A pair that wore out in step. A resync that ran in the wrong direction. An enclosure that died with two perfectly good disks locked inside it. Not one of those is a lost cause. The halves are copied separately, compared against each other on the evidence rather than on a counter, and your files come off whichever proves the better source.
A mirror is quoted, never guessed at. The diagnostic costs nothing, the written figure comes after it — from £500 + VAT counted by disk, logical rebuilds keeping no fix, no fee.
A mirror makes precisely one promise: the volume keeps running when a single disk dies. Every position below steps around that promise rather than through it, which is exactly why they end up in a padded box on the way to Manchester.
Two disks ordered on the same purchase order, spun up on the same morning and worked identically ever since do not fail years apart. They fail within a short distance of one another, and whatever finished the first is frequently only days behind the second.
On a two-bay NAS, or a mirrored desktop case, whatever fails first is generally the controller or the bridge rather than either member. Two entirely sound disks finish up sealed in a box that no longer talks to anything, and each then has to be read out on its own terms.
Put the wrong member back, or leave the unit to work it out for itself, and the blank or stale disk gets faithfully duplicated over the good one. The mirror performs exactly as designed while erasing the thing it was bought to protect.
Where the halves ran degraded at different points, they finish up carrying different versions of the same folders. Merge them carelessly and each spoils the other. Copy them apart and each is free to surrender whatever only it still holds.
A deletion, a corrupt write, a bad patch, an encryption run: mirroring exists to reproduce writes the instant they happen, so every one of those reaches both halves together and neither disk is spared anything.
A NAS firmware update that fails partway through, or an operating system that eats its own boot volume, can leave a dead box with two healthy disks locked inside. The route back is to copy each disk and raise the volume on hardware that works.
From the moment the parcel is opened the halves are treated as two separate jobs. Each goes onto a hardware imager, with any physically failed member visiting the clean bench beforehand, and the resulting copies are then compared: which of them holds more of your data, and which holds the more recent version of it. The stronger copy becomes the foundation, the weaker supplies whatever is missing from it, and neither original is written to at any point in the process.
Mirroring answers one scenario and one only: a disk dying while the volume carries on underneath it. Everything else strolls past it. Delete a folder and both halves lose it together. Push out a bad update and both take it. An encryption run goes through the pair in one pass, and a dead controller removes the two of them from service at the same moment. That is why a broken mirror is a genuine recovery job rather than a swap-and-carry-on, and why the old rule has not budged: the backup that saves you is the one on separate media, kept somewhere else entirely.
Most pairs reaching this bench came out of a two-bay Synology, QNAP or Netgear unit, or a small office server, sent in from studios in the city centre, practices at Crosby and Ormskirk, workshops at Widnes and Runcorn and offices over the water in Birkenhead. Put both disks in one parcel addressed to Manchester Data Recovery, Peter House, Oxford Street, Manchester M1 5AN, roughly 35 miles along the M62 or a working day in the post. The wider picture sits on RAID recovery and NAS recovery, and every band is published on the data recovery cost page.
The mechanism is simple enough to describe in a sentence. Each write the operating system issues is sent to both members, and the operation is not reported as finished until both have taken it. Reads can then be served from whichever disk is free, which is where the modest speed benefit comes from. What no part of that arrangement does is form an opinion about whether the write was a good idea. Deleting a folder is a write. Encrypting a file is a write. Saving a corrupted document over a good one is a write. All of them are duplicated with perfect fidelity, in the same instant, on both disks.
This is the whole of the argument against treating a mirror as a backup, and it is worth stating plainly because the misunderstanding is so widespread. A backup is a copy taken at a moment in time and kept apart. A mirror is a copy taken continuously and kept together. The first lets you go back to yesterday; the second only lets you carry on through a dead disk. They solve different problems, and only one of them helps on the afternoon somebody empties the wrong folder.
Somewhere on each disk the controller keeps a small record: an event counter, a timestamp, a state flag saying whether that member is in step with its partner. When a mirror is assembled, those records are compared and the disk with the higher counter is treated as authoritative. It is a sound design, and it explains most of the ways a resync goes wrong. Pull a disk out for a few days and put it back, and its counter is behind; the box concludes it is stale and overwrites it, which is correct. Rebuild the array after a controller failure, though, and those counters may be gone, damaged or misread, and the box can just as easily conclude the wrong thing.
Recovery does not rely on that record at all. Both members are imaged first, and the question of which is the better source is settled by looking at the contents: which copy has a coherent file system, which holds the later versions of files whose dates you recognise, and which shows signs of having been overwritten. That takes longer than reading a counter and it is considerably harder to fool.
Because reads can be served from either disk, a mirror can hand back data from a member that has begun to rot without anything obvious happening. Bad sectors develop slowly and a controller with no scrubbing routine may not notice until it needs the affected region. The symptom, when it eventually arrives, is odd rather than dramatic: files that open on one occasion and not on another, a volume that checks clean and then does not, an application complaining about data it wrote itself. Where that pattern is what brought you here, both disks matter more than usual, because the good version of a given file may only exist on one of them.
Consumer NAS units rarely implement a textbook mirror. Synology stacks its own layer over the standard Linux arrangement and then puts a modern file system on top; QNAP does something similar with different names; Netgear's X-RAID adds its own expansion logic. All of them are readable, and none of them is readable by simply plugging a member into a Windows machine and hoping. The practical consequence for you is small: send both disks rather than one, note which bay each came out of, and say which make and model the unit was.
What not to send is the enclosure, unless the format is genuinely proprietary, and a telephone call on 0800 689 0668 settles that in under a minute. What not to do is try each disk in turn in a desktop machine to see which one works. Windows will offer to initialise them, and initialise is the one word on this page that turns a routine job into a difficult one.
Both halves copied apart, the stronger one raised, the weaker one filling the gaps. Diagnosis free.
Nearly everything on this bench arrived by tracked, insured post. It is the calmest way to move a failing drive, and a parcel handed in anywhere on Merseyside is normally on the bench the next working day.
Still screwed inside a laptop, tower, MacBook, iMac, server or a CCTV recorder? Take the drive out and post that on its own. We do not strip machines here, and any computer shop will do it in a few minutes. The one job nobody can take on is flash soldered straight to a logic board, as on Apple Silicon Macs and a handful of very slim laptops: if the storage will not unbolt, there is nothing to send.
↓ Print the shipping & booking-in form (PDF)
Address it to Manchester Data Recovery. It is roughly 35 miles from Liverpool along the M62 if you fancy the run, and next working day by tracked post if you do not. We ring or email the moment it is booked in, and the free diagnostic closes 2 working days after that.
Not sure what belongs in the parcel? Call 0800 689 0668 before you tape it shut, or step through the free online diagnostic first.