Unpowered SSD endurance investigation finds data loss and performance issues

Up through 2019 or so, I was relying on BD-XL discs, sized at 100GB each. The drives that created them could also write out M-DISC archival media, which was fearsomely expensive as a home user, but could make sense to a small business.

100GB, spread over one or more discs, was plenty of capacity to save the critical data, if I were judiciously excluding disposable stuff, such as ripped CD audio.

Tapes are also way more reliable than hard drives.

That's not so enticing when I could get 3 16TB hard drives for half the price, with a full copy on each drive plus some par3 files in case of bad sectors.

and consumers would rather have more storage

No one from the manufacturers tells them that the "more storage" - multiplicatively more - lasts exponentially less.

For the same price, would you rather have a 1TB drive that will retain data for 10 years after having written 100PB, or a 4TB one that will only hold that data for 3 months after having written 2PB?

That requires support software they may not want to build

The software is already there if you know where to look.

and support requests from users complaining that toggling SLC mode lost all the data or toggling QLC mode back on did similarly

Do they also get support requests from users complaining that they lost all data after reformatting the drive?

It’s a valid business decision to not support that kind of product feature.

The only "valid business decision" is to make things that don't last as long, so recurring revenue is guaranteed.

Finally, the "smoking gun" of planned obsolescence: SLC flash requires nowhere near as much ECC and thus controller/firmware complexity as MLC/TLC/QLC. It is also naturally faster. The NRE costs of controllers supporting SLC flash is a fraction of those for >1 bit per cell flash. QLC in particular, according to one datasheet I could find, requires ECC that can handle a bit error rate of 1E-2. One in a hundred bits read will be incorrect in normal operation of a QLC storage device. That's how idiotic it is --- they're operating at the very limits of error correction, just so they can have a measly 4x capacity increase over SLC which is nearly perfect and needs very minimal ECC. All this energy and resource usage dedicated to making things more complex and shorter-lasting can't be considered anything other than planned obsolescence.

Contrast this with SmartMedia, the original NAND flash memory card format, rated for 100K-1M cycles, using ECC that only needs to correct at most 1 bit in 2048, and with such high endurance that it doesn't even need wear leveling.

Also consider that SLC drives should cost a little less than 4x the price of QLC ones of the same capacity, given the lower costs of developing controllers and firmware, and the same price of NAND die, yet those rare SLC drives which are sold cost much more --- they're trying to price them out of reach of most people, given how much better they actually are.

Prosumers/enthusiasts generally have three options for large-scale data backups (18TB+), and none are as remotely affordable as the original storage medium:

* A larger storage array to hold backups and/or versions as needed (~1.25x the $ cost of your primary array to account for versions)

* Cloud-based storage (~$1300/yr from Backblaze B2 for 18TB; AWS Glacier Deep Freeze is far cheaper, but the Egress costs per year for testing are comparable to B2)

* LTO drives ($3300 for an mLogic LTO-8 drive, plus media costs)

Of those, LTO drives are (presently) the only ones capable of having a stable "shelf life" at a relatively affordable rate. Most consumers with datasets that size likely aren't reading that data more than once or twice a year to test the backup itself, and even then maybe restoring one or twice in their lifetime. LTO is perfect for this operating model, letting users create WORM tapes for the finished stuff (e.g., music and video collections), and use a meager rotation of tapes for infrequent backups (since more routinely-accessed data could be backed up to cloud providers for cheaper than the cost of an associated daily LTO backup rotation). LTO is also far more resilient to being shipped than HDDs, making it easier to keep offline copies with family or friends across the country to protect your data from large-scale disasters.

It's the weird issue of making it cheaper than ever for anyone to hoard data, but more expensive than ever to back it up safely. It's a problem that's unlikely to go away anytime soon, given Quantum's monopoly on LTO technology and IBM's monopoly on drive manufacturing, making it a ripe market for a competitor.

I'd still love to see someone take a crack at it though. The LTO Consortium could use a shake-up, and the market for shelf-stable tape backup could do with some competition in general to depress prices a bit.

Sometimes, unfortunately, a user will value the data more than they can afford to properly replicate/secure it it, and compromises must be made. Also, Tape does require a higher buy in than most (individual) SSD's/HDD's demand before you can even start investing in actual storage media, even if going used, so there are absolutely valid contexts where the "right"/best available approach is just throwing $200 at second HDD and cloning/mirroring it, but best available compromise in any given specific context is a separate discussion from general best practices or best value in broad terms.