Three disks, one dead, and nobody left in the business who knew how the array had been set up. If it was striped the answer was going to be short — so establishing what it actually was came before anything else could happen.
A business ran its file storage on a Windows machine using software RAID — no hardware controller, the array managed entirely by the operating system. Three disks. One stopped responding altogether, and afterwards Windows no longer presented the volume at all.
The client reconnected the failed disk without success, then called. They could not tell us the configuration: whoever built it had left, there was no documentation, and their working assumption was that it might be striped. That assumption is why they were as worried as they were.
Worth being blunt about striping, because the fear was well founded.
A striped volume has no parity and no redundancy of any kind. Data is written across the members in fixed-size chunks — 64 KB by default on Windows — in rotation, with nothing held spare. Lose a member and you lose every chunk that lived on it, and there is nothing anywhere from which to compute them back.
The practical consequence is worse than the headline percentage suggests. On a three-disk stripe, losing one member removes every third 64 KB chunk across the entire volume. A document under 64 KB might survive whole if it happened to land on a surviving member. Anything above about 128 KB — which is most photographs, every video, every database and most modern office files — is guaranteed to have holes punched through it at regular intervals. You do not lose a third of your files. You lose a third of the content of almost all of them.
So anyone offering to rebuild a striped volume from parity is describing something that does not exist. What can genuinely be done is recovering the failed disk itself so its chunks become readable again — which is a physical job, not an arithmetic one.
Including the owner’s, and including the operating system’s. The configuration is recorded on the disks themselves and can be read back.
Windows software RAID uses dynamic disks, and a dynamic disk reserves the last megabyte of itself for a database describing the whole disk group — which disks belong to it, which volumes exist, what type each volume is, and which regions of which disks make up each one. Every member carries a copy, so the configuration survives the loss of any single disk.
All three were imaged read-only first, including the failed member, and that database was read from the images rather than from the running system that had already failed to interpret it.
Then it was confirmed empirically, which is the part worth describing. Two working images were compared byte for byte at matching offsets. Striped members carry different data in alternating chunks; mirrored members carry identical data at identical addresses. The difference is unmistakable within a few megabytes of comparison, and it requires no metadata to trust.
They were identical. The array was not striped. Two of the three were a mirrored pair, with the third serving a separate function in the disk group — and the redundancy the client did not know they had is what saved the job.
With the structure established, the volume was reassembled from the images entirely outside Windows. The operating system’s own handling was set aside deliberately: it had already declined to present the volume, and a system that has decided an array is invalid is not a useful participant in proving that it is not.
The failed disk was still worth recovering, and was. Opened under filtered air, fitted with a matched donor head assembly, imaged in graduated passes. It contributed the material that was not duplicated on the mirrored pair.
The file system was then rebuilt from the assembled volume, with damaged structures reconstructed from the backup copies NTFS maintains for exactly this purpose.
About 99%. The mirror carried most of it and the repaired member supplied the rest, with the shortfall confined to regions unreadable on the failed disk and not duplicated elsewhere.
The lasting point here is not technical. Nobody in the business knew what their own storage was doing, and the gap between their assumption and the reality was the gap between a near-total loss and a near-complete recovery. Twenty minutes writing down what the array is, which disks are in it and where the documentation lives would have removed a week of genuine fear.
RAID recovery is from £500 +VAT after a free 48-hour diagnostic.
Send the device over with a short account of what happened to it. The diagnostic costs nothing and commits you to nothing, and the figure that follows is fixed in writing before anything is opened.
Nothing can begin until the device is on the bench, which makes this the only step that needs anything from you. Pack it properly, put your details in with it, and send it over. What follows is a free diagnostic and a written figure, in that order.
Posting it? Use something tracked and insured — whatever is on the drive is worth considerably more than the postage. Bringing it in? Weekdays, 9am to 5:30pm, and it still wants packing as above for the journey.
Tell us what the device is, what it is doing, and anything already tried — an engineer will come back with a realistic view of the odds and a price band, before you commit to posting anything.
We’ll be in touch shortly. For anything urgent, call 0161 871 0788.
Only by recovering the failed disk itself. RAID 0 has no parity and no mirror, so nothing can compute the missing stripes back — the only route is making that physical disk readable again. Where it is genuinely destroyed, that share of every file is gone.
Not really — it is a normal starting point. The layout is written into the disks and can be read back. Stripe size, member order, and whether disks are mirrored or striped are all determined by comparing the images, regardless of what anyone remembers.
By comparing members directly. Mirrored disks hold identical content at identical offsets; striped disks hold different data in alternating blocks. The difference shows up within a few megabytes of comparison.
Usually not. Windows software RAID keeps its configuration in metadata on the disks, and when a member disappears the operating system frequently stops presenting the volume entirely — while the data on the surviving members is untouched.
No. Reconnecting an intermittently failing member can prompt the operating system to write to the set or resync in the wrong direction. Power it down and get every disk imaged before anything else is attempted.
From £500 plus VAT, fixed in writing after a free 48-hour diagnostic. Where a member needs clean-air work there is a 50% deposit with the balance due only on success.
Start with an instant online quote, or call and talk it through with us first. You'll have a clear, fixed price before any work begins.