Is it possible to predict when my SSD is going to fail ?

[oliveryuan]

Hi,
I have heard that SSDs fail suddenly and without any warning.
Its been some years I purchased my SSD.
Is it possible to predict when my SSD is going to fail ?

 

[drescherjm]

Is it possible to predict when my SSD is going to fail ?

I say No. It can work fine with no negative symptoms at all for 5 or perhaps 10 years and die the next day.

 

[kirbyrj]

You could always look at something like SSDLife. It's not going to help with the random failure, but if you're getting close to the write limits, it will tell you.

 

[drescherjm]

You must have accidentally clicked on underline in the toolbar. Click edit, select your text and click the first button on the left above the edit text box.

 

[likeman]

you can't really predict random fail (like my 840 evo flat out failing) most SSDs are problem free upto 300-400TB of Written data for TLC drives , MLC drives can go into the PB of data ranges (samsung Pro drives above 850 can do over 2PB of data before they fail)

also don't trust that they go into read only mode when they fail (rarely ever works from what i have found they work or they don't)

backup your data (second bigger HDD with full disk imaging software like backupper or acronis true image or alike and a file based backup as well like Windows File History)

 

[_Gea]

Not really predict but there is a statistical relation between some smart counters and failure rates.
At a critical server you may replace disks then. At home do backups and wait until a disk really fails.

ex https://www.backblaze.com/blog/what-smart-stats-indicate-hard-drive-failures/

 

[kirbyrj]

Not really predict but there is a statistical relation between some smart counters and failure rates.
At a critical server you may replace disks then. At home do backups and wait until a disk really fails.

ex https://www.backblaze.com/blog/what-smart-stats-indicate-hard-drive-failures/

Is that relevant for SSDs?

 

[_Gea]

Only for the general question if you can predict failure rates (as an average rate not for a single disk/ssd)

The backplaze stat is for disks but you find similar (not with same number of disks) for ssds that show a statistical relation between failure rates and some error counters,

ex
https://www.usenix.org/system/files/conference/atc17/atc17-mahdisoltani.pdf

https://www.snia.org/sites/default/...Chan-Analysis_SSD_HealthPrediction_Life_2.pdf

 

[oliveryuan]

Thanks for your reply and suggestion.

 

[Zepher]

Here, the one with the most use is going to last more than twice as long, lol

The 128GB was my OS drive and the 256GB was my Game drive.

Both of those drives are still working.

 

[oliveryuan]

Here, the one with the most use is going to last more than twice as long, lol
View attachment 248448
The 128GB was my OS drive and the 256GB was my Game drive.

Both of those drives are still working.

I'll have to try. Thanks!

 

[CraigHB]

My oldest SSD has maybe 10% wear. They seem to last plenty long enough in a personal computer. Chances are you'll replace them due to buying a new computer or bigger drive before they wear out.

You can't predict random failues with SSDs, but the MTBF is quite a bit better than a hard disk drive (MTBF is mean time between failure which is a manufacturer's average of how often a part fails over time). I think it's about tens times longer than a hard disk in terms of consumer parts. The controllers do seem to run pretty warm on NVMe m.2 drives so a heat sink can probably do some good in avoiding random failures.