NAND flash prices expected to plummet 8-13% in Q3 While semiconductor supply chains remain a mess, if you’re in the market for a new SSD, you may be in luck. The threat of a recession and increased competition in the memory business is causing NAND flash pricing to plummet, TrendForce reports. According to data from the market research biz, slowing demand across all segments …
About pricing things high and then making them difficult to even buy is that everyone starts to just make do with what they've got.
We've also reached a point where unless you've got a very intense workload upgrades aren't the big technical leap they used to be every few years.
"And with more chipmakers bringing 176-layer QLC SSDs to market"
How reliable is this cost-cutting technology? Out of 8 ssd's I own, 2 are TLC, and both return corrupted data, which is worse than outright failure.
(1*samsung,1*WD)
Can I buy some more MLC drives anywhere? Or do I have to buy a large pool of QLC turkeys and run 2* ZFS-Z3 to get reliable data storage?
Question about layers that I couldn't find the answer to: How many layers per cell?
> Can I buy some more MLC drives anywhere? Or do I have to buy a large pool of QLC turkeys and run 2* ZFS-Z3 to get reliable data storage?
If this is anything like the situation with MLC NAND that I've worked with [NB: IANAE, and this was not recently], setting the device to SLC mode alleviates some problems at the expense of storage capacity (and is now supported in kernel 5.8 according to https://www.phoronix.com/news/Linux-5.8-NAND-MLC-SLC-Emulate).
...Some time ago, I worked on board bringup for a system with such a device and the manufacturer was using a kernel that predated the patch that refused to work with it. Some of my older colleagues later noted the board had the codename "Procrustes", and winks would sometimes be exchanged if everyone was out of the line of sight of management :(
On a later assignment for a different employer, our client was stuck on a legacy OEM kernel where the board had the same device and same kernel* - I would have seriously considered running away screaming had I been forewarned! Patches to enable SLC mode emulation for this board were available by then though, and ultimately what we suggested got deployed [...and whether this is still recommended is no longer something I need to know].
(* plus, I learned, a PROM requiring a FAT filesystem which was conventionally mounted read-only by the kernel if at all. To this day I'm not sure if the "random" read disturb [yep - a hardware-level "fix" where adjacent cells were unaffected but the problem didn't go away] meant a regular filesystem check on it would risk more damage than leaving it unmounted ... in the first case we didn't know enough to consider testing things to destruction; the second time round I had to answer every request to try with "do we have a second spare board yet?")
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
