A RAID's most hazardous phase isn't the loss of one of its disks: it is the subsequent rebuilding process. RAID systems will limp along on a weakened leg for months until they fail totally once another disk fails while rebuilding at 60-something percent completion. The bench's first rule for RAID data recovery was created specifically for this story: before any RAID logic runs, every disk in the RAID is imaged.
The redundancy in RAID creates RAID's trap. RAID 5 will continue to function with only one lost disk; therefore, it tolerates the loss. But then it puts a strain on its remaining disks during rebuilds, which is exactly when your second disk (which is typically the same age, batch and level of wear as your first) gives up. This is why many RAID 5 rebuilds fail on their second disk: a classic case. RAID 0 contains zero redundancy. If any one disk fails, RAID 0 is dead. Two-disk 'performance' external hard drives often use RAID 0, without ever telling anyone about it. RAID 1 and RAID 10 mirror data. While they provide protection for the failure of a disk, they also mirror any deletions or corruptions that occur. And beneath all of these are controller metadata, a mapping of how stripes are ordered, how large blocks are and where the disks exist in sequence. The failure of a controller or an over-zealous 'initialize' button can erase the metadata while leaving the actual data alone.
Every individual disk in your array is cloned and saved prior to beginning the recovery process, ensuring that the original state of each disk will be maintained permanently regardless of what occurs subsequently. Once the disks have been cloned, the virtual array is rebuilt using the images of those disks, working out the stripe order and parity rotation from data contained within the disks themselves when the metadata is no longer present; the failed member disk is created via the mathematical application of parity and the entire file system is validated before calling anything recovered. We are asking for just one thing from you: label each disk with its bay order number (1, 2, 3...) as you remove them, and under no circumstances should anyone force online or reinitialise the array to see if it will return. It won't return. It gets overwritten.
No, but stop now. If you are rebuilding an array and the rebuild fails before it's finished, then another drive is failing. That drive will be working harder when you try again. Recovery makes an image of all the disks in the array, including the bad one. It then virtually recreates the entire RAID by using those copies and rebuilds the missing stripe by calculating with parity. This happens so often we can't even begin to count how many times this occurs.
No. First get a quotation for your disk recovery (via our website quote form or over the telephone). We will require you to send us the labelled hard disks only: do not send the server or the controller. The striping order, block size and parity rotation are all recoverable from the physical data on the disks themselves in most cases. It may be helpful if you have noted the make and model of the controller used; however, this is not necessary, as we recover this information from the disks themselves.
Depends on what was written where, and honesty is required: it could be a bad situation. Typically initialisation will write to array metadata and possibly more than that. Turn everything off now, note exactly what buttons were clicked and let the diagnostic map out what survived; partial recovery is often expected even after wrong moves.
First, a free diagnostic, then a fixed written quote before any work is done. This means no surprises on the invoice (because we don't give you an invoice until you approve the number).
Examples of some of our other work from the bench that are similar to this kind of job.
It will take only two moves to place your device at our service area, and once that has been accomplished, the clock on your free look starts.
First get a quote, over the telephone or using the online quotation form. When sending your internal drive, remove all cables, caddies and power supplies, as they will have little use when sent to us. But when sending an external drive, send the complete item, including its original casing, together with the supplied cable and power supply. Ensure that the device is securely wrapped and unable to move during transportation. Either post it insured to 4 Redheughs Rigg Westpoint, South Gyle, Edinburgh EH12 9DQ or call into our offices Monday to Friday, 9am to 5:30pm, with no appointment needed. Please include a brief description of how your device was damaged. If you have printed a shipping form, please attach it along with the device.
Once the arrival of your device has been confirmed, the diagnostic will run free of charge, and then when we make the phone call to you, we'll name the problem and tell you the fixed price to fix it. The bench will remain dormant on your device until you give us permission to proceed.
You will receive free diagnostics along with a written quote prior to doing anything else. All members are imaged prior to attempting to rebuild your storage system.