3 ms·

For a 5-year life span 1M hours MTBF (114 years) mean the probability of an error for a single drive is about 4.3% (1-e^-(5/114)). That equals 0.86% Annual Fai

• TylerE 7y ago
  Those rates are surprisingly low considering that's spinning iron. I'd love to see a similar data set for SSDs. Of course, it also depends on what you count as a failure. Is a drive that can no longer write but can still read a "failure"? Probably - but that's a considerably softer way to fail than a hard drive with a bad bearing or head.
  • imtringued 7y ago
    A lot of SSDs that fail early, fail because of their firmware. That MTBF probably only covers the reliability of the flash chips but not the controller.
• Dylan16807 7y ago
  > 0.86% So my comment is 100% nitpick but I do want to make a point here. X year failure rate divided by X is not the same as AFR. AFR will always be higher, because it has to compensate for each year's pool of potential failures shrinking. The correct equation is already there, 1-e^-(n/114), and the correct number is .87%. And while that's only a percent off, it's lucky to only be a percent off. If MTBF was 113 years it would be .86 vs. .88, which is the difference between two digits of accuracy and one and a half digits of accuracy. Losing a quarter of your accuracy is not great!