Almost everyone has had disk errors at one time or another. There are the file system errors but this page is concerned with mechanical failures of hard disks. We purchase hard disks frequently primarily to take advantage of capacity increases.
The photo is the ST-506 5MB hard disk ($1,500) that was the less expensive of the choices available in 1980. Mainframe hard disks came down in price rapidly and PC users adopted them quickly.
The old PC XT came with a 5MB or 10MB hard disk. The ST-412 10MB hard disk (1981) was larger and less expensive but was interleaved 5 times. The disk controller was too underpowered to keep up. Eventually a better card allowed 1:1 interleave which materially improved the loading time.
The IBM PC-AT came with the larger ST-225 which was half-the height of the older PC-XT hard disks. It also was double the capacity at 20MB. This meant it was possible to install a pair of hard disks in the machine.
The original PC-XT and PC-AT used a DOS program, SETUP, to configure the BIOS. There was no UI in early personal computers until the mid 1990s. Early machines used a few jumpers on the motherboard.
It was not until IDE disks and the 3½” Quantum LPS52AT (950-5-9105). IDE disks adopted IDENT which returns a 512 byte sector with information about the disk. This can be used by the BIOS to setup the disk automatically.
It several more years before all disks adopted SMART which includes several parameters that can suggest pending fault.
Seagate was late with the ST-157A 40MB IDE. IDE disks often needed a custom or user defined drive type. IDE saw many changes to hard disks.
Over time we have seen a lot of hard disks fail. Only with several disks can a good degree of fault tolerance be established.
The actual MTBF is probably much higher than our experience as disks are taken out of service that are still functional. This means the real service life of the disk may be much longer. We operate 4 disks in the gaming machine primarily for media libraries and backup purposes. RAID cards are popular however modern TB class disks are able to consolidate data easily.
Disks that are taken out of service as operational technically right censor the operating life. This skews the table considerably as hard disks are generally designed for a 10 year operating life. Many hard disks have been in service for even longer. For example we have seen some 40 year old hard disks for sale in working order.
Hard disk docks are very low cost and they can be used to make copies of data quickly. We use the Startech SATADOCK525 which is installed in a spare front DVD drive bay. PCI and PCI Express SATA cards are readily available to support the front panel dock.This allows for easy connection of hard disks disks without opening the machine up and risking damage.
USB docks are very popular and models are available for EIDE and SATA disks. USB docks may have some issues with disk limitations. This was discovered with some USB cases for notebook disks. Many USB docks offer flash card reader slots which are a low cost extra but some are obsolete and are unable to handle the larger capacity models. Storage limitations are a pervasive issue.
More recently USB front docks for 3½” floppy bays are now available with flash card readers included for $10. These make it easy to use bare notebook hard disks for backups.
Storage cases for hard disks are available. Off-line storage of backups is possible with a dock. A low cost stationary cabinet is ideal.
Using redundant copies of important data seems prudent. Downloading games from Steam is slow, especially given the size of modern games. NASA reported a failure rate of 75% for an archive. So its important to consider the value of your files and take the steps necessary.
OUR HARD DISKS
The table includes both 2½” and 3½” disks. Even with a smaller population of disks failures are common. While EIDE is now obsolete, old machines can be retrofitted with an adapter to use a modern SATA disk. Check for hard disk limits which may be a problem.
Many have pointed out the need for a better understanding of what disk failures look like in the field. Yet hardly any published work exists that provides a large-scale study of disk failures in production systems. Changes in disk replacement rates during the first five years of the lifecycle were more dramatic than often assumed.
We base failure rates on 24/7 operation with 8760 hours each year. Obviously if the disks are not powered 24/7 then the probability of failure will be lower.
We now have a front panel SATA slot on the workstation that can handle both 2½” and 3½” disks. This makes it easy to recover data or to check disks for faults. It also affords a convenient way to use hard disks for backups as well. We have accumulated a lot of 2½” hard disks from SSD upgrades to notebook class machines.
Low cost metal brackets can allow the installation of 2½” hard disks is desktop machines designed for the older 3½” disks. These are widely available for less than $2. Seagate is now providing 15mm thick disks for desktop machines making brackets mandatory.