HDD/SSD reliability (safety) tests and statistics

sich

n00b
Joined
Jun 13, 2016
Messages
20
Are there any sites, blogs, resources about disks safety and theirs percentage of failure? I know only quartal reports of backblaze.
 
I don't know of percentages exactly, but you can write many many TB's to a drive before failure
This is true, but I wonder how many drives (of a definite number) goes down after a few years of active use. For example, we know, that seagete averaged less reliable than hgst, but i wonder in percentages.
 
You don't need to bump threads, this forum gets enough activity as it is, if someone has some info to offer you'll know it. ;)

I think Backblaze is the one that most people are interested in considering the "torture chamber" they put their drives through to produce the reliability studies every so often. While they're far from typical usage, individual sites like the TechReport one linked above show long term reliability of their own "torture" of a few particular SSDs and is useful to most people as a yardstick to measure reliability and long term usability from.

In "average Joe" usage situations I doubt most drives will just up and die on someone randomly - it's usually something else that triggers a failure whether it's component failure on the drive's electronics (actual mechanical "death" for a hard drive is fairly rare believe it or not), an issue with the host machine's power supply (bad power supplies kill more machines than most anything else by pushing bad/dirty energy of unreliable amounts or sometimes too much), bad capacitors in the power regulators, and so on.
 
Hardware.fr has RMA rates for PC components
Pugetsystems also has released similar reports.
Google has done a study on hard drive reliability (not very useful for consumer use though).

Something to keep in mind with SSD "reliability" tests is that the write to failure method isn't very useful (and really misleading). The actual issue with SSD write wear is that the cells wear out over time and therefore affect data retention. So unless you test for latter claiming failure at X number of writes is misleading.
 
Back
Top