Hard drives fail!

As with the certain inevitability of taxes and death, every hard disk is going to be either forced into early retirement, and some will die of old age or some other hard-drive-related disease. But many drives will unexpectedly fail at some point, due to poor quality manufacturing or defect, overheating or physical abuse. Due to the precision involved in modern hard drive manufacture, the operating tolerances are minuscule and easily, adversely affected by sudden movement or shock, or extreme temperatures.

Overheating

Overheating is probably responsible for most hard drive failures we encounter as a data recovery company- although as portability increases through advances in miniaturisation, and lifestyle changes are progressively responsible for data storage devices being subject to more exercise than they are wont, we are slowly seeing a relative reversal of overheating-related failures in favour of dropped, shocked or otherwise-neglected storage devices.

The fans in top-end computer servers make a cantankerous noise when running at full steam. This is not designed to annoy everyone within 500 metres; the fans are doing their level best to prevent the hard drives (and other, internal electronic components) from passing out in the heat – and possibly even suffering terminal heat stroke.

Modern consumers are increasingly demanding totally mute storage and computer equipment – yet are blind to the effect this “passive cooling” has on the working parts. Stories abound of NAS storage devices being relegated to the office stationery cupboard to satisfy the Office Manager’s demand for minimalism in the workplace. No airflow, the gradual build-up of heat- and suddenly the NAS is NOT. These units should be stored in the open, with good, free-flowing air, and in as cool a spot as possible.

Passive cooling DOES NOT WORK. Manufacturers of external drives and NAS units are lying to you when they make these claims. We know – we test these things. In the photograph above is an external drive. The two hard disks have been mounted almost touching; there is scarcely a millimetre clearance between the drives and the top cover. There is no active cooling. Small wonder, then, that one of the drives overheated and caught fire.

Enterprise drives and RAID storage systems

One cause of failure that is not widely known relates to the hard disk’s method of error correction. When a modern desktop drive encounters a bad sector (a tiny, “bad” area on the physical recording medium) it will mark that area as bad and will no longer use it. However, this process can take up to 60 seconds. Furthermore, during this reallocation process, the hard drive is effectively hiding itself from the rest of the system. In the meantime, RAID controllers are often programmed to “fail” a non-responsive hard drive – but after only a few seconds of inactivity. As a result, the controller takes out that hard drive and marks it bad.
For this reason, you should only use Enterprise-class hard drives for RAID storage.

---

Firmware failures

A drive’s firmware can be considered to be its “translator” between the operating system and the physical storage medium. It “talks” to the computer and to the drive’s internals. Sometimes it loses the plot, though; either because of a problem at the manufacturing stage (perhaps incorrectly programmed) or it encountered some obscure operation of the hard drive that it simply did not understand. Maybe the read/write head / slider was a bit optimistic in how high it could fly above the platters (something controlled by the firmware, depending on ambient conditions and other reasons too complex to mention here) and the firmware gave up on itself, rendering the drive inaccessible. Seagate recently entertained the Data Recovery industry with a huge spate of firmware failures with their 7200.11 series. In their favour, they did correct the issue – but not before many users had emptied their pockets and took out a second mortgage to have the data professionally recovered.

---

Physical damage

This is an area that is screaming up the charts of Main Causes of Drive Failure.
In an era where anyone old enough to walk is carrying around more computing power than is required for a manned space mission to Mars, mangled laptops and MP3 players are rushing through the doors of data recovery companies. Hard drives aren’t really designed to be banged about, or worn on the person during a rugby match. They are designed to be cocooned in a quiet, peaceful area of tranquility. This picture is a close-up of the inside of a laptop hard drive that was dropped whilst powered down. As you can see, the read/write heads (the tiny black things are called “sliders” and the read/write heads themselves are embedded into the trailing edge of the slider) have been severely damaged.

Up a notch to laptop computers. In the movies, you will see some advertising honcho prance around with his latest audio-visual presentation on a laptop, which he plonks down on each desk of his colleagues. Unfortunately, this happens in real life, too – and a drive is not accustomed to that sharp knock which, if the drive is running, is very likely to cause some media damage caused by a head crash.

Desktop, tower, server computers – should all be left in situ and not moved when powered on. Not even slid aside gently to gain access to the rear. Always power down first. Fortunately, most serious servers and enterprise systems are rack-mounted, drilled deep into concrete floors that would effectively withstand the impact from a truck.

Physical damage also occurs when a component fails; either due to a manufacturing anomaly (for this, read “the use of sub-standard components, bought in as cheaply as possible”), or simply a bad batch of drives manufactured in China, and which did not go through the correct (or any) process of quality control. This is the type of physical damage we experience most often with high-end servers and storage systems. IT Administrators in larger corporations are usually extremely responsible individuals, and it is purely bad luck on their part.