Data Recovery Case File · NAS & Network Storage · No Controller, No Documentation

Recovering a RAID Array From the Year 2000

His enquiry was doing something quietly generous, on behalf of a family member who is no longer able to do it himself, and it described one of the harder jobs in this archive with disarming casualness. "I've got an old system from around 2000. It was a RAID system comprising six or seven hard drives, and had all the family photos on. I'm just wondering roughly how much you charge for this sort of old RAID?" Twenty-five years is a long time in storage, and almost everything about this case is unusual: the interfaces predate anything current, the controller that assembled the array is long obsolete, no documentation exists, and even the number of disks is uncertain. The encouraging part is that the missing controller — which sounds like the fatal problem — is the one thing that genuinely does not matter.

The decode: why the controller does not matter, and what actually does

Why the missing controller is not the problem: the instinct is that an array assembled by a particular controller needs that controller to be read, and it is wrong. What a controller does is apply a layout — a rule for how data is distributed across members: which disk holds which chunk, in what order, with what block size, at what offset, and where parity sits if there is any. That layout is expressed in the data itself, and it can be worked out by analysing the members rather than by asking the hardware. So the approach is not to hunt for a twenty-five-year-old controller card. Every member is imaged individually, and the layout is derived from those images by testing candidate configurations against the filesystem's own structures until the volume resolves coherently. Proprietary metadata written by the original controller helps where it is present, but the reconstruction does not depend on it.

What genuinely does matter — one: the member count. Six or seven is not a detail to be resolved later, because it changes what any reconstruction means. If the array had seven members and only six survive, the outcome depends entirely on the level: a striped array missing a member loses part of every file, while an array with parity can be reconstructed from the survivors. So every disk that exists should be gathered, and any bay ordering or labelling preserved — though where it has been lost, the order can usually be determined mathematically from the images.

What matters — two: the interfaces. Drives of that era and class use connections that no current machine provides, and a cheap adapter is not the answer for disks this old and this valuable. Era-appropriate native connection is the difference between reading them properly and provoking them.

What matters — three, and most: dormancy. These disks have been stationary for a very long time. Bearings stiffen, lubricants migrate, and heads can adhere to platters during storage — so each member's first power-up is the moment that decides its contribution to the whole. That is why they are assessed and imaged individually, on equipment that can watch what each one does, rather than assembled into anything and switched on. It is also realistic to expect that on a set of this age, one or more members may need mechanical work before they will read at all.

On the cost question: honestly — a legacy multi-drive array is priced per member, because each is a separate patient before it is part of an array, and the reconstruction is additional. An indicative range can be given, but the binding figure follows a free assessment of how many members there are and what condition each is in. On a set this old, that assessment is genuinely worth having before any commitment.

On the bench

The disks were treated as individual patients before they were treated as an array. Each was brought up on an era-appropriate native connection rather than a general-purpose adapter, assessed on the Atola Insight Forensic before anything was asked of it — the first power-up after long dormancy being the reading that matters — and imaged individually on the PC-3000 Express under Data Extractor, with mechanical service where a member required it. Only once every image existed was the array question addressed: candidate layouts tested against the filesystem's own structures until member order, block size, parity rotation and offsets resolved coherently. The volume was reassembled offline from the copies, and the family photographs were validated by rendering and delivered in date order.

The outcome

Every member imaged individually, the array layout derived by analysis, and the photographic archive reassembled, validated and delivered. Free assessment, one fixed written figure including VAT; where a drive has to be opened, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The decode, for anyone holding a very old array: the missing controller is not the obstacle it appears to be, because the layout it applied is expressed in the data and can be derived by analysis from images of the members; what matters instead is gathering every disk, since the difference between six and seven can decide the outcome; era-appropriate connection rather than a cheap adapter; and above all dormancy, because after twenty-five years each disk's first power-up is the moment that determines what it can still contribute.

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.