More RAID 5 volumes are lost while someone is attempting a recovery than at the point of the original failure. That happens either by rebuilding the RAID onto older disks that can't take it, or by rewriting the RAID metadata as part of a repair. Our approach takes this risk away. First each member is imaged, and then we reassemble the RAID in software. Under no circumstances do we ever write to the original disks.
No RAID case is quoted by us without being evaluated first. A free diagnostic comes before any quoted cost, and the written figure after it starts from £500 + VAT, depending on the number of members. For logical rebuilds, in most cases you only pay if we recover your data.
RAID 5 was developed to cope with losing a single disk. The problems arise with what goes wrong after that, and these are what fill our bench. The examples below are some of the most common RAID 5 failures we see coming in from server rooms in Wiltshire.
When a replacement disk is installed and the rebuild starts, the ageing survivors are worked very hard for hours until one of them gives out before the rebuild completes. It is common enough for a degraded RAID 5 to become dead in a single afternoon, and this is the most frequent RAID 5 emergency we deal with.
Disk controllers, host bus adapters (HBAs) and backplanes tend to fail less gracefully than the disks they control. Each of the disks could be perfectly good, but the layout and parity information would need to be recovered from the member disks themselves before a usable volume exists again.
For example, if one of your member disks has been missing for several weeks, it still holds a snapshot of what existed at that point. Forcing this disk back into operation makes the controller mix old stripes with live ones, which creates bad parity across the entire set.
Parity is designed to recover from one read error on any one member of a stripe. However, when two members of a stripe have errors in the same location, parity is unable to fill these areas with valid data, so we make sure we image around each fragile area on each disk before building any stripes.
Clicking the "Repair", "Re-create" or "Initialise" buttons can quietly overwrite the metadata describing where your data is located. If you have already clicked one of them, please don't click anything else.
Adding disks to an existing RAID set or changing the RAID level is a lengthy and often slow process. If it is halted part way through by a power outage, a system crash or someone pressing the stop button before it has finished, the RAID set can get stuck between its old layout and the new one it was moving to, and it will not mount under either configuration.
None of your original disks will ever run within a live array again. Each one is read sector by sector on dedicated imaging equipment (after head-swap bench repair first, if that member has physically failed) to create an image of each disk. All further work takes place using those images. By examining them we work out the disk order, stripe width, rotation and parity direction, and then build a virtual volume from that information so your files can be retrieved from it, while the untouched original disks are kept back as the last layer of protection.
Putting in a new disk and letting the controller go through a rebuild once more can look like progress. However, on a set that has already used up its redundancy, it can be the most dangerous button in the building, as the rebuild puts heavy stress on the remaining disks at a time when they are at their weakest. So if you get a second failure while the rebuild is running, what would normally have been a routine job becomes a much longer one. If your rebuild stalls or you get an error message during it, shut down the system and leave it switched off.
Most of the RAID recoveries we take on hold the main day-to-day storage for a small business. These include HP ProLiant and Dell PowerEdge servers, the odd self-built tower, and NAS units made by Synology, QNAP and Netgear, sent in from towns such as Swindon, Chippenham, Cirencester and Royal Wootton Bassett. Once you've had a quote by phone or through the free online diagnostic, the member disks of the array, each labelled with its bay, come to our lab in Oxford. The lab is under an hour from Swindon up the A420, and the disks can be dropped off at reception Monday to Friday, 9am to 5:30pm, with no appointment needed. If you prefer, they can be sent overnight by tracked and insured post. Business-critical arrays can also move ahead of the queue for our bench. More detail on the wider service is under RAID recovery and NAS recovery, and the costs can be found on the data recovery cost page.
First power down the server or array, then get a quote from us either over the phone or through our free online diagnostic. Next, post the member disks to us, clearly labelled by bay. We will reassemble the set in software, and the price will be given to you in writing before any work starts.
Most customers get a quote first, by phone or with the free online diagnostic, and after that they post or courier their media to us.
Sending a drive from a computer, laptop, MacBook, iMac, CCTV/DVR or server? Please remove the hard drive or SSD from your device and just send us the drive (from a server, every disk, each one labelled with its bay), as we do not offer a drive removal service. We do not recover storage that is soldered to the motherboard, for example on T2 and Apple Silicon Macs and some thin laptops, and only drives that can be removed and sent to us will be accepted. If you're not sure which kind you have, ring us first.
↓ Print the booking-in & shipping form (PDF)
Please mark the package for the attention of Oxford Data Recovery so that we can ring you as soon as we've diagnosed your media.
If you're unsure what to send, please ring 0800 689 0668 before sending anything, or use our free online diagnostic.