Does a SSD only store data for 3 months without power?

Question

On this document published by Dell, in section 6, it states:

I have unplugged my SSD drive and put it into storage. How long can I expect the drive to retain my data without needing to plug the drive back in?

It depends on the how much the flash has been used (P/E cycle used), type of flash, and storage temperature. In MLC and SLC, this can be as low as 3 months and best case can be more than 10 years. The retention is highly dependent on temperature and workload.

Does this mean that bit rot will begin as early as 3 months? I have not heard of this happening with SSD drives before. What physics would cause data loss at different rates?

I have an SSD that's been without power for a year yet remains perfectly fine. Does that essentially disprove this theory? — PointlessSpike, May 25 '16 at 07:59
The spread is 3 months to more than 10 years. @PointlessSpike 's SSD is only 10% of the way to the theoretical lower-end of the maximum.. — user1666620, May 25 '16 at 08:58
@user1666620- So 10% of my data should be unrecoverable, right? If that's evenly distributed across the whole drive, the OS should be unusable. But I started it up again a week ago to get it ready to give to someone else, and it was fine. — PointlessSpike, May 25 '16 at 09:01
Well, of course the time the memory can retain data is limited, because we are talking about capacitors, and these leak. But that does not answer *how long* the data retains. — , May 25 '16 at 09:21
It means that bit rot will begin as early as months in SSDs **that have reached their rated P/E cycle (i.e. the rated end of life)**. That's what the column heading "Data Retention @ rated P/E cycle" means. It means the 3 months referes to data retention at the rated P/E cycle. — 410 gone, May 29 '16 at 08:18
@PointlessSpike I think that would be the point at which bit rot exceeds the rated unrecoverable bit error rate - usually one error in 10^14 or 10^15 bits read. You certainly won't be seeing errors in 10% of sectors. — SomeoneSomewhereSupportsMonica, May 29 '16 at 23:19
Temperature matters significantly too. From [here](http://www.eeweb.com/blog/eli_tiomkin/industrial-temperature-and-nand-flash-in-ssd-products), expect ~250x faster degradation at 70C vs 25C. The data decay specs are probably based on whatever the specced maximum storage temperature is. — SomeoneSomewhereSupportsMonica, May 29 '16 at 23:25

score 46 · Accepted Answer · answered May 25 '16 at 09:34

The claim is not true.

To meet JEDEC (Joint Electron Device Engineering Council ) requirements the data retention requirement for a client SSD is one-year at 30°C (source), and for enterprise storage disks it is 3 months at 40°C at the end of its prescribed service life.^*
That last addition is crucial. The data retention attribute changes over the lifetime of a NAND flash-based device (source). When a NAND flash device is new from the factory, it can retain data in an unpowered state (at specific temperature conditions) for many years. However, data retention specs are not given for a new device. Only when the SSD has reached its end-of-life (i.e. after a specified amount of data was written to the drive, we are typically talking about Petabytes here), do these specs come into play.

The second document I quote goes in detail into the relation between endurance and data retention.
Note that it specifically warns:

The implication is that old and well-worn NAND flash storage devices are not intended for archival data storage. At the end of a device’s service life, the device should be retired and the data onboard should be transferred to a new device. The three-month JEDEC spec is set specifically to give the system administrator plenty of time to accomplish this data transfer, if necessary.

^{* So it's easy to see where those '3 months' came from}

I think that the document you have is entirely consistent with the doc quoted in the question. Your document's reference to "the end of its prescribed service life" is equivalent to the OP's "@ rated P/E cycle". It's just that the OP has misunderstood that document. Your document doesn't contradict theirs: it only contradicts their misunderstanding of it. (As does their own document.) — 410 gone, May 28 '16 at 14:32
@EnergyNumbers I think so too. I have seen 'interpretations' click-baiting us with [SSDs can lose data in as little as 7 days without power](http://www.extremetech.com/computing/205382-ssds-can-lose-data-in-as-little-as-7-days-without-power) — , May 29 '16 at 14:01
I would also note that temperature matters a lot: http://www.eeweb.com/blog/eli_tiomkin/industrial-temperature-and-nand-flash-in-ssd-products — SomeoneSomewhereSupportsMonica, May 29 '16 at 23:25

Does a SSD only store data for 3 months without power?

1 Answers1