Slashdot Mirror
SSD Latency, Error Rates May Spell Bleak Future
Lucas123 writes "A new study by the University of California and Microsoft shows that NAND flash memory experiences significant performance degradation as die sizes shrink in size. Over the next dozen years latency will double as the circuitry size shrinks from 25 nanometers today, to 6.5nm, the research showed. Speaking at the Usenix Conference on File and Storage Technologies in San Jose this week, Laura Grupp, a graduate student at the University of California, said tests of 45 different types of NAND flash chips from six vendors using 72nm to 25nm lithography techniques showed performance degraded across the board and error rates increased as die sizes shrunk. Triple-Level NAND performed the worst, followed by Multi-Level Cell NAND and Single-Level Cell. The researchers said MLC NAND-based SSDs won't be able to go beyond 4TB and TLC-based SSDs won't be able to scale past 16TB because of the performance degradation, so it appears the end of the road for SSDs will be 2024."
Because there could *never* be a breakthrough discovery/invention found within the next 10 years.
There will be other solid-state storage solutions. The only reason NAND is currently used is its relative cheapness and reliability.
An old study (well, executive) showed that there was a world wide demand for "maybe 6" computers. This might all be true at current technology levels but technology will have changed an awful lot by 2024.
... always denies other areas of innovation. The same way processors were thought not to scale down to x nm and we're at 20'ish nm now. The same way hard drives were thought only to have x capacity and we're now in the terabytes. If nand is really so limited then something different then nand will take it's place. But a few terabyte will be more then enough for 99% of applications and hard disks will be for packrats and those who need large amounts of longer term storage.
Yes, please send your SSDs to me for disposal, thanks.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
Still waiting for the Holographic Memory that should have been hear a decade ago.
-- By all means let's be open-minded, but not so open-minded that our brains drop out.
Yes. They'll all stop working then and it will become impossible to make any more.
Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
From the article, "This will reduce the write latency advantage that SSDs offer relative to disk from 8.3x (vs. a 7 ms disk access) to just 3.2x.". Yeah, doom and gloom.
But I'm choosing to ignore it all, entirely based on font.
http://cseweb.ucsd.edu/~lgrupp/CV.pdf
Yeah, about that 4TB limit, I think these folks will be surprised that their 5TB and 10TB drives won't be possible in the next few years....
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Can't scale past 16TB? Why not just stack them?
Well, not so much that but rather than hard drive rotational latencies will finally catch up to nand. With our disks spinning at a paltry 100,000,000 rpm, latency will finally be a worry of the past.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
No, no one remembers it. It's an invention of fantasy, not memory.
http://en.wikiquote.org/wiki/Bill_Gates#Misattributed
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
The funny thing is that this "myth" was well established long before he published any sort of rebuttal.
This is ancient history. It happened decades ago and finding evidence now would be difficult even if you knew where to look. Chances are that any such corroboration faded away by the time that rebuttal was published.
A Pirate and a Puritan look the same on a balance sheet.
I have an 8MB SD card from the first camera I bought (in about 2003). Because of the small size, we immediately replaced it. I found it the other day (late 2011) and I was able to read a couple test pictures just fine over 8 years later. I can read my first CD-R's still too. I don't believe any of this digital media rot stuff. I haven't seen it happen at all in anything that supposedly rots.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
While they discuss individual SSDs, modern flash storage arrays ( http://www.violin-memory.com/products/6000-flash-memory-array/ ) can hide all the write latency and its effects on read latency. When you start talking about 16TB SSDs the same techniques can be used.
As far as bandwidth and IOPs, they use a 4K/8K write size for MLC/TLC, but MLC already exists with 8K pages, as well as having the ability to write more than one plane at once, which doubles the write bandwidth. Double the page size again and you double the BW.
Now bigger page sizes only help on the reads if you can use more than a single user read worth of data in the page, which might be possible depending on what the system knows about access patterns. But without making assumptions about the ability to store data together that's likely to be read together, garbage collection, which can wide up reading more bytes than the user does, can use most of the data in a page.
So there are factors of 2X, 4X maybe 8X in performance that the paper misses out on.
As far as density, it is not necessary to go to smaller features to get more bits per chip by using 3D techniques such as Toshiba's P-BiCS (Pipe-shaped Bit Cost Scalable) MLC NAND which allow vertical stacking which increases density without using smaller features with their worse performance and lifetime.
The group at UCSD that authored this has done some nice work so I don't mean to be too negative, but they are trying to predict too far from a limited and faulty set of assumptions which unfortunately negates much of the validity of this paper.
jon
p.s. in the interests of full disclosure, I make the arrays in the first link :)
-jon
Actually, that's when I realized that the guy writing the article didn't have a clue. Since when is throughput measured in IOPS?
Since always. Throughput is always operations per second, or transactions per second. Bandiwdth is measred in Mbps or MBps.
Retention time in 2003-time-frame flash is tens of years. Retention time for the latest 25nm flash is measured at one year. Much less if you wear it out. Your 8MB SD card likely hasn't had the level of cycling needed to see reduced data life.
Prices haven't dropped in a couple years.
Prices are now down to about $1.50/GB for standard 2.5" SSDs. And you can sometimes find them for $1.25/GB. That's lower then the $2.50-$3.00 of 18-24 months ago.
Sure, it's expensive compared to the $0.10/GB of bulk storage like 1/2/4TB drives, but when you compare it to things like 10k RPM SATA/SAS and 15k SAS (about $1/GB) it starts to not look so expensive. The only things that make me nervous about them is that SSDs still have some controller issues and it's a younger technology compared to traditional hard drives.
At $1.50/GB, that means you can purchase a 120GB SSD for about $180. For a lot of people, that's big enough and cheap enough in exchange for vastly improved performance. And if you can keep the users from storing stuff locally, you could go with one of the 64/80GB units which are in the $100-$125 range.
I've converted a few users over to SSD over the past 2 years. It's been worth the money every time. The machines are far more responsive to user input, they don't sit there and spin, and it generally means that the CPU starts being the bottleneck again. Not all of these are power users, either.
I paid about $1.75/GB for my 250GB SSD. Do I wish it was bigger? Sometimes. But it turned a 4-year old laptop from something that I hated using due to the slowness of the old 500GB 5400 RPM hard drive into something that is fast and responsive. For work it made me much more productive.
Wolde you bothe eate your cake, and have your cake?
Related reading: http://www.tomshardware.com/reviews/ssd-reliability-failure-rate,2923.html
It features statistics from different data centers on the failure rate of SSD's.