"Micron's 232-layer QLC NAND is a key enabler of our high-capacity DirectFlash Module," said Bill Cerreta, general manager of Hyperscale at Pure Storage. "Thanks to Micron's NAND innovation, Pure Storage can take one more step on its quest to replace all HDDs in the data center by 2028"

With this horrible strategy of continuing with QLC, no innovation, besides increasing their own profit margins... No, they won't.

Denver"Micron's 232-layer QLC NAND is a key enabler of our high-capacity DirectFlash Module," said Bill Cerreta, general manager of Hyperscale at Pure Storage. "Thanks to Micron's NAND innovation, Pure Storage can take one more step on its quest to replace all HDDs in the data center by 2028"

With this horrible strategy of continuing with QLC, no innovation, besides increasing their own profit margins... No, they won't.

They absolutely will. Your comment reads like you don’t work with this stuff much, or maybe you missed that it was pure not micron that said it.

Pretty much all warm tier storage is already these kinds of drives. It was heavily adopted by all the big providers and is wide spread in bare metal land.

HDDs are considered cold already for the big 3. Hot tiers are your DC class SLC drives but they are in tiny capacities and usually only used for syncing AI clusters or buffering DB shards.

Edit:: oh I see from the other thread, your just on some shitting rampage.

DenverI really hope QLC doesn't become the industry standard. If it were my choice, no product would incorporate QLC: Sacrificing two-thirds (66%) of the useful life for a mere 33% increase in capacity is a foolish trade-off that only serves to line the pockets of companies. It's detrimental to the environment and downright harmful to our pockets.

No doubt QLC is crap, but the discounts on these technologies when purchased in bulk is more than there lower density faster counterparts. QLC may not be good for you, or me, but when they are filling racks with them the cost benefit and access times make it worth there while.

Solaris17They absolutely will. Your comment reads like you don’t work with this stuff much, or maybe you missed that it was pure not micron that said it.

Pretty much all warm tier storage is already these kinds of drives. It was heavily adopted by all the big providers and is wide spread in bare metal land.

HDDs are considered cold already for the big 3. Hot tiers are your DC class SLC drives but they are in tiny capacities and usually only used for syncing AI clusters or buffering DB shards.

Edit:: oh I see from the other thread, your just on some shitting rampage.




No doubt QLC is crap, but the discounts on these technologies when purchased in bulk is more than there lower density faster counterparts. QLC may not be good for you, or me, but when they are filling racks with them the cost benefit and access times make it worth there while.

Nah, let's face the numbers coldly. We are talking about only 33% more density, and 1/3 of the useful life, how is this worth it? From what point of view? Consider this: you'd need three QLC SSDs to match the lifespan of one decent TLC SSD.

When you weigh the price against this trade-off, it's clear the math doesn't quite add up.

Edit:: oh I see from the other thread, your just on some shitting rampage.

What? Where's the moderation? :p

QLC isn't necessarily bad, especially for read sequential workloads like many do with HDDs, though a lot of consumer-grade ones suck.
This single 60TB 2.5" SSD has decent latency and 65PBW endurance (36TBW/day for 5 years). It has 7000MB/s for seq. read and 1M IOPS 4K-QD256 random read.

n-sterQLC isn't necessarily bad, especially for read sequential workloads like many do with HDDs, though a lot of consumer-grade ones suck.
This single 60TB 2.5" SSD has decent latency and 65PBW endurance (36TBW/day for 5 years). It has 7000MB/s for seq. read and 1M IOPS 4K-QD256 random read.

But that's not consumer class SSD. Look if Scamsung or someone else could deliver an 8TB QLC with 5 year warranty and 2400TB wear rating and it performed like a good PCI-E 3.0 ssd I'd buy one if it were under $600.

n-sterQLC isn't necessarily bad, especially for read sequential workloads like many do with HDDs, though a lot of consumer-grade ones suck.
This single 60TB 2.5" SSD has decent latency and 65PBW endurance (36TBW/day for 5 years). It has 7000MB/s for seq. read and 1M IOPS 4K-QD256 random read.

65 PBW is not remotely impressive for an enterprise SSD of that size. I have 15.36 TB drives that get 33.6 PWB. The linked QLC drive is getting less than half the endurance per TB.

It's also an $8,000 SSD, which makes it irrelevant for 99.9999% of people. At that price it's purely an enterprise drive and a drive for the very very very few ultra-entusiasts who have $8K to spend on an SSD (which doesn't make much sense either when you can get smaller cheaper TLC drives and raid them together for much cheaper).
He's using consumer scenarios as if a regular consumer is going to drop 8K on an SSD that large to begin. Drives like that are designed to increase density but density is not a problem outside the data center. Your average consumer or even your extreme homelabber doesn't need to drop $8,000 on an SSD when they can just get 4 drives for half the cost and with double the endurance.

Another nonsense point made in the video is that a regular person would never write enough data to the drive to kill it but it ignores the fact that anyone spending $8,000 on a drive is not normal. He uses video storage as an example but if you are dropping $8,000 on a video storage drive you are insane or have more money than brains.

You are definitely correct in that QLC is good for read intensive use cases but the scenarios and reasoning presented in the video are completely out of touch and I say that as someone who has a bunch of overpriced SSDs.