What finishes off a parity set is almost always the attempt to save it: another complete rebuild requested of already exhausted discs, or a menu option that wipes the very description a reconstruction needs. None of that happens here. A copy is made of each member prior to attempting anything else. The layout is worked out from those copies and then demonstrated as opposed to assumed, and not even one byte will be written back onto the discs you posted.
We quote a parity set, and never guess at it. The diagnostic costs nothing, and the written figure follows it: from £500 + VAT, counted by disc, and on logical work, in most cases you only pay if we recover your data.
One dead member is exactly what this layout was bought to absorb, and on its own it rarely results in a recovery job. What really fills up the diary is the afternoon that follows. These six between them make up nearly every parity set that comes to this bench.
The spare is inserted and the controller settles down to reading every remaining disc from start to finish. On today's capacities that is many hours of unbroken work demanded of drives the same age as the one that just died, and one of them declines. What was merely degraded at lunchtime becomes too much for the controller to handle by close of business.
Backplanes, host adapters and controller cards run on their own schedules, and those often run years ahead of the drives they were bought to look after. No member of the set is faulty, yet the volume never appears again. The reason is simply that the details of how it was built died with the hardware.
A member that dropped out three weeks ago is holding a three-week-old photograph of the set. Bring it back online and the controller starts folding that old photograph into live stripes, poisoning the parity across the whole array as it goes, and quietly.
A single unreadable sector on a single disc is exactly the case parity was created to cover. Two unreadable sectors at the same stripe address on two different discs is a gap the arithmetic cannot close, which is why each and every soft region on each and every member is imaged around before any stripe is calculated.
Controller menus have options labelled initialise, re-create and repair, behind warnings that don't begin to describe what they do, and any of the three can flatten the metadata recording where the data actually lives. Therefore if one has already been run, that is the moment to stop rather than trying the next.
Adding a disc (or moving from one level to another) requires hours of constant rewriting. Any interruption to this process caused by a power cut, a crash or an engineer losing patience will leave the set stranded between two geometries and no longer valid in either.
Parity is commonly explained as a spare copy of the data. It is not. For each stripe written across the set, the controller creates a single additional block based upon the other blocks, effectively a running total, and writes it on whichever member's turn it is. In the event of one block in that stripe later going missing, the total along with the surviving blocks can be used to calculate what was lost. A very elegant solution, requiring little overhead in terms of space and strictly limited: it will cover one absent block per stripe and no more. Two gaps in the same stripe and the sum has more unknowns than equations, which is a fact about arithmetic rather than about firmware.
The reason why the arrangement is fragile in practice has nothing to do with the mathematics. Discs that were purchased together will typically fail together. They were produced in the same batch, went into service on the same afternoon, and since then have done identical work at an identical temperature in the same warm cupboard. Therefore, when one disc wears out, the others are not innocent bystanders: they are the same age with the same mileage, and they are about to be asked for the hardest hours of their lives.
Rebuilding a replaced member means reading all the data the other discs hold, not just samples or the portions which were previously written, but each and every sector. Four four-terabyte drives make that twelve terabytes of continuous sequential reads, comfortably the most demanding operation those drives have ever had to perform. Weak sectors that have been sitting unnoticed in a rarely visited corner, and any drive with a few years of power-on hours has some, are found during that sweep. When one of these sectors is discovered mid-rebuild on a set with no redundancy left, there is no warning to act on. The rebuild stops, and frequently the volume is lost with it.
There is a much quieter version of the same problem, and you will find it by looking at the spec sheet. Consumer drives are sold with an unrecoverable read error rate of roughly one sector in every hundred trillion bits. That is so low you can't see it during normal use. Ask for twelve terabytes in a single uninterrupted pass, though, and it stops being a footnote and becomes a reasonable expectation. Hence the advice repeated on this page: an array which has been powered down is just a group of discs all in the exact condition they were in when the first disc died, and each of them can then be copied slowly, on its own, with as many retries as a soft patch happens to need.
Much of what is received here is the full trading record of a company that never anticipated becoming a recovery customer: Dell PowerEdge and HP ProLiant servers, towers somebody assembled themselves, as well as Synology, QNAP and Netgear units. These come from the business parks along the A329(M) and Western Road, from the software and instrument companies on the M4 corridor between Reading and Slough, from engineering units around Camberley and Frimley, and by tracked post and courier from every London postcode shown on the London page. First get a quote over the phone or complete our free online diagnostic. The discs, out of the box and labelled by bay, then travel to Guildford Data Recovery, Building 2, Ground Floor, Guildford Business Park, Guildford GU2 8XH, either dropped off (forty minutes from Bracknell using the A322 and the A3, Monday to Friday, 9am to 5:30pm, with no appointment needed) or by post within one working day. Other layouts are covered on RAID recovery and NAS recovery, and every figure behind them can be found on the data recovery cost page.
No disc of yours is ever placed back into a running array in this lab. Every member is first put onto a hardware imager, and any drive with a physical fault goes onto the repair bench before that happens. Once everything has been imaged, all of the work occurs on the copies. The originals are stored on a shelf untouched, as a fallback nobody expects to need, and each attempt at the reconstruction can be tried, discarded and tried again without costing anything except time.
A collection of images is not yet a volume. Four properties need to be pinned down before anything can be read out of them. The slot each member occupied, since the bay labels and the truth inside the metadata part company more often than anyone expects. The size of the blocks, which depends upon both the controller and whoever configured the unit years ago. How the parity travels, to the left or the right through the rotation, and whether it is synchronous with the data or displaced from it. And how far into each disc the data begins, because almost all controllers reserve the first part of each member to store their own bookkeeping.
Get one of those four wrong and the failure is a quiet one. Up comes a partition table, a directory tree appears, and files appear to have sensible names and dates. However, their contents are gibberish. There are no warnings. This is why none of the four is entered on a hunch here: each one is demonstrated. Structures with a known shape are found across the images, their fragments are matched against one another, and every candidate description is tested until only one is left. The file system is then raised on that description, and only once real files open correctly does anyone begin to talk about which folders are coming home.
Not every RAID 5 lives on a card in a server rack. Intel RST on a desktop board, mdadm on Linux, Windows Storage Spaces and the manufacturers' layouts inside Synology and QNAP units all act as RAID 5 and describe themselves in completely different ways. Metadata can reside at the beginning or the end of each disc, or even both. Several use a geometry altered from textbook parity. All of it is readable, but none of it is readable through guesswork, and it makes no difference whatsoever to how the discs should be handled before they get here.
Turn off the box and leave it off. Don't authorise a rebuild, and don't accept any menu item offering to initialise, re-create or repair the array. Write on every disc which bay it was removed from, because this is by far the most useful piece of information you can pass along and the easiest to lose. Get a quote over the phone or using our free online diagnostic before anything is packed, then keep all of the discs together in one carton, including the disc that failed first, since it is frequently the most useful disc in the box. Take note of the make and model of your controller or NAS, and if a dialogue appeared and someone clicked through it, say so on the booking-in form. The two minutes you spend writing notes will save hours at the bench, and now and then even decide whether a marginal set comes back whole.
Shut down, get a quote first, box up every disc (including the dead one), and let the set be rebuilt in software instead.
Hardly anything walks in here; it arrives in a padded box. Once you have a quote from us over the phone (or from our free online diagnostic), the least stressful option for storage that has already given up is tracked, insured post, and a parcel handed over at a Bracknell post office before the last collection usually reaches Guildford the following working day.
Is the hard drive still bolted into a machine, whether that is a laptop, a tower, an iMac, a MacBook, a rack server or a DVR stored under the till? Get a quote first, then pull the drive out of the machine and send only the bare drive (never the entire machine): from a server the disks come out labelled by bay, and from a DVR you just send the disk. Stripping hardware is not something our lab does, though any repair shop on your high street can do it in about ten minutes. The one case with no way round it is memory chips soldered flat onto the mainboard, which is how T2 and Apple Silicon Macs and certain ultrabooks are built, and that storage is not taken on. Call us first if you're not sure which kind you have. Where the storage cannot be unbolted, there is no parcel to make up.
↓ Print the shipping & booking-in form (PDF)
With your quote in hand, the name on the parcel wants to be Guildford Data Recovery. Driving it over from Bracknell takes roughly forty minutes along the A322 followed by the A3, and you do not need an appointment; posting it costs you a stamp and a day. Either way, a message goes out to you as soon as the package has been logged onto our system, and two working days after that the diagnostic is finished.
Unsure which items should go in the box? Call 0800 689 0668 whilst the box is still open, or work through the free online diagnostic and let it guide you.