Despite manufacturers’ advertised ratings, disk drives in consumer and commercial appliances may not offer the reliability or predictability that users expect. The annual failure rate (AFR) metric printed on many drives show an expected failure rate below one percent (0.88 percent), meaning about one in 114 drives is expected to fail in a given year.
Large-scale studies from Google and Carnegie Mellon challenge that assertion, suggesting the rate is much higher and providing a case for increased vigilance and a need for computer users to back up their data.
Beat the heat?
A 2007 Google study of approximately 100,000 of its own drives used their Self-Monitoring, Analysis, and Reporting Technology (SMART) capability to self-report errors such as platter surface defects and reallocation of data. Among miscellaneous health indicators, SMART can detect bad drive sectors and measure temperature.
Despite SMART’s capability, researchers found that many drives failed with no signals at all; they also could not find a link between running temperature and failure. Moreover, researchers noted that, even with the assumption that a significant number of drives ran exceptionally hot—above 104 degrees Fahrenheit (40 degrees Celsius)—statistics would still not reliably join heat and drive failure to the point where users could rely on the metric to predict the fate of their own drives.
The effects of old age
Meanwhile, Carnegie Mellon reported in a similar study that disk failure rate does not directly correspond to a disk’s age or type. Its researchers found that many drives failed before one year, in an “infant mortality” group, but that those which survived could be expected to live through old age—five to seven years—before statistics showed a rise in their average failure rate.
These two studies transcend their own years to represent a corpus of important industry information that others have not reproduced to scale. Moreover, they provide this industry-challenging set of figures:
Overall, Google researchers determined that the average failure rate of its drives to reach three percent. In lockstep, Carnegie Mellon expected a fail rate of between two and four percent. Both those reports provide evidence that a more realistic AFR is more than double the industry AFR of 0.88 percent.
Be proactive with cloud backups
Users may notice corrupted files or slow reading and writing—or even suddenly fried circuits—as signs that their drives have stopped working or will soon follow that path. Although the above studies report moderate links between symptoms and drive health, it is important for individuals and businesses to be proactive in their protection of important data. Failure can occur at any time.
Although no one can predict the exact moment a hard drive will show signs of wear or crash completely, individuals and businesses can take steps to protect their data with the Mozy cloud-based backup service. Users can back up and sync their files across all of their desktop and mobile devices. Learn more at Mozy by Dell.