SoylentNews is people
SoylentNews
Using a glass substrate instead of aluminum could allow 12 platters to be crammed into a 3.5" hard disk drive enclosure:
Even if many modern systems eschew classic hard drive storage designs in favor of solid state alternatives, there are still a number of companies working on improving the technology. One of those is Hoya, which is currently prototyping glass substrates for hard drive platters of the future which could enable the production of drives with as much as 20TB of storage space.
Hard drive platters are traditionally produced using aluminum substrates. While these substrates have enabled many modern advances in hard drive technology, glass substrates can be made with similar densities, but can be much thinner, leading to higher capacity storage drives. Hoya has already managed the creation of substrates as thin as 0.381mm, which is close to half the thickness of existing high-density drives.
In one cited example, an existing 12-terabyte drive from Western Digital was made up of eight platters. Hoya believes that by decreasing the thickness of the platters through its glass technology, it could fit as many as 12 inside a 3.5 inch hard drive casing. That would enable up to 18TB of storage space in a single drive (thanks Nikkei).
When that is blended with a technology known as "shingled magnetic recording," 20TB should be perfectly achievable.
Toshiba is reportedly planning to release a 14 TB helium-filled hard drive by the end of the year.
Also at Network World.
(Score: 2) by takyon on Thursday September 14 2017, @04:43AM (3 children)
If it doesn't reach a certain level of endurance, it simply won't be adopted by enterprise customers. They are going to be the first ones to get their hands on QLC NAND, not you. And they have more demanding workloads (full drive writes) than typical consumers. Facebook is particularly thirsty for QLC NAND [tomshardware.com] (alt [techtarget.com], note: "We asked for QLC flash a few years ago, and we ask for it again").
Toshiba has suggested that QLC could have a similar endurance to TLC [soylentnews.org]. A full 1,000 write cycles instead of the 100 write cycles that had been predicted. That sounds unbelievable, but then again we've heard of labs working on methods to try to get endurance something like a million times better [ieee.org]. Or maybe the endurance problem doesn't scale exponentially like the 2^n states do, and 1,000 write cycle QLC is believable.
If you meant speed, most consumer PCs don't really need super high transfer speeds beyond 500 MB/s. It was the increased random IOPs that really made the difference for people switching from HDDs to SSDs. Even a crappy SSD should be able to do 2-3 orders of magnitude better than an HDD on random IOPS.
There are also tricks to make the QLC SSD perform better, especially in consumer or enthusiast drives meant to hit higher speeds/IOPs. For example, QLC could emulate SLC/MLC (using 4-8 of 16 states to pretend to be 1-2 bits per cell). Or you can include SLC/MLC as a cache, DRAM as a cache, Intel 3D XPoint as a cache, etc. You can do extreme overprovisioning to help boost speed and endurance since QLC drives are going to have high capacities.
When QLC SSDs do hit the consumer market, review sites will be all over them on day 1 if not before. So unless you preordered your SSD, you should at least have an idea of what you are getting.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Azuma Hazuki on Thursday September 14 2017, @05:11AM (2 children)
Where *is* 3D XPoint, by the way? I remember them wittering on about how it was going to eat SSDs alive, but all that's out there isn't even bootable. Just some Optane cards that are basically the SSD portion of "hybrid" SSHDs on massive amounts of 'roids.
When they can make a 10 TB hunk of 3D XPoint at a reasonable (well, reasonable by datacenter standards...) pricing, *then* we'll see some fireworks.
I am "that girl" your mother warned you about...
(Score: 3, Informative) by takyon on Thursday September 14 2017, @05:46AM (1 child)
3D XPoint [wikipedia.org] a post-NAND technology [theregister.co.uk] (also referred to as storage class memory [tomshardware.com]) along the lines of Crossbar's RRAM or HP's Memristors. It occupies a memory/storage tier in between DRAM and NAND. It is denser than DRAM, cheaper than DRAM, slower than DRAM, and non-volatile unlike DRAM. It is less dense than NAND, more expensive than NAND, and anywhere from somewhat faster than NAND to order(s) of magnitude faster, highly dependent on the type of workload. This article explains it pretty well [arstechnica.com].
Now you have had several companies working on post-NAND for years. Crossbar [wikipedia.org] has been hyping [theregister.co.uk] their shit for years. They have promised all around better specs than NAND and the ability to store multiple terabytes in a postage stamp-like form factor [theregister.co.uk].
So Crossbar, HP, Micron, IBM, Crocus Technology, Unity Semiconductor, Samsung, Toshiba, Spin Transfer Technologies, and other companies have all FAILED to get a legitimate post-NAND technology onto the market outside of some megabyte-sized cache products. Intel comes along with their concept and has apparently rushed it to market (although it has been delayed somewhat). It's in the post-NAND category, but it's not a true replacement for NAND since it doesn't try to compete with NAND on capacity/density and cost per bit. Intel is definitely not giving us terabytes of this stuff in postage stamp size anytime soon.
I will say this much for XPoint. They have 16 and 32 GB modules at reasonable prices [soylentnews.org]. They can be bought today - the 32 GB one is around $75-$80. So if your motherboard happens to have an M.2 slot, and you wanted to try this out as a kind of non-volatile boot drive, it would seem like you can do it fairly cheaply. Here are benchmarks [legitreviews.com]. DRAM prices have increased in the last 2 quarters, btw.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by Azuma Hazuki on Thursday September 14 2017, @04:11PM
LOL, riiiight after posting my last post I went "hmm, maybe I should Google this..." and found all that information. It's odd how the 16 and 32GB devices have worse sequential write than an NVMe NAND SSD, but the read speed looks good. And the 4K, low-queue-depth reads especially. I have only a 20GB root partition, so this would be viable for a Linux boot drive.
I am "that girl" your mother warned you about...