MHDR / Specialist cases / RAID 0 data recovery

Specialist · RAID 0 stripe recovery

RAID 0 data recovery in Maidenhead. A stripe keeps no second copy, so every member travels to the bench.

RAID 0 buys one large, quick volume by cutting every file into blocks and dealing them across the disks in rotation, holding nothing back. Around Maidenhead they turn up as two-bay boxes set this way for the capacity, editing scratch drives run for the throughput, and older desktops whose board RAID utility offered it. When one member stops, the volume stops with it. So every disk sent in from Maidenhead is imaged first, the layout worked out from those copies, and the set put back together in software rather than in the controller that lost it.

No fee unless the data comes back Diagnosis free, then a fixed quote Post it in from Windsor, Slough or anywhere else

Describe the fault to an engineer
0800 6890668

What each RAID 0 symptom is telling you.

None of these? Run the triage →

The symptom	The likely cause	Start here
One drive died and the whole volume vanished	Nothing in a stripe is duplicated, so the set fails as a single unit	Power down; keep every disk
The array is gone from the BIOS after a board swap	The definition lived with the old controller; the data did not	Do not let it create a new set
A two-bay NAS set to RAID 0 no longer mounts	The volume spans both disks in a format Windows cannot read	Send both disks, or the box whole
A RAID utility is offering to rebuild it	There is nothing to rebuild from; it writes fresh metadata	Decline it and pull the power
Large video files open corrupted	Something assembled the set with the wrong order or block width	Stop; the layout must be proved
A scratch array dropped out mid-project	One member is failing, and no file can be completed without it	Stop the work; image the disks

If you post it in: use a tracked, insured service addressed to our secure intake lab, and the return leg is paid at our end; if you would rather check the packing first, an engineer will go through it by phone. Each step is set out on the postage page.

RAID 0: two disks or more, and not one spare copy.

How a striped set worksEach file is cut into fixed blocks and dealt out to the disks in turn, so both are read at once and their capacities add up. Nothing is written twice, and no record is kept of what was there before.

Why one failure ends itThere is no parity and no mirrored half to fall back on, so the set survives exactly as long as its least healthy disk. Add a third member and you have added a third way for the whole volume to go at once.

Before a single file appearsThree things have to be established first: which disk the blocks began on, how wide each block is, and how far into the disk the data starts. All three come out of the images, from file system structures and whatever description the array left behind, then get tested against files whose contents we know.

NAS boxes striped for spaceA two-bay unit set to RAID 0 is a small Linux machine presenting one volume across both disks, usually ext4 or Btrfs over an mdadm stripe. A member on its own carries alternate blocks, so Windows sees an unformatted disk and offers to help. Decline, and send both.

What happens between arrival and return.

The most recent cases →

Booked in, then assessed at no cost Free

Nothing becomes chargeable until you have said yes to it. Each item takes a case number of its own the moment it arrives, an engineer establishes what has actually failed, and you are given a candid view of which files stand a realistic chance of returning. The single quote that follows is fixed, written down, and costs nothing.

Diagnosis at no costOne fixed quote, in writingNothing signed

Image every member, alter nothing

Every member is copied read-only, the weakest of them on hardware that can be slowed right down and told to step past unreadable ground. Your own disks are never written to.

Every member imaged read-onlyOriginals never written to

Prove the stripe geometry

Order, block width and starting offset are derived from the copies: partition and boot records, directory runs, and the points at which a long file's contents break off on one disk and resume on another. Candidates are tested, not assumed.

Stripe width establishedOrder and offset proved first

Assemble it in software, then open files

The images are interleaved back into a single volume, the file system riding on them repaired where it is torn, and then files are opened — a long video, a database, a drawing — because a tidy listing is no evidence that the layout is right.

Volume assembled in softwareFiles opened, not listed

Opened, verified, and sent home

Nothing is billed until you have seen the recovered file list in full and told us to go ahead. What comes back arrives on new media, with the return journey at our cost, and the case stays open at this end until you have opened every file yourself.

You approve the list firstReturned on fresh mediaThe return post is ours

What tends to show up first

Anything wider than a block is scattered — a file larger than the stripe width leaves pieces on every disk in the set, so one member that will not read puts regular holes through nearly every one of them.
Striped is not spanned — a spanned volume fills one disk before the next, so a dead member costs you what sat on it. A stripe interleaves everything, so a dead member costs part of nearly every file. The two sit a line apart in a NAS menu.
There is nothing here to rebuild — no parity, no second copy, nothing for a controller to work back from. A utility offering to rebuild or repair a striped set is really offering to write a fresh array over the top of the one you want returned.
No files appear until the layout does — a flawless image of every member still shows you nothing recognisable until the sequence, the block width and the starting point have been established from the data.

What we can and cannot promise: where every member still reads, a striped set usually comes back complete — imaged, geometry proved, volume assembled, files opened and checked. Where one disk is physically failing, the outcome is decided by how much of it an imager can recover, because what is missing from that member appears as gaps in every file wide enough to have crossed it. Small files that sat inside a single block often survive. You are given the list of which is which, and the free diagnosis is where that line is drawn.

From the casebook, lately.

MH · MHD-2026-6846ON RECORD ✓

The two-bay NAS that had been striped for capacity, not safety

A small practice had run its two-bay NAS as a stripe since installation, for the one large volume it gave them, and after a power cut the box stopped appearing on the network. One disk had been slow for months. Both were imaged, the weaker over two days at reduced speed, and the layout proved from the ext4 structures before use. Current drawings came back intact; three older archives returned with gaps, each named.

Both members copied first3 days from arrival

Before the RAID 0 disks come to the lab.

Do these now

Cut the power to the box or the PC
Write the bay number on each disk before removal
Send every member, including the one that will not spin
Tell us what created the array and roughly when it was first put together

Things best not done

Asking the utility to rebuild the set
Creating the array again to make it appear
Letting Windows initialise or format a member
Scanning one disk on its own for files

The questions that arise most often.

One disk in the striped set has failed. What is left?

We will be straight about this: a striped set holds no second copy of anything, so what comes back is decided almost entirely by the disk that failed. If an imager can still lift most sectors off it, the rebuilt volume may be very nearly entire. Where that member gives up nothing, files small enough to have landed inside a single block on a surviving disk return whole, and everything larger comes back with gaps at regular intervals. Which of those applies to your set is settled at the free diagnosis, before any money is involved.

Do you need all the drives, or only the one that failed?

All of them, the stopped one included. A stripe is defined by its members together, and one disk on its own holds alternate blocks that no software can turn back into files. Send them as they are: nothing formatted, nothing scanned, no further attempts in the enclosure, and each disk marked with the bay it sat in. A two-bay NAS may travel whole if you would rather. Bench equipment will usually read a disk the machine itself has written off, and the amount it recovers decides how complete the rebuild can be.

How do you work out the stripe size and the disk order?

From the data, on the images, never on your own disks. The blocks sit in rotation, so the sequence shows itself where a long file's contents break off on one member and resume on the next; partition tables, boot records and directory structures give the same answer from another direction, and host RAID often leaves a description of the set at the end of each member. Candidate layouts are assembled and tested against files whose contents are already known, because a wrong block width produces a volume that mounts and looks reasonable while every large file in it is ruined.

Our array was built by the motherboard's Intel RAID. Does that matter?

It changes where the set was described, not how it was written. Intel RST keeps its definition in a small area at the end of each member and in the driver, so a board swap, a cleared BIOS or a switch to AHCI can make the volume vanish while your data stays exactly where it was. What matters is that nobody lets the utility create the array again to bring it back. Send all the members and the layout is read off the disks.

More of the jobs we've closed.

Data off servers → SQL & mailboxes → Servers & RAID sets → Media we work with →

Switched off, it keeps every chance it has.

Each further power-up asks more of a drive that has already begun to fail. Open the case first and let the free diagnosis tell you what is left.

0800 6890668

Ask for a free diagnosis Posting it to us

Hardware and media we recover

Typical faults

Evidence and enquiries

Practical matters