I never thought there would be such a disagreement with good published statistics. I will just STFU since my suggestion sucks and everyone has better ideas...
Data is data and while it's not absolute, BackBlaze's data set is massive. It's quite possibly the largest study of hard drives in the world. Of the drive models specified, their failure data is an indicator of what might be expected. Sure, one could buy an exact model where BackBlaze's failure rate was at or close 0.0% (the WD or Toshiba models for example) and it might fail quickly in a desktop. That's the nature of statistics and probability, "nothing is absolute". However, their data is a strong indicator of the probability of failure, at least where the models in their data set are concerned.
Further, there are reasons why BackBlaze is compiling this data and it's certainly not for the computing community. This data is being compiled, analyzed and actively applied to save money, period. Toward that end, in reducing tech labor hours, down time, etc., it's an active effort toward buying drives that last longer, even marginally, and influence OEM drive prices. It's a matter of trade offs related to cost. Otherwise, compiling such a massive data set would be an expensive and foolish waste of time. (If that was the case, being so poorly managed, BackBlaze would be out of business.)
While It's possible to poke holes in the collection techniques behind any data set, given the content of this thread, it also appears that there's a great deal of variation in individual interpretation of results.
In balancing out the picture, a quote from Mark Twain comes to mind:
"There are lies, damned lies, and statistics."
Here's an example of what he was talking about. -> Example
(Even if its representation seems distorted, the underlying raw data is accurate.)Good backup takes the "drama" out of computing
OMV 3.0.99 Erasmus
ThinkServer TS140, 12GB ECC / 32GB USB3.0
4TB SG+4TB TS ZFS mirror/ 3TB TS
OMV 3.0.99 Erasmus - Rsync'ed Backup
R-PI 2 $29 / 16GB SD Card $8 / Real Time Clock $1.86
4TB WD My Passport $119
one could buy an exact model where BackBlaze's failure rate was at or close 0.0% (the WD or Toshiba models for example) and it might fail quickly.
I just wonder if anyone can tell me, what is the max capacity of RAID can be built up on a smal ARM development board? Like RPi or Odroid, e.g... I, personally, have recentelytried 1TB, but I wonderwhere are the limits...There is a way to make NAS work efficiantly and stable! Eltechs way!