAID 5 stripes with one parity block. You lose one drive's worth of capacity and can survive one failure. The catch is the rebuild (see below).
- RAID 6 stripes with two parity blocks. You lose two drives' capacity and survive any two failures. The right default for anything with large disks.
- RAID 10 mirrors then stripes. Best real-world speed with redundancy, but you only keep half the raw capacity. Overkill for a media/document NAS, ideal for databases or VMs.
- SHR is Synology's layer on top of RAID. It lets you mix drive sizes without wasting the extra space and makes expansion painless. It is proprietary, so it only exists on Synology boxes.
The URE trap (why RAID 5 is risky now)
This is the part that changes the decision on modern drives. When a drive fails in RAID 5, the rebuild has to read every sector of every remaining drive to reconstruct the lost one. Consumer HDDs are rated at roughly one unrecoverable read error (URE) per 10^14 bits. A single 10 TB drive holds about 80 trillion bits, so reading it in full has a meaningful chance of hitting a URE. In RAID 5, one URE during a rebuild can kill the whole rebuild, because there is no second parity to reconstruct that bad sector.
- On 8 TB and larger drives, treat RAID 5 as unsafe. The rebuild reads are too large and the failure probability is real, not theoretical.
- RAID 6 solves this: during a single-drive rebuild it still has a second parity, so a URE is reconstructed and the rebuild continues.
- Rebuild time also matters. Large arrays can take a day or more to rebuild, and that whole window is when a second drive is most likely to fail (they are the same age, same batch, same workload). RAID 6 covers that second failure.
Bottom line: for bulk storage on big drives, RAID 6 is the modern default. RAID 5 only makes sense on small arrays of modest drives with a solid backup.
DS413j vs UNAS Pro: what actually changes
| Synology DS413j (2013) | UniFi UNAS Pro | |
|---|---|---|
| Bays | 4 | 7 (2.5/3.5" HDD/SSD) |
| RAID options | SHR, 0, 1, 5, 5+spare, 6, 10, JBOD | RAID 5, 6, 10 only (+ hot spare) |
| Mixed drive sizes | Yes, via SHR | No, classic RAID (size = smallest drive) |
| Network | 1x Gigabit | 1x 10G SFP+, 1x Gigabit |
| CPU / RAM | ARMv7, 512 MB | Quad-core A57 1.7 GHz, 8 GB |
| Form factor | Desktop | Rackmount (2U) |
| Ecosystem | Full DSM (apps, Docker, VMs) | Storage only, tight UniFi integration |
| Off-box backup | Hyper Backup | Remote UNAS, SMB/CIFS, S3, Dropbox |
Three things stand out for the move:
- No SHR on the UNAS Pro. It uses classic RAID, so every drive is treated as the size of the smallest one. Buy matched drives. You lose SHR's easy mix-and-expand flexibility.
- RAID 5/6/10 only. No RAID 1, no single-disk redundancy pool. Minimum is RAID 5 (3 drives) or RAID 10 (4 drives).
- It is a 2U rackmount, not a desktop unit. Fine if you have a rack or shelf; louder and bulkier than the Synology if you do not. Verify current firmware for in-place expansion/RAID migration before you commit, since UniFi Drive has historically been less flexible than DSM here.
What config to run on the UNAS Pro
For a capacity-first media and document NAS with 7 bays and large drives, the answer is clear:
- RAID 6 across all 7 drives. Usable = 5 drives' worth, survives any 2 failures, and rides out a URE during rebuild. With 8 TB drives that is about 40 TB usable; with 12 TB drives about 60 TB. A large jump from a full 4-bay.
- Want zero-touch recovery? Run RAID 6 on 6 drives plus 1 hot spare (usable = 4 drives). The array rebuilds itself automatically the moment a drive dies, without you touching it. You trade one drive of capacity for that. Given you are moving off a near-full box, filling all 7 into RAID 6 and keeping a cold spare on the shelf is the higher-capacity play.
- Skip RAID 5 on drives this size (URE trap). Skip RAID 10 unless you are running VMs or databases off it, because it halves usable capacity for speed you will not notice over SMB.
Migration path: there is no in-place SHR to UNAS conversion. Build the UNAS Pro fresh in RAID 6, then copy data across the network (10 GbE on the UNAS side helps, though the old Synology tops out at Gigabit). Afterwards, repurpose the DS413j as a backup target or retire it, and set the UNAS Pro to also back up off-box to S3 or a remote share. RAID handles the disk; the backup handles everything else.
Setup checklist
- Buy matched drives (same size and model) sized for RAID 6 capacity, plus one spare.
- Create one RAID 6 pool across the bays.
- Decide hot spare vs cold spare (auto-rebuild vs one more drive of capacity).
- Enable snapshots for a first line of defence against deletion and ransomware.
- Configure an off-box backup (remote UNAS, SMB, or S3/Dropbox). This is the non-negotiable step.
- Test a restore from that backup before you trust it.
Further Reading
Reference pages and the two devices in play.
