Data Recovery Case File · NAS & Network Storage · Redundancy Against the Wrong Thing
Dropped NAS: When Both Mirror Members Are Damaged
His enquiry was three short lines and states the problem exactly. "Have two disks from a NAS, mirrored. NAS was dropped and disks have failed. 3TB disks. Please provide cost for recovering as much data as possible." A mirrored pair is the configuration this archive most often calls fortunate — each disk holds a complete copy, so one failing leaves the other intact. That protection has a specific shape, and a drop is precisely the shape it does not cover. Redundancy defends against one disk failing on its own schedule, not against a single event that reaches both at once. But there is a genuinely hopeful consequence of a mirror that applies even here, and it is why his phrasing — as much data as possible — is likely to be answered better than he expects.
| Media | Two 3TB disks from a mirrored two-bay network storage unit — unit dropped while assembled; both members reported as failed |
| Reported situation | Impact to the complete unit · both mirror members affected by the same event · maximum partial recovery sought · cost requested in advance |
| Fault class | Correlated impact damage across both members of a mirrored pair — redundancy defeated by a single event; per-member damage likely to differ in extent and location |
| Equipment used | Each member inspected under the laminar flow bench before power · matched donor head service per member as required · PC-3000 Express with Data Extractor prioritised imaging of each · images merged sector-wise, taking the best read of every region from either member · losses reported per file |
The decode: what a mirror protects against, and the advantage that survives
What redundancy actually covers: a mirror is designed for independent failure. Disks fail at different times for different reasons — one develops bad sectors, one's bearing goes, one's controller gives out — and a mirror means that when it happens to one, the other carries on. That is genuinely valuable and it is what the configuration is for.
What it does not cover: a single event reaching both members simultaneously. A drop, a fire, a flood, a power surge, a theft, or a mistaken deletion propagated instantly to both copies — these are correlated failures, and against them a mirror provides no protection whatsoever, because both copies are in the same box experiencing the same thing. This is the substance of the phrase RAID is not a backup, and it is usually explained abstractly. His case is the concrete version: two complete copies, one fall, both damaged.
The advantage that survives anyway: and this is the part worth being encouraged by. Even with both members damaged, he still has two physical copies of every sector — and the damage is very unlikely to be identical on both. Each disk fell in its own position, with its own heads and its own surfaces, so the regions that failed on one are frequently readable on the other. That makes his recovery meaningfully better than a single damaged disk would allow: each member is imaged separately, and the two images are then merged sector by sector, taking the best available read of every region from whichever disk supplied it. A file unreadable on one member and intact on the other comes back whole. So a mirror that failed to prevent the loss still improves the recovery — not as redundancy, but as a second attempt at every sector.
Why "as much as possible" is the right framing: because the honest deliverable here is a map rather than a percentage. Where a region failed on both members, that material is gone and should be named. Where either member yielded it, it comes back. He should expect a per-file account of what is complete, what is partial and what is missing, rather than an estimate offered in advance.
The instruction that matters now: neither disk should be powered again to test it. After an impact, each start attempt risks converting head damage into surface scoring — and here that risk applies twice over, because every sector lost on both members is a sector genuinely lost.
On the bench
Neither disk was powered on arrival. Each was inspected under the laminar flow bench before anything was switched on — heads and platter surfaces examined, the drop's damage located and any scoring measured — and serviced with matched donor parts where the finding required it, as two separate patients with two separate conditions. Each was then imaged on the PC-3000 Express under Data Extractor, head-mapped and prioritised. Only afterwards were the two images merged sector-wise, taking the best available read of every region from whichever member supplied it — the step that converts two damaged copies into one substantially better result — and the remaining gaps were resolved to the individual files they affected.
The outcome
Both members serviced and imaged individually, the images merged to take the best read of every sector, and the result delivered as a per-file map. Free assessment, one fixed written figure including VAT, 50% of parts and labour upfront with the balance only on successful recovery. The decode, for anyone relying on a mirror: it protects against one disk failing independently and offers nothing against a single event that reaches both — a drop, a fire, a flood, a surge, a theft — which is what people mean when they say RAID is not a backup; but even with both members damaged you still hold two physical copies of every sector, and the damage is rarely identical, so imaging each separately and merging the best reads beats either disk alone. Keep both members, and stop powering them.
Mirrored disks damaged by the same event
Keep both disks and stop powering either of them. It's tempting to write off the more damaged one and concentrate on the "better" member, and it's the wrong instinct: even two damaged copies give you two attempts at every sector, and the regions that failed on one are often readable on the other. Imaging each separately and merging the best reads produces a materially better result than working from either alone, so both are worth sending. Label which bay each came from. Don't let the unit rebuild or resync anything, and don't insert a replacement disk. And take the wider lesson from what happened: a mirror protects you against one disk failing on its own schedule, not against a drop, fire, flood, surge or theft that reaches both at once — which is exactly why a copy somewhere else remains necessary.
Send both — call Glasgow Data Recovery on 0141 404 0294; each inspected before power and imaged separately, then merged to take the best read of every sector, with losses named per file.
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.