Question SSDs & data retention in various setups (laptop/desktop, internal/external SSDS)

[sirhawkeye64]

I know that SSds need power to retain their data over long periods of time but was wondering what the retention period usually is on average for a few different scenarios, such as:

1.) When it comes to internal SSDs in laptops and desktops, do these computers generally feed power to the SSD to retain data even when the PC or laptop is powered off (but still plugged in)?

2.) What about external SSDs or internal SSDs that are not plugged in full time (ie. used in a USB HDD dock). How long can these retain data without power?

I'm thinking of using a larger SSD for cloud syncing in a computer that may not be turned on all the time (maybe once per week or at least once per month. Now I figure once per month is probably enough to maintain the data on the SSD but wanted to be sure. I know that using SSDs for storage -- or long term storage -- is not a good idea, and that's not exactly what I'm doing here -- it's only for syncing to the cloud as I have everything else backed up on regular HDDs. I only am going to use the SSD because I need a 2.5" drive and the largest HDD I see in that size is only 2TB. So it looks like SSDs are the only option in this case for this use, short of going to an enterprise-level 2.5" HDD.)

 

[helpstar]

depends on your ssd nand technology.
at least one year
worst case scenario, powered on at 25°C then powered off and stored at 55°C -> only weeks

https://www.starline.de/en/magazine...ata-retained-on-an-ssd-without-a-power-supply

 

[Misgar]

1.) When it comes to internal SSDs in laptops and desktops, do these computers generally feed power to the SSD to retain data even when the PC or laptop is powered off (but still plugged in)?

It's unlikely the +5VSB supply is used to maintain M.2 SSDs, but I'm not going to poke test probes on any of my drives to prove/disprove the theory.

You should be OK leaving a consumer-grade SSD powered off for 1 year if the ambient temperature is 30°C or lower.
https://www.anandtech.com/show/9248/the-truth-about-ssd-data-retention

 

[Maxxify]

I know that SSds need power to retain their data over long periods of time but was wondering what the retention period usually is on average for a few different scenarios, such as:

1.) When it comes to internal SSDs in laptops and desktops, do these computers generally feed power to the SSD to retain data even when the PC or laptop is powered off (but still plugged in)?

2.) What about external SSDs or internal SSDs that are not plugged in full time (ie. used in a USB HDD dock). How long can these retain data without power?

A normal consumer drive will rely on the host clock and/or a power-on state to determine if flash needs to be read and reflashed. So, this requires more than passive power, although theoretically it's possible to have external devices that can wake drives. In real world use, people rely instead on simply accessing the drive from time to time. A modern consumer TLC drive with minimal writes (e.g. let's say, 100 or less drive writes) can maintain data for a long time, but I wouldn't want to rely on a guess. It's probably in the 10+ year range. JEDEC ratings aren't valid for this as they are done after intense and accelerated flash wear where the fitness rating would be 1/10th at best (hence, 1 year JEDEC could be 10 year with a basically fresh drive). This is one reason many people have flash drives in drawers that are readable a decade later (older drives could be SLC/MLC based, which I have going back even further, and those retain even longer, but older drives used inferior ECC and management algorithms as well).

 

[sirhawkeye64]

Ok so it sounds like at least one year, and probably longer (as the drives are kept at or below room temperature( so under 80 degrees F at most, which is about 26-27 degrees C.

Does powering up the drive and then storing it for another year sort of "reset" the proverbial timer on the data rentention, ie. it should be able to hold it for another year or does it not work this way? The drives in question likely will be powered on atl east once every few months, but they won't be used daily or weekly (monthly is a possibility).

 

[Maxxify]

The drive stores date stamps as metadata ("last read") and maintains metadata in terms of groups of blocks wherein the blocks in any group share characteristics based on page sampling. This is because not all word lines in a die perform or act the same, e.g. top layers program slower but have better data retention vs lower ones. Some blocks can handle more cycles than others, etc. So basically it knows when to check and rewrite data based on various metadata but it pulls any date time from the host in most cases. It also has to run through its startup procedure successfully to refresh data. There are patents that describe all of this in greater detail, I'm given a rough overview.

Many things impact data retention. Environment being one: humidity and temperature, as well as cosmic rays if you want to really get down to it (space SSDs have different requirements here). Lower temperature being better, which is what JEDEC hints at. But even workload type can impact retention, which is why JEDEC has different values for client and enterprise workloads. But basically it's just a matter of the cell charge not deviating so far from thresholds that it's not reparable through ECC and if needed, parity rebuild. The thing is, drives can be very good at rebuilding because data skew is predictable to some degree (e.g., threshold skew from just being powered off is predictable) so with a time update a drive could reasonable restore data even after many years if the NAND hasn't had much wear.

For the record, if keeping data like this it is best to do a full read/scan in order to force the drive to check and refresh data, and you can also do a re-image (as in backup then restore that to the original drive so all data is refreshed once, which counts as a drive write) to refresh data.