SSD unpowered Data Retention

[Wise Old Owl]

I thought I would post some recent findings as there has been a ton of contradicting information about the ability of an SSD as long-term data storage.
Now for context, I started using SSDs back during the HDD drought about 15 years ago? due to flooding in Taiwan closing all the HDD production, with my first being an Intel 320 series 120GB (which is still going by the way)
Anyway, with a ton of older, small capacity drives lying around, I started looking to use them for storage of large files like .RAF which can go over 250GB each.
During this I just found an SSD from an old PC build I did for my DAD years ago to act as its C: drive.
This drive is an OCZ Agility3 120GB one, and it was pulled from that PC and put into storage on the 19th of Jan 2016 and never touched since. Also, this drive had been in use for several years before it was pulled.
Today (7th April 2024) I just plugged it in and all data is present and correct almost 9 years later.

 

[arknor]

Depends on the drive. it can be up to 20 years data retention, although probably not for consumer level products

I'm assuming they will have a capacitor(s) inside that hold the charge needed to power the memory cells.


I bet they use so less power when the drive is inactive that it would be possible to power one via a built in thermoelectric engine, like some watches are, or via a similar method

 

[KompuKare]

Problem with the OP's sample of one is, well, that it is a sample of one.

I rather suspect the NAND manufacturers themselves have an idea of what each of their cell designs can do.

Whether the SSD and controller manufacturers do? Maybe the fully vertically integrated ones like Samsung, Micron/Crucial, WD/Sandisk, etc. might do.

But I feel that the each cell design part is important.

So not even as simple SLC/MLC /TLC/QLC and/or how many layers, but almost each revision of the basic cell.

 

[Cyber-Mav]

your test shows unpowered for 8 years. indeed data retention is not an issue at this age however you should have benchmarked the drive.
i think i did a thread on here with my findings on a samsung sd card that sat in my camera for over a year and not used. the photos were all intact but the read and write speed of the card had tanked hard.
after a full format the card had its speed back.

 

[Wise Old Owl]

your test shows unpowered for 8 years. indeed data retention is not an issue at this age however you should have benchmarked the drive.
i think i did a thread on here with my findings on a samsung sd card that sat in my camera for over a year and not used. the photos were all intact but the read and write speed of the card had tanked hard.
after a full format the card had its speed back.

I had it connected via a USB 3 adapter, and I was able to hop around random files and open them in their apps without noticing any performance difference. A lot of over analysing on here, I merely posted the information as this is a real hard drive, with real use that has had a real near 9 years sat in a cupboard. This is also a pretty naff drive and small capacity. Most threads online about this involve simulated data. Whilst I am not telling anyone how to go about storing their data, I'm sure there are those that will find it of comfort that such a drive can indeed still hold its data after nearly a decade without power. It's a very different thing to say it should do this or that vs it actually happening.
As for benchmarking it, why? I would never entertain using it again as a functioning drive, rather I will continue to use it for long term storage.

 

[Cyber-Mav]

As for benchmarking it, why?

the whole point of a long term storage test, to see if data is retained and at what performance cost. And opening random files is not a good test. need to test every file, who knows the few files you opened could have been the only good ones on the drive?

 

[Mr Evil]

I always link to this article when SSD retention comes up: https://www.anandtech.com/show/9248/the-truth-about-ssd-data-retention

My own experience is that there is wide variation, with some drives keeping data intact for many years, and others for much less. I've had some cheap USB thumb drives that will lose data after only a couple of months. Newer flash squeezes more data into less space, and so has worse retention.

Personally, I wouldn't trust an SSD to hold data unpowered for more than about a year. Note that just because you can read the drive doesn't mean it hasn't lost data. You need to checksum everything, otherwise you might not notice if there are a few missing bits and bytes here and there.

 

[BananaDunka]

I think the whole thing has long been overblown by forums and internet chatter without anyone actually providing any meaningful metrics and objective testing. People just read an article online and the rest is history, suddenly SSD retention becomes a thing for unpowered drives yet we've had unpowered flash media for decades, when was the last time a USB stick lost all its data after being in a drawer for years?

Exactly!

IMO it's mostly a non-issue whereby in all these years I've yet to personally see or hear of data loss on any sort of flash media that hasn't been powered for x amount of time. I've also worked in IT where we store laptops and desktops as backups that literally never get used because they are emergency backups to hotswap in, these are running SSDs and go years without being powered on yet still retain all function when booted up one day to replace a broken active machine.

The other problem is cheap USB drives are often cheap for a reason, the controllers aren't engineered for longevity. This is why you always buy something from a reputable brand and not cheap out, even things as basic as USB drives.

Edit* The link above, it specifically states:

"Remember that the figures presented here are for a drive that has already passed its endurance rating, so for new drives the data retention is considerably higher, typically over ten years for MLC NAND based SSDs."

So yes, of course once the endurance rating has been reached then of course data loss is to be expected, the health of the data blocks on the flash medium are past their shelf life (literally).

 

[Mr Evil]

...without anyone actually providing any meaningful metrics and objective testing...

I'd love to provide better data, but doing so would require buying a lot of drives and then waiting for years. I have neither the time nor the patience for that, and I doubt any other normal users do either.

...we've had unpowered flash media for decades...

We have, but as flash cells shrink so it becomes easier for the electrons to fall out. Each new generation also stores more levels per cell, so fewer electrons need to fall out before they cease to be in a valid state.

...when was the last time a USB stick lost all its data after being in a drawer for years?..

As I said in my previous post, I've had it happen to me, with both USB sticks and SSDs, after varying periods of time. It's never all the data disappearing though, but rather files becoming corrupt - maybe just a few, or sometimes a lot of them.

 

[Drumroll]

Had few OCZs showing issues. Long gone. Surprised yours still alive, despite the long time unused, but still. The manufacturer died before your SSD or your data.

 

[OspreyO]

I've had a rare USB FlashDrive become corrupt. Same with a SansDisk SSD. But both ok after reformatting. I usually have these encrypted. So that takes a bit of time to recreate.

The issue if you get partial corruption, is if you don't notice and it gets copied to synced drive. I've had to restore back to an older copy a couple of times.

 

[ATIorNvidia]

I had a WD Blue 3D SSD that had the bug where old data reads extremely slowly (single digit kb/sec speeds) but no actual data corruption