Data Recovery Case File · NAS & Network Storage · The Array Outlives the Box
Data on the drives, not in the appliance: a dead NAS decoded — why "might or might not migrate" is a gamble with the only copy, how RAID 5 is rebuilt offline from five images, and the sequence that keeps the hardware decision for afterwards
His enquiry was precise, and its central problem is one thousands of NAS owners will meet. "I have 5 × 3TB drives set up as RAID 5 in a five-bay Synology NAS that encountered a fault on the main system board and no longer powers on. The HDDs are not supported in the latest Synology model bays. Synology have said I can buy a new NAS and it might or might not migrate. How much would it be to get the data off these HDDs?" Credit where it belongs: the vendor's answer is honest — migration compatibility across generations genuinely is uncertain, and they said so rather than selling him a certainty. But honesty about a gamble does not make the gamble wise, and this page explains why: an attempted migration is performed on the drives themselves, which are currently the only copy of everything. The principle that resolves his whole question is simpler than the hardware: the array lives on the five drives, not in the box that failed. Get the data off first; decide about appliances afterwards, with nothing at stake.
| Media | Five 3TB hard drives in a RAID 5 array from a five-bay NAS appliance — appliance main board failed, unit will not power on; drives believed healthy; array not supported by current-generation bays |
| Reported situation | Appliance dead at board level · vendor advises a new unit may or may not adopt the existing array · owner seeking extraction cost rather than hardware replacement |
| Fault class | Appliance failure over an intact array — no drive fault reported; reconstruction from member images indicated; migration attempts contraindicated pending extraction |
| Equipment used | All five members imaged write-blocked in parallel (Atola TaskForce 2, RAID autodetect including parity handling with a missing member) · array parameters — member order, stripe size, parity rotation, offsets — determined from the images · volume reconstructed offline and verified · data delivered to media of the owner's choosing |
The decode: where the data actually lives, why migration risks it, and what parity buys him
The appliance is a computer, not a vault: a NAS is a small computer bolted to some drive bays: its board runs the software that assembles the array and serves it on the network. When that board dies, nothing whatsoever happens to the data — which is written across the five drives, in a layout the array software defined. The drives are, in the fullest sense, the storage; the box was only the reader. This is why the correct question is his own — what does it cost to get the data off the drives — and not the question the situation invites, which is which appliance to buy next.
Why migration is the wrong first move: the vendor's "might or might not" describes a procedure carried out on the live drives. If a new unit adopts the array, excellent. If it doesn't — because the generation, the software version or the volume format differs — the failure modes are not neutral: a unit that fails to recognise an array may offer to initialise the drives, may re-lay volume structures, or may begin a rebuild that overwrites members. Each of those is performed on the only copy in existence. In effect, migration asks him to bet an unbacked-up archive on a compatibility question the manufacturer has already declined to guarantee. The professional order inverts it: extract first, verified, onto independent media; then migrate, buy, or rebuild at leisure — because a failed migration after extraction costs a Saturday, while a failed migration before extraction costs everything.
What RAID 5 gives him, and what it doesn't: RAID 5 distributes data with parity across the members, which means the array can be reconstructed even with one member missing or failed — a genuine safety margin, and good news given his drives are believed healthy. What parity does not survive is a stale or partially rebuilt member being reintroduced, or the array's layout being altered by a well-meaning appliance. Reconstruction on a bench avoids both hazards entirely: each drive is imaged individually and write-blocked, and the array is then rebuilt from the images — the member order, stripe size, parity rotation and data offsets determined by analysis rather than assumed, and every experiment performed on copies where a wrong guess costs nothing. (One practical note for owners: labelling bay order before removing drives is helpful, but not fatal if forgotten — parity mathematics can determine the correct order from the images themselves.)
On the bench
The five members were imaged write-blocked and in parallel on the Atola TaskForce 2 — the array frozen safe before a single reconstruction hypothesis was tested — and the drives' health confirmed as the enquiry had suggested. Array parameters were then derived from the images themselves: member order, stripe size, parity rotation and offsets established by analysis, with the platform's parity handling standing ready to reconstruct even had a member been absent. The volume was assembled offline from the copies, mounted, and verified by opening across the estate — shares, documents, media, the household and business data the appliance had served — before delivery onto media of his choosing. The dead NAS never turned on again, and never needed to: it had only ever been the reader.
The outcome
The RAID 5 array reconstructed offline from five member images, verified and delivered — the migration question rendered harmless by being asked after the data was safe. Free assessment, one fixed written figure including VAT, no recovery, no fee. The principle, posted for every owner of a dead NAS: the array lives on the drives, not in the appliance, so a failed board is an inconvenience rather than a loss; "might or might not migrate" is an honest warning about a procedure performed on your only copy — decline the gamble and extract first; RAID 5's parity is a real margin that survives a missing member but not a mistaken rebuild; and once the data sits verified on independent media, every hardware decision that follows is free of consequence.
NAS appliance dead — drives fine, and a new unit suggested
Don't put the drives into a new appliance yet. Migration between generations may work, but when it doesn't, the failure can initialise your drives, re-lay volume structures or start a rebuild — all performed on the only copy of your data. Take the drives out, keep them together, and note the bay order if you can (parity maths can recover the order from images if you can't). Don't attempt a rebuild, don't "test" a member in another machine that might write to it, and don't power the failed unit repeatedly hoping. Have all members imaged write-blocked and the array reconstructed offline from those images, where wrong hypotheses cost nothing. Then, with your data verified on independent media, buy whatever hardware you like — and let the migration question be interesting rather than dangerous.
The array outlives the box — call Leeds Data Recovery on 0113 322 3083; all members imaged write-blocked, parameters derived, volume rebuilt offline and verified — one written figure, no recovery, no fee.
Request a quote online →
Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.