The hand-over point is Manchester (the Oxford Street drop-off, or the post), and from there all drives are processed the same way; no matter if they arrive from Manchester, Burnley, or Bolton. This is the "tour": a description of what each piece of equipment is, how it works, and why it decides which recoveries work.
Each function has its own station: intake write blocking, hardware imaging, firmware repair, and mechanical jobs. Each job goes through the stations it requires and bypasses the ones it doesn't.
Roughly three nanometres will explain most of what we do here.
There is just one number that describes almost everything about how mechanical recoveries work. A modern hard drive's read/write head hovers approximately three nanometres above the platter — a space many thousands of times thinner than a single strand of a human hair. An air-cushion created by the platters' rotation maintains this small space.
This is why you can't open a disk on a table; normal room air contains particles hundreds of times larger than the gap, and one of these under a head will dig into the surface upon which your data resides. This is why if you drop a drive while it is still running, it will be catastrophically damaged. On the other hand, a drive that is powered down when dropped usually will not suffer damage. Finally, it is easy to see why a drive that has been clicking for two weeks will be a vastly different creature than one switched off after only ten minutes of operation. Each revolution with a failing head passes the whole track underneath it once again.
The most asked question and the honest answer to this.
Placing a failing hard drive into a computer and dragging files to another location results in failure due to a very simple fact. The operating system was developed with healthy hardware in mind. When you hit a "bad" sector on a failing disk, Windows continues to try reading it until all hope is lost. During that time, each attempt to read that sector puts more stress on an already fragile disk that has limited reads left in it. All the effort put forth during each retry is wasted, as the whole copy is aborted and hours of work discarded simply due to the presence of one unreadable block towards the end of the file system.
Purpose-built imaging hardware reverses the order of things as well. The timeout for bad sectors is measured in milliseconds, rather than seconds, so when you have a bad sector on your drive, that bad sector costs you a fraction of a second, rather than a minute. Bad sections are documented and bypassed instead of being repeatedly attempted. In addition, reads may target a single head at a time; therefore an undamaged surface is captured completely while damaged surfaces are worked around. Lastly, all operations performed against these drives are strictly "read only", which means that the original data cannot be altered regardless of what happens to the software or application attempting to do so (no regular operating system will guarantee this).
That's why an otherwise perfect mechanical unit can be totally hidden.
There is much more going on inside a hard drive than just magnetically storing data. There is actually a real program running on the hard drive called firmware. It runs on its own processor and RAM. This firmware also stores its code and information in a protected service area on the platter(s), not in chips. That firmware holds the defect map. The firmware contains translation tables to convert logical addresses to physical locations, and adaptive calibration values, specific to each individual drive.
If some portion of that stops being readable then the disk will not complete booting. The disk spins fine, it appears to be working just fine, and displays either zero capacity or an empty "model" string. This is due to the inability of the system to characterise itself. Your data remains unaffected by a program which will not run. Diagnostic modes provided by the manufacturer allow for the repair of that area; modules are rebuilt or replaced with corresponding parts from a donor device, translators are regenerated from surviving defect lists, and ROM adaptives are transferred to a replacement board if necessary. The same is true for SSDs.
The reconstruction stage and the theory behind it.
When we receive a failed RAID or NAS, the part that used to describe the array is the one that has already proved unreliable, so we treat its information about how the array was created as less reliable than other possible sources of information about the state of the array.
The layout is based on the data. Stripe size appears to be a repeated structural period throughout the array's members. The disk order is determined by testing different configurations until the file-system structures appear to be organised and not "noise." To confirm the parity rotation, we test each possible configuration to determine which will allow us to reconstruct the original stripe(s) and ensure that when we do so, the parity correctly computes. Each possibility is run through the same test process as each other, with the correct solution announcing itself through successful mounting of the resulting file system, while all incorrect possibilities result in an unmountable file system or a failure to mount altogether.
Immediately above those are whatever additional layers the manufacturer may have built on top — LVM pools, SHR provided by Synology, thin provisioning provided by QNAP, and BeyondRAID provided by Drobo without documentation; each layer needs to be stripped away in sequence as errors occurring at lower layers present themselves as "nonsense" at higher layers. All of this occurs using read-only copies of the storage medium, thus completely outside any hardware that has failed.
Yes you can insert the drive into your computer and on a good drive it should work. However once the drive starts to fail, the operating system repeatedly attempts to read each bad sector for seconds at a time, then ultimately fails to complete the entire backup, wasting hours of previous work and creating significant strain on a drive with limited remaining reads.
The space a hard drive head floats over a platter due to air pressure from spinning disks. The reason we can't open hard drives in regular air and the reason a dropped hard drive spinning at full speed is worse than when it's stopped.
Your hard drive has lost the ability to read its own firmware. Most hard drives write firmware to a special service area on the platters for the drive rather than storing it in a chip. If any part of the firmware becomes unreadable then the drive will be unable to describe itself. Thus, it will report zero capacity and your data will sit untouched behind it.
No, because the RAID controller is the component that has already failed. Therefore, any configuration information provided should be used as a hint. The configuration is determined through analysing the data that is stored across the members of the array.
No, and recovering your data does not mean repairing the drive. Repairs only go as far as reading the data one time, and a drive with donor heads or transplanted platters is not a disk to trust. Your data will be extracted from the failed hard drive onto new media.
No. Our engineering staff carry out all of the work in-house, with nothing subcontracted, and the data never leaves the UK at any time.
The first station is free – diagnostics will be completed within 48 hours. A written quote will be produced prior to commencing with any recovery.