When a RAID array goes down, your data isn’t sitting on any one disk — it’s spread across all of them, and putting it back together is a puzzle, not a copy-and-paste. Whether you’ve lost a second disk in a degraded array, had a controller fail, or watched a rebuild go wrong, the single most important thing is to stop before that damage becomes permanent. We rebuild RAID arrays of every level and controller offline, from images, for businesses across the UK. Post the disks in, or drop them off.
A rebuild onto an already-inconsistent array can overwrite good data for good. Power it down and let it be imaged first.
RAID spreads your data across several disks — striped for speed, mirrored or parity-protected for resilience — according to a specific geometry: the RAID level, the block or stripe size, the order the disks sit in, how parity rotates across them, and where the data starts on each disk. While it’s healthy, the controller holds all of that and hides the complexity. When it fails, that geometry is exactly what has to be reconstructed, because no single disk holds a usable copy of your files — each one holds only its slice of a larger pattern.
That’s why a failed array is a job for specialist recovery rather than a data-copy. We image every member disk, work out the array’s parameters from the data itself, and rebuild it virtually, offline — never on the original disks, and never by trusting a controller that has already shown it can’t be trusted. Get it right and the array comes back whole; get it wrong — usually by letting hardware ‘fix’ itself — and a recoverable array becomes an unrecoverable one.
A second disk fails in a degraded array. The commonest disaster. One disk drops and the array runs degraded on its redundancy — then a second disk fails before the first is replaced, and the array falls offline. Because the disks were bought together and aged together, the second failure often follows the first by days. On the failed disks themselves, imaging the weak ones is the first step.
Controller failure. The RAID card dies and takes the array configuration — the metadata that defines the geometry — with it. The disks are fine; the map that ties them together is gone, and has to be rebuilt.
A rebuild that went wrong. A replacement disk is added and the rebuild fails partway, is interrupted by a power cut, or runs against an array that was already inconsistent — and now good and bad data are mixed. This is one of the most damaging situations, and one of the reasons not to let an array rebuild before it’s been imaged.
Lost or ‘foreign’ configuration. After a power event, a firmware glitch or a controller swap, the array shows as foreign, missing or reconfigured, and the disks won’t assemble.
Accidental reinitialisation or reconfiguration. The array is recreated, cleared or reconfigured by mistake — recoverable, but only if nothing has since been written over it.
File-system corruption on a healthy array. Sometimes the disks and geometry are fine but the volume itself is corrupt — a logical recovery on top of a working array.
More RAID data is destroyed by well-meant recovery attempts than by the original fault. If your array is down, please resist the following until it’s been assessed:
Don’t let the controller rebuild onto a replacement disk. If the array is already degraded or inconsistent, a rebuild can calculate and write bad parity across good data, overwriting recoverable files permanently. It is the single most common way an array is lost.
Don’t reinitialise, ‘clear the configuration’ or recreate the array. These can wipe the very metadata we use to reconstruct the geometry. If a controller offers to initialise disks, decline.
Don’t change the disk order, and don’t keep swapping disks in and out. Label each disk with the bay it came from before you remove anything, and keep that order. Forcing failed disks back online repeatedly can also finish off a disk that’s barely readable.
Don’t run CHKDSK or repair tools on the array volume, and don’t format it. On an array that’s assembled incorrectly, a ‘repair’ can make the corruption permanent. Power the system down, note the disk order and any error messages, and let it be looked at — on a RAID, doing nothing is almost always safer than doing something.
The first step is always to image every member disk, read-only, exactly as it is — and any disk that’s mechanically failing gets the clean bench and careful-imaging treatment first, because a RAID recovery is only as good as the reads underneath it. Nothing is written to your original disks at any point.
From those images we reconstruct the array’s geometry: the RAID level, block size, disk order, parity rotation and data offset, worked out from the data on the disks rather than trusting a lost or damaged configuration. We assemble the array virtually from the images, verify the reconstruction is correct by checking that real file structures fall into place, then rebuild the file system and extract your data. Because every stage happens on copies, we can try, check and refine the reconstruction as often as needed without any risk to the originals — which is the whole reason it’s done this way.
We recover every common configuration — RAID 0, 1, 5, 6, 10, 50 and 60, plus JBOD and spanned volumes — across both hardware and software RAID. Hardware controllers including Dell PERC, HP/HPE Smart Array, LSI / Broadcom MegaRAID, Adaptec and Areca; software and OS-level RAID including Windows Storage Spaces, Linux mdadm, and ZFS and Btrfs pools.
Arrays don’t only live in servers, either — we recover the RAID inside NAS units (including Synology SHR, QNAP and ReadyNAS), servers and SANs, whatever the underlying disks or file system. If it stores data across multiple disks, we can reconstruct it.
RAID recovery starts with a free diagnostic: we image and assess the disks, work out whether the array can be reconstructed, and give you a fixed written quote before any chargeable work — with a file listing to check that the reconstruction is right. Pricing is per case, because a clean two-disk rebuild and a five-disk array with mechanical failures and a botched rebuild are very different jobs. We understand a downed array often means a business is stopped, so a priority service is available when time is critical, along with NDAs and confidential handling as standard. Disks can come in by insured courier — we recover arrays for businesses right across the UK and Ireland.
Not necessarily. A redundant array (RAID 1, 5, 6, 10) is designed to survive a single disk failure, so if only one disk has gone, the safest move is often to replace it carefully and let the array rebuild — provided the array is otherwise healthy. The danger is a second failure during that rebuild, which is common because the disks aged together. If two disks have failed, or the rebuild misbehaves, stop and have it imaged before anything else.
Only if you’re confident the array is otherwise sound. If there’s any doubt — a second disk showing errors, an inconsistent array, a rebuild that has already failed once — a rebuild can write bad parity over good data and make things permanently worse. When the data matters, the safe path is to image every disk first, then reconstruct offline. Power it down and ask before rebuilding.
Often, yes. A failed or interrupted rebuild is one of the most common jobs we see. As long as the original disks haven’t been reinitialised or overwritten since, we can usually image them, work out the correct geometry, and reconstruct the array from before the rebuild went wrong. The key is to stop making further changes to the disks.
Yes, provided the disks are readable. RAID 0 stripes data with no redundancy, so a single failed disk takes the array offline — but if that disk can be imaged (including with clean-bench work where it’s mechanically failed) and the others are intact, we can reconstruct the stripe and recover the data. It’s more precarious than a redundant array, which is exactly why RAID 0 data should be backed up.
Yes. NAS boxes run RAID under the surface — Synology SHR, QNAP, ReadyNAS and others — usually on Btrfs or ext4. We recover the RAID inside them the same way: image every disk, reconstruct the array and the file system offline, and extract the data. Bring the whole unit and its disks, or just the disks in their original order.
No. Our lab is in Belfast, but RAID recovery is done by post or courier, so we work with businesses right across the UK and Ireland. Send the disks in with insured, tracked delivery in their original order — the service and pricing are the same wherever you are.
Power the array down, note the disk order, and get in touch. We’ll image the disks, work out whether it can be reconstructed, and give you an honest, fixed quote before any chargeable work — priority service available when a business is stopped.