October 22, 2018

Drive Failure

hard-disk-attritionCloud backup provider BackBlaze provided some very interesting statistics this week in regard to their independent testing of hard drives.

Hard drive manufacturers provide a measurement called MTBF (mean time between failures) for their hard drives.  Based on manufacturer information drives should last on average, somewhere between 11 years and 110 years, following what is known (scientifically) as a bathtub curve.  A bathtub curve says there will be a very high number of failures almost immediately (because of manufacturer defects) and then a second high number of failures toward the end of life (like the walls of a bathtub).  This says drive failures are expected to be, according to the manufacturers, most common within the first few weeks or after a few years.

We all recognize that the moving parts of a computer are the most likely to fail, but assuming, as manufacturers advertise, the average life is somewhere between 11 and 110 years, we don’t worry about it.

However, BackBlaze found, from their experience, that these estimates are incredibly optimistic.  From their experience the failure rate of hard drives after 3 years ranged between 5% and 25%.  Translated: somewhere between 1 in 4 and 1 in 20 drives crashed and burned within 36 months.

According to Wikipedia, there are currently less than 8 hard drive manufacturers in the world and we continue to undergo mergers and acquisitions.  There are simply not that many options and no manufacturers that seem to lead the more or less likely to fail predictability.  According to the BackBlaze study, Hitachi drives proved best by a small margin, but Hitachi recently sold their drive manufacturing to Western Digital.

Given these failure rates, what should a small business do?

  1. Have a backup.  Anytime your data is on a single hard drive, you should have cause to loose sleep!
  2. Do not trust backup to a single backup drive, always have a local and off-site back.  Cloud storage options are now between $1 and $2 per gig per year, it is worth the investment.
  3. Centralize data.  Having important company files spread across various workstations magnifies your risk of loss.  Centralize company data to a server, either hosted onsite or in the cloud.
  4. Invest in system images.  If your employees use a number of specialized applications that take time to install and configure, invest in the few extra minutes to create a system image of their computer.  This is basically a “manufacturer reset” image, but instead of coming from the manufacturer, it is the image with all your key settings.  These images can be stored on a server or NAS for easy retrieval.

As prices of drives come down and size of storage goes up, the quality and reliability is sliding.  Hard drive reliability is something many of us have taken for granted, but as this study shows, failures are becoming more and more the norm.