Smaller NAND and media retention

I had a thread about this a couple years ago, but wanted to bring it back up. As NAND flash gets smaller (50nm > 34nm > 25nm > 20nm...), the cells get closer and endurance goes down. All we ever read from the manufacturers are the "tricks" they use to get around the problem, like wear leveling and other algorithms. But they can't escape the laws of physics.

Does anyone know long these newer SSD (AND flash) drives will retain their data without any powered applied? How many months, or hopefully years of being stagnant? Sure, an SSD is unlikely to see that much downtime in a PC, but I have flash drives that I would love to know if the data will still be present a year or two later.

Case and point: I'm creating a self-booting restore USB flash drive for my Dad's computer and his backup is around 28GB. I figure a 32GB flash drive could hold that and the backup program quite easily. Obviously any form of optical media would be too small, but this would be perfect. It's a flash drive that'll see NO use though until it is needed, if ever.

I have no idea how small the NAND flash is these days on 32GB+ flash drives, but I am interested in how long they are supposed to maintain data integrity. Anyone?

I don't have am answer to your flash longevity question, but there are 50GB and 100GB BD-R's now if you decide that flash media isn't a viable option for you.

ryan_975 said:

I don't have am answer to your flash longevity question, but there are 50GB and 100GB BD-R's now if you decide that flash media isn't a viable option for you.

Click to expand...

That's good to know...thanks. In this instance though, the laptop does not have a Blu-Ray drive.

I'm interested in knowing the answer to this question too. How long will data be accessible on a 128GB USB key or 256GB SSD drive with no power applied?

Greetings

Valnar said:

As NAND flash gets smaller (50nm > 34nm > 25nm > 20nm...), the cells get closer and endurance goes down. All we ever read from the manufacturers are the "tricks" they use to get around the problem, like wear leveling and other algorithms. But they can't escape the laws of physics.

Does anyone know long these newer SSD (AND flash) drives will retain their data without any powered applied? How many months, or hopefully years of being stagnant? Sure, an SSD is unlikely to see that much downtime in a PC, but I have flash drives that I would love to know if the data will still be present a year or two later.

Click to expand...

I have seen a few articles published about this (probably by Anandtech) when there were transitions from one nm size to the next smaller one usually around the time new SSD's came out using the smaller cell size, and I can't remember exact figures. AFAIK there are JEDEC standards that mandate minimum times for retention in an unpowered state and this would be the only authoratative source for this topic.

Temperatures also seem to have a huge impact as well, have a read of this especially page 27 which has a temperature chart, it appears longevity is also affected by the temperature of the SSD whilst in storage and it mentions two additional JEDEC testing standards.

Due to all of this if an SSD has a published retention date of 12 months I think if would be prudent to give it a large margin of error given rises and falls in temperature due to natural weather changes and in that case I'd only

(a) be confident of it retaining data for 3 months
(b) be hopeful of it retaining data for 6 months
(c) be apprehensive of it retaining data for 9 months
(d) be resigned to the fact that it would at least have some errors after 12 months

Too many unknowns I think to be betting on how long data is actually going to survive.

Also does anyone know if simply re-connecting it and supplying power to it would restart the clock or would you have to erase and re-write each cell for the clock to be reset.

Valnar said:

Case and point: I'm creating a self-booting restore USB flash drive for my Dad's computer and his backup is around 28GB. I figure a 32GB flash drive could hold that and the backup program quite easily. Obviously any form of optical media would be too small, but this would be perfect. It's a flash drive that'll see NO use though until it is needed, if ever.

Click to expand...

and

Valnar said:

In this instance though, the laptop does not have a Blu-Ray drive.

Click to expand...

I think you might be better off with an external USB slim-line BD drive and a bootable BD rom image if you can squeeze it into the 25GB disk size unless dual layer 50GB ones are also available, most newish laptops can boot of them if you set the BIOS appropriately.

Cheers

A brand new SSD - even 25nm TLC one - will have about 10 years of retention when you start using it. Only when the flash degrades, will the retention drop significantly. When all the write cycles have been exhausted - on paper at least - you should still have 12 months of retention. So it generally starts at 10 years, then drops to 12 months.

Special SSDs with eMLC NAND memory have only 3 months of guaranteed retention time, but offer more write cycles as compensation.

Do note that if you cause the SSD to lose power unexpectedly, without the SSD featuring power-safe capacitors, this may cause damage to the flash that was currently active, causing it to still work but have vastly decreased retention time versus its siblings. This kind of damage cannot be detected by the SSD controller and will go undetected. You can see the number of unexpected power-loss in your SMART data. You do not need to have had a single power failure for multiple unexpected power-loss to have occurred. Those two have little relationship with each other, despite popular thought.

Well, since my necrothread was brought back...

HobartTas said:

Also does anyone know if simply re-connecting it and supplying power to it would restart the clock or would you have to erase and re-write each cell for the clock to be reset.

Click to expand...

That's still the big question. Can I power up the SSD/Flash drive for 5 seconds, or should I let it sit for 5 hours? How often does this supposed micro-firmware run and check for errors or worn out cells? What happens if you shut your PC down during the cleanup cycle (as people do this every day)?

And for that matter, what if you have a USB flash drive that doesn't have a wear leveling algorithm? Does that mean keeping it in a drawer or keeping it plugged into your PC powered up (with no changes) is the exact same thing? Cells would still die in a year or so.

If I knew exactly how long I have to put electricity through these dormant drives, it would be cake to setup a schedule. The same with refreshing/rewriting the data.

Good questions. I have been told that the retention while powered down is down to 1 year for some products. I did not think of asking how long I have to power the thing up for a complete refresh, and obviously it depends on what you are doing with the thing during that time. I am sure that when something looking like benchmarking hits the drive all safety housekeeping goes out the window.

For what it is worth, products like the Zalman VE400 use a hard disk as storage but can emulate an optical diskdrive. You put the ISO on the hard disk and the enclosure presents the contents of the ISO in the usual way an optical disk drive would, allowing you to boot from it. (I don't have it myself)

@Valnar
Obviously, if you need to store things mid-term, you certainly need to hunt down older SSDs.
I am personally a fan of the Intel 160GB G2 series. I bought a bunch for my ESXi boxes. I traded performance for reliability.

That said, I plan on buying a few Evo pro to host my software build VMs as I don't have to worry about losing anything with those.

Considering that manufacturers are now putting write limits on SSD warranties, we need to take heed of the fact that these things are not meant to last.

Yah, the Intel X-25 G2 series is my bread and butter. I have an old 32GB X25-E (expensive for its time) in my DOS/Win98 box. I rarely boot it up but I figure a good 50nm SSD would have more longevity than a bunch of magnetic platters which don't see juice too often.