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.
You heard me.
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.
This has been a pain heading down the pipeline for cellphone designers for a couple of years now, and its not like you can still buy stuff built on the old processes either unless you ordered it a couple of years ago and I only know one cellphone company confident enough to pre-order a few years of storage.
Storage is getting larger, whoopee. It's also becoming massively slower and less reliable. Two attributes you can't really make up for in volume on a phone. Even so-called 'high end' SLC eMMC is miles slower than the raw NAND we used to be able to get. You get 10x as much now, but you can only use it 1/20 as long.
Blah blah, with years of revenue, disks will become cheaper because the production lines, investment risks and other burdens of innovation implementation will be paid for. Same thing happened with LCD production, and we get better LCDs now than we used to have.
"16TB ought to be enough for anybody."
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.
as reported here and here? I thought people have been busy about it for quite some time.
There is a spark in every single flame bait point.
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
Surely somebody will come up with a new way to store data in the next 12 years. Just a hunch, but I'm thinking that somebody is going to dust off some old research or idea and we are all going to be rushing out to replace our flash drives for something faster long before we start hitting this limit. This sounds like some researcher just threw down the gauntlet. Innovation will prove that his assumptions where incorrect because their conclusion looks like ones I've heard in the past.
Anybody recall the "640kB out to be enough for anybody" quote?
One thing does stand out though, 2 TB limits on NAND flash drives? Really? 16TB on TLC? Given standard drive sizes seem to be multiplying by 2 about every year, there will be serious issues even before the 2024 date.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
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?
I only ask because Apple is the largest flash customer/reseller in the world and they just bought this company
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
Along with error rates, what will happen to retention times as the cell size shrinks?
Supposedly, flash memories have expected retention times as short as 5-10 years or so (if not refreshed by re-writing), thanks to gradual leakage of the trapped charges they use to record data; this value is expected to drop as flash cells get smaller. I've had gadgets whose firmware mysteriously become corrupted after sitting around for a few years, and sometimes they could be revived by re-flashing them -- I sometimes wonder if this kind of retention problem could have been responsible.
I've wondered what spinning disk could do in a vacuum chamber and with a non-contact magnetic bearing.
Rod Taylor
Because, maybe, one of those children will innovate, and solve this problem?
now we need to go OSS in diesel cars
I will be happy (but not satisfied) when just a mere 1TB SSD does not cost the same as a house or a (decent) car.
A Pirate and a Puritan look the same on a balance sheet.
... microelectronics fabrication, making them vulnerable to inductive effects?
now we need to go OSS in diesel cars
I was kind of hoping we'd have something better than NAND Flash within 5 or so years. Maybe something using memristors? NAND is just too expensive to be useful. Prices haven't dropped in a couple years.
Moore's law has been extended to the point where it is a sociological or economic observation, not a physical one.
the preceding comment is my own and in no way reflects the opinion of the Joint Chiefs of Staff
Actually, that's when I realized that the guy writing the article didn't have a clue. Since when is throughput measured in IOPS?
Type one: bandwidth sensitive. OS files, application files, cached application data, etc.
Type two: bandwidth insensitive, e.g. streamed. E.g. Video, audio.
Store type one on a SSD. Store type two on a higher-capacity magnetic drive. How likely is type one data expected to grow? Perhaps not that fast. My home machine is an ~8 year old Dell laptop with a 60G disk. It's not even half full. That includes an OS, browser, Office suite, and a few other applications.
There's been tons of research on alternative technologies, including phase-change memory (http://en.wikipedia.org/wiki/Phase-change_memory) and magnetic tunneling junction (http://drl.ee.ucla.edu/index.php?page=research&function=sttram) memory. Obviously commercializing them is expensive, but some progress has already been made there. I'm sure other competing technologies will be developed in that time as well.
only use the ssd partitions where it makes sense. no reason to install the whole OS to the ssd. just mount /tmp or /var/lib/mysql on an ssd slice. use plain old sata raid for the rest.
Join the Slashcott! Feb 10 thru Feb 17!
You still have the limits of the disk to deal with. That's why optical media like DVDs and CDs aren't getting any faster. The disks are already spinning as fast as they can.
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
latency will finally be a worry of the past
To be replaced by fear of decapitation, no doubt.
I dont think microsoft had much to do with it, people got tired of using a 2 gig pendrive as their mass storage device and couldnt run any of their programs with a half a distro and a celphone UI
The centripetal forces would require an amazing materials engineering advance.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
Yeah at almost .1c for the edge of a 3.5" drive I imagine the relativistic effects would cause some problems as well
it's RELIABILITY issues they need to fix first. I just sent my first SSD back for RMA after only 7 months usage. An OCZ Vertex 3 MI and it was junk from day one. But just started tossing smart errors like mad. Bought a Vertex3 non MI and it's been rock solid, now I just have to wait for the other one to get back so I can re-raid them and then be happy. Another thing that sucks about SSD's are write speeds. They go from awesome on a bare drive to crap once the drive starts filling up. When I get the MI and benched it, it shows the proper 500+MB/s read/writes. When I sent it back it was showing 450+ Read and 200+ writes :( And I could feel it.
This one is showing 500+ reads and about 350 writes alone wich isn't bad. When I raided the two before sending it back, I was getting about 1000+ reads and 350+ writes. The reads are pretty phenomenal but the writes are pretty meh. Hopefully when I get the replacement drive back next week [or the next :(] I'll see awesome writes again. I asked them to send me back a non-mi drive in return and they agreeed. This way I have 2 matching drives with matching firmare and matching everything :) But if last weeks test is anything to go by. I am REALLY looking forward to getting this drive back :)
I'd file this under "no shit sherlock" but its a grad student getting their conclusion without clearly be trained by a knowledgeable industry insider. Then again, maybe they were just doing research that proved what they already knew. A lot of people know this already and actively implement features to work around these problems. I'd disclose more but um lets just say I can't... :)
That said I'm not aware of any SSDs using more than 2 bits per cell. My understanding is that it's not practical... yet. I see OCZ claims they're going to do it, but its a lot easier to say you're going to do it than to actually do it and provide a reliable drive that isn't going to get RMA'ed in droves. You have to consider the development cost is going to be higher as you pull your hair out trying to prevent customer data loss. Beating the hell out of NAND takes a lot of time.
I agree to this 150% and more. I was not sure if I want a SSD in my new iMac, but I went for it. Anyone who hasn't use a SSD yet, does not know how blazing fast this is. Booting in 2s, start Photoshop in 2s and meanwhile restore Chrome with 10 windows and 50 tabs and what not else. This is just amazing.
Every time I have to reboot my work Mac I really want to ask for a SSD. So slow ... so horrible slow.
"Freiheit ist immer auch die Freiheit des Andersdenkenden" - Rosa Luxemburg, 1871 - 1919
But I have to agree that this is a quite rare exception...
I beat on it all day for SEO, backlink analysis, video editing
Two guesses what business AC is in :-)
Be careful. People in masks cannot be trusted.
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.
Biggest laptop manufacturer using NAND = Apple
Report claiming NAND unreliable commissioned by = Microsoft.
Hmmm....
"We live in a global world" - Harvey Pitt, former Securities and Exchange Commission Chairman
I would have modded you funny but i've already posted. so instead:
A: porn
There is no meaningful "parity writing overhead" with SSDs, because their IOPS are so high.
Based on what todays spinning disk form factors? If SDDs become standard we can rethink the formfactor for the harddrive. My understanding among other reasons ~3.5" disks one out because they were selling in higher volumes and advanced quicker than the huge mainframe disks of the day. They then became the standard size and everything has been basing itself out of that. But there is no reason why SDD couldn't be a different form factor if you were willing to pay for it. For example: how about a disk that is the size of your motherboard? You open up your case from the side and there it is. It could be on a hinge so you just swing it out of the way whenever you need to work on something else but on a full sized desktop that would give you about 10 disks worth of space to play with. Or you could have a data "cube" next to your box, and the box would probably be getting correspondingly smaller what with smaller formfactor computers already out, more integration into the CPU, etc.
This comment is more true than you realize. I recently downloaded all the works of Isaac Asimov as a 10MB PDF file. It really puts things into a perspective when a life's work of one of the most prolific authors can be summed up in so little disk space. So I can safely say no one will ever need more than 10 MB.
However, people will want more than 4TB. And more importantly, they will pay for it.
PlusFive Slashdot reader for Android. Can post comments.
2047 or 2048 - 2^11 - would be a good year to mark the end of the road.
Never seen an industry so incompetent in all my life. SSD was introduced decades ago and still today we have to suffer low capacity and high prices and lack luster performance.
Compare this to the HDD industry which keeps surprising people with much higher capacities, performance, and better cost per/byte ratios then anyone could imagine 10 years ago.
Compare that to the CPU industry which is squeezing out obscene levels of processing power per watt efficiency and shrinking their die sizes down to near single atom structure.
I don't understand why they can't slap a bunch of cheap non-volatile memory chips in a massive RAID like stripe set in a circuit board and deliver it for pennies on the GB with crazy performance? Is that difficult? Is there some fundamental roadblock that prevents this kind of simplicity in design?
Maybe they got to move away from NAND and start thinking about something new in solid state storage, obviously NAND seems to be a dead end. Also why not fire every single engineer working on SSD. Obviously they are, as a collective, dumber then sand and stuck in a rut.
I haven't thought of anything clever to put here, but then again most of you haven't either.
True. But even multilayer means a significant hit on equipment costs as new equipment is needed, and that equipment stays busy longer building a single device up in layers.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
I get ~457917 meters per second on the edge, so more like ~0.0015 c.
But still too much to not cause vehicular manslaughter on container fail.
I have nothing to lose but my bindings.