SSD Write Cycle question.

[Coffee999]

I understand that an SSD only has so many write cycle's in it's life. But if a drive is filled with data that rarely changes, and is only accessed but not written to or erased, would this mean that it's life is greatly extended because there are very little write cycles? And, if this is the case, would it be out of the question that an SSD lives for 5-10 years? Thank you.

And I guess I have to ask: Has anyone had a SSD fail because of a too many write cycles, or other death?

 

[Denpepe]

If you want to see how long an SSD lasts, check this out https://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead/

in short If you use then normally you should be able to use it for 50 ish years bye which time you would have already replaced it for something better, but ofc there is the odd outlier that will fail early so make backups.

 

[aztekk]

I have been using SSD's since 2008 and have not had one outright fail on me yet, but it has happened with USB flash drives on several occasions. I would say that controller failure is the primary thing you should be worried about under normal use, this is what most often kills flash storage in my experience.

Another thing to consider is data corruption, which used to be a big problem: decades ago I was using early Sony flash storage cards/cartridges (<8MB) which later evolved into the Memory Stick standard and those casually corrupted all data beyond readability within a few months. This has improved a lot in the past 20 years with new better designed controllers and NAND chips. For example the Samsung K9WAG08U1A SLC NAND chips that were featured in my first SSD drive 11 years ago show a guaranteed data retention spec of 10 years in the datasheet. This may be lower for MLC/TLC/QLC chips (in theory, I'm sure the manufacturers will still spec to 10y anyway) so I would be conservative and refresh any data you need to keep on them every 3-5 years. Of course a backup is always a good idea.

It's important to note that the failure mode for write cycle depletion is not catastrophic like with hard drives. Instead when the counter reaches the maximum figure the SSD will go into a read-only mode allowing you to transfer your data off assuming the electronics are functioning otherwise.

 

[Maxx]

Yes and no. Yes because, obviously, you're not doing a lot of writes. No because the data will get fuzzy over time and the controller will read the data and if it has difficulty doing so - that is, the data is fuzzy enough to require ECC - it will rewrite it. This is known as static data refresh and was first utilized on the second fixed firmware for the TLC-based Samsung 840 EVO. In that case, it was noticed because performance suffered - ECC will have a read latency penalty that grows with sensing level. Modern controllers/drives will periodically read and refresh. This does not apply if the data is "cold" (not powered) as then you can eventually suffer bit rot (where the drive thinks the NAND is worn and marks it as a bad block).

I've never had a SSD fail. My original pair from 2010 are not even worn 10%, although they utilized compression which reduced write amplification to sub-1.0 and are also MLC-based. Some of my newer, TLC drives are around that point, but usually vendors tie SMART life remaining to TBW which generally underestimates write endurance significantly (2-3 times). Good 3D 64L TLC - say, Intel/Micron or Samsung - can be good up to 10,000 P/E in reality. Which could be centuries with typical consumer use. The drive will fail in some other way before that.

 

[EniGmA1987]

I understand that an SSD only has so many write cycle's in it's life. But if a drive is filled with data that rarely changes, and is only accessed but not written to or erased, would this mean that it's life is greatly extended because there are very little write cycles? And, if this is the case, would it be out of the question that an SSD lives for 5-10 years? Thank you.

And I guess I have to ask: Has anyone had a SSD fail because of a too many write cycles, or other death?

I had a SSD fail last year that was only used for around 2-3 years. It was a plan OS drive usage. The computer had a second SSD that hosts shared files, and 2 other computer accessed those files daily. I suspect that the files are opened as a temp file on the OS drive, and was causing extra writes. Either way, I was quite surprised that a 250GB drive with barely 70GB of my own files (including OS ones) would fail in a couple years from what I thought was so little access. But Windows popped up a warning one day about low write life left, so I used Crystal Disk Info to check and sure enough it was at 5%. As I watched over the next week it went down to 2%, so something was hitting it quite hard with what seemed very little usage to me.
So ya, they can and do wear out. I typically replace drives every couple years so I have not had another one wear out on me, but I suspect that if I had been keeping drives longer that I would have seen many more failures.

I do have some drives going on 5 years old in my personal gaming PC that have not worn out, but they are just used for gaming drives. My OS SSD in my personal gaming computer gets replaced every year or two as I need a bigger one or faster one so I havent seen if that would wear out too.