Nice to see that the WD20EADS has the lowest rate among 2TB drives; I have have 3 of those in my Drobo now. Overall I’m surprised how high these numbers are; is “failure” defined as DOA or any failure within the warranty period?

I don’t think it’s necessary to break down the SSD’s by capacity. This is good enough to get the feel of what’s happening. Everything I’ve read about this issue has said that SSD’s are about as reliable as HDD’s. 1 year ago, they were less reliable, so there’s hope that they’re getting better.

Interesting, but there are several issues that, from a ‘study design’ perspective render these data all but uninterpretable. Among these are:

1. How ‘failure’ is defined (I.e. were all failures true failures, or were some proxy events, like returned drives that might or might not have failed)

2. Who measured the failure events, and therefore was the definition consistently applied

3. How were the failure events identified; passively by having consumers return them to a retailer, or actively by interrogating consumers about their experiences, this latter method being less likely to result in surveillance bias, and

4. How many samples (drives) are represented in each category, that is, what is the denominator in each category of drive. The fewer the the number of drives, the more sensitive the results are to small changes between categories. For example, if 2 of 4 drives fail in one group, and 1 of 5 in another, it is reported as 50% vs 20%, a seemingly large difference, even though the numbers in each group are small - and over time may have resulted in 2 vs 1 out of 200 in each group, if failure events are infrequent. Denominators matter.

It would also be helpful to know how the drives were used (e.g. internal laptop vs external), which might affect longevity.

In short, the data may not have collected in such a way as to make the attempted comparison, and may reflect little more than the prevlence of background noise.

It does seem, on the surface, that the SSDs should be more resilient than mechanical drives, and was one reason why NASA chose solid state drives for their Mars rovers.

More information about these data would make the report more interpretable.

What you’re saying is useful when analyzing why the drives failed, but for everything else, simple numbers are good enough. If a vendor sells 1,000 of one model drive, and 50 come back as not working, and another that sold 1,000 have 20 come back, then the first drive is failing more often. As consumers, that’s all we need to know.

It’s very significant how a drive fails. Normal consumer can’t just swap the failing disc from a raid. Most people don’t take enough backups.

For all of my hard drive failures I have been able to recover all or most of the data, before sending the malfunctioning drive back. The situation is very different if the drive stops working suddenly and completely.

goodjob !
but i do not belive hitachi is the Black Brand(under 1T)
IT IS Very stable~
At least I never meet Flaw one few years