Report: SSDs will run into performance issues in future

Solid state drives are supposed to be so much better than conventional and mechanical hard drives for PCs. SSDs are supposed to be faster, less prone to failures and use less power. So far, the only real issue with SSDs has been their much higher cost compared to regular hard drives.

Now a group of researchers have posted word that the future of SSDs could be shorter than first believed. Computerworld.com reports that a new research paper claims that as the circuitry of NAND flash-based drives keeps going down, more and more performance issues come up, including latency and read and write errors.

The group, which includes a team member from Microsoft Research, tested 52 such drives from six different companies, with circuity ranging from 72nm to 25nm. The results show that latency in those drives increased as the circuity got smaller.

The research report then predicted that SSDs would have 6.5nm circuity by 2024. By then, the latency issues will have doubled and bit error rates would have tripled for those drives. The report gives the opinion that "it's not going to be viable to go past 6.5nm ... 2024 is the end."

The good news? SSDs are expected to have storage capacities as high as 16 TB by 2024. By that time, researchers believe that we will have to pick between "capacity or performance" for those kinds of storage drives.

Report a problem with article
Previous Story

Inside of Foxconn Apple plant revealed in new video

Next Story

Microsoft extends Windows 7 support to 2020

47 Comments

Commenting is disabled on this article.

Haha. I have a 80GB hard disk from 2005 and it's still working as if it was new. As long as I use XP, I won't have any space issues, so if my hard disk survives another seven years, it's a beast.

I guess were at the point where everyone knows exactly what nm means and that doesn't need to be explained anywhere in the article. Now that you've all out geeked me, someone want to explain? I'm too lazy to Google it on Bing.

jimmyfal said,
I guess were at the point where everyone knows exactly what nm means and that doesn't need to be explained anywhere in the article. Now that you've all out geeked me, someone want to explain? I'm too lazy to Google it on Bing.

It means....N-ever M-ind.

jimmyfal said,
I guess were at the point where everyone knows exactly what nm means and that doesn't need to be explained anywhere in the article. Now that you've all out geeked me, someone want to explain? I'm too lazy to Google it on Bing.

nm = nanometer = 1/1,000,000,000 meters

When talking about transistors, circuitry, etc. the number is used to describe the "minimum feature width" (roughly something like how small you can make a circuit element in the chip).

jimmyfal said,
I guess were at the point where everyone knows exactly what nm means and that doesn't need to be explained anywhere in the article. Now that you've all out geeked me, someone want to explain? I'm too lazy to Google it on Bing.

Wait, you run a computer assistance program and you don't know what "nm" stands for, you used "Google it on Bing", and you claim to be "geek from the start".

Holy **** you fail, and it is people like YOU that have ruined the computer professional economy of the 90's. Yes, everyone is a computer geek, except people that are too dumb and too lazy, that means you JimmyFail.

I don't know why they just don't embed the SSD chips right on the motherboard. Would eliminate the need for anything faster...and you can still put another hard drive in an open bay.

texasghost said,
I don't know why they just don't embed the SSD chips right on the motherboard. Would eliminate the need for anything faster...and you can still put another hard drive in an open bay.

.... what?

texasghost said,
I don't know why they just don't embed the SSD chips right on the motherboard. Would eliminate the need for anything faster...and you can still put another hard drive in an open bay.

Let's see:

-Adds costs to motherboards
-Miniature capacity in order to be of ANY real world use
-Fragmentation of your storage (quite a few people prefer the two-partition system: a) OS + Applications b) data ( c) backup))

I personally couldn't stand having this miniature partition I'd probably never use for anything real other then my OS that would be better put on a normal SSD that's probably higher quality, capacity and exchangeable.

GS:mac

I've been using a Kingston 128GB SSD-Now for the past year and I even defragment it despite the warnings that SSD's don't need to be defragmented. I find the device still performs better when its fully defragmented and I've never noticed any loss of performance. In fact, the performance of my SSD is way better compared to my regular drives. As a system disk it boots up faster and gets to my desktop faster. Apps launch faster and to defragment is faster.
On a side note, I used the same SSD drive to test out the Windows 8 developer preview and boot up times were <10 seconds like they said.

netsendjoe said,
I've been using a Kingston 128GB SSD-Now for the past year and I even defragment it despite the warnings that SSD's don't need to be defragmented. I find the device still performs better when its fully defragmented and I've never noticed any loss of performance. In fact, the performance of my SSD is way better compared to my regular drives. As a system disk it boots up faster and gets to my desktop faster. Apps launch faster and to defragment is faster.
On a side note, I used the same SSD drive to test out the Windows 8 developer preview and boot up times were <10 seconds like they said.

is that from a cold boot or their super hybrid sleep BS boot up times

the reason they say DON'T use defrag on an SSD is for a good reason by defraging your SSD your reusing the same nand cells over and over again shortening their life span considerably ( remember each cell only has a limited number of write cycles before it becomes inoperable) just let the SSD's drive wearing software do it's thing

Theory would suggest mirrored SSD's would fail about the same time, so perhaps swapping one out with anew and let it rebuild each year would increase reliabilty for server purposes. Sounds like SSD data recovery might be impossible.

Never going back, at least not for my OS drive. My x2 128GB is a year and half old and is at 99%. As for storage, I need TB and have to HDD. The landscape will be significantly different in a few years much less a dozen. Look back 1 dozen years and see where we were.

Raa said,
That's why I don't select SSD. I'll stick with HDDs for now!

obviously you have never experience ssds.

virtorio said,
Damn, I thought 80 GB would be enough to late me until at least 2050

Well, for you 80GB might be more than enough. Just don't use a computer for anything other than email and you should be in great shape!

I remember way back when as analysts were saying that Intel and their Pentium chips weren't going to get any smaller or faster. New tech comes along, advances keep happening, the world keeps turning.

error404ts said,
I remember way back when as analysts were saying that Intel and their Pentium chips weren't going to get any smaller or faster. New tech comes along, advances keep happening, the world keeps turning.

There is a difference between the real researchers and "analysts"+"journalists".

Pink Floyd said,
wow 12 years! I bet SSD will be replaced by something else before 2024!

Because HDD's only lasted like 5 years, right ?

Errrr I thought everybody already knew this? The companies that make these drives are already experimenting with newer technology.

TheAleas said,
it almost sounds like there is no choice but to go smaller, is anyone forcing them?

Moore's Law... yeah, this concept some dead guy once said has somehow become this magical law upon which all innovations in storage and circuitry must follow. Or death ensues.

Sadly, self-serving businessmen with no engineering background keep reasserting and blackmailing the researchers over the dire consequences of failing to follow through on Moore's Law. Like the world even cares.

As long as prices for these things are as high as they are I can't justify getting one with the price per gig as it stands.

"as the circuitry of NAND flash-based drives keeps going down"

what does that even mean? what exactly is going down? it's size? it's cost? ponies?

quintesse said,
"as the circuitry of NAND flash-based drives keeps going down"

what does that even mean? what exactly is going down? it's size? it's cost? ponies?


Yes the ponies will go down, It will be their downfall.

quintesse said,
"as the circuitry of NAND flash-based drives keeps going down"

what does that even mean? what exactly is going down? it's size? it's cost? ponies?

Somewhat poorly phrased, but I believe they were referring to the lithography. The fabrication process. "Circuitry". I know. I had to reread it a couple times to get it, myself.

quintesse said,
"as the circuitry of NAND flash-based drives keeps going down"

what does that even mean? what exactly is going down? it's size? it's cost? ponies?


Before I really read any article. I look up the author and find the sources they are using. unfortunately, more often than not, the authors are not qualified to be reporting on the subject matter.

This is assuming that nothing changes between now and 2024... Come on...

I'm sure that as the technology fully matures, these issues will be all but eliminated.

Xenosion said,
This is assuming that nothing changes between now and 2024... Come on...

I'm sure that as the technology fully matures, these issues will be all but eliminated.


Agreed. I see SSD technology as still relatively undeveloped; I think once they make more advancements on it regular HDD will be all but obsolete. Never know...

Xenosion said,
This is assuming that nothing changes between now and 2024... Come on...

I'm sure that as the technology fully matures, these issues will be all but eliminated.

well you can say that but if your limited my matter and the laws of physics you cant get around them

silicon will end at something like 11-16nm you cant make chips smaller than that

DrakeN2k said,

well you can say that but if your limited my matter and the laws of physics you cant get around them

silicon will end at something like 11-16nm you cant make chips smaller than that


There are already replacement technologies for silicon being refined. Silicon might be limited but are you honestly suggesting that progress will end with it?

este said,

Agreed. I see SSD technology as still relatively undeveloped; I think once they make more advancements on it regular HDD will be all but obsolete. Never know...


wonder if they've checked the ones that plug into a PCI-E slot on a desktop motherboard. those are always awesome to look at. they like look like actual SSD's compared to the others that wanna look like Hard Drives.. Drives are never meant to last. I hope by that time I can have a 16 tb blu ray or another format by then or just something in the cloud provided the RIAA doesnt kill it this year.

DrakeN2k said,

well you can say that but if your limited my matter and the laws of physics you cant get around them

silicon will end at something like 11-16nm you cant make chips smaller than that

And I'm sure the physicists were saying that at CERN when they discovered something that traveled faster than light.

Amodin said,

And I'm sure the physicists were saying that at CERN when they discovered something that traveled faster than light.

Inpussibull !

Xenosion said,
This is assuming that nothing changes between now and 2024... Come on...

I'm sure that as the technology fully matures, these issues will be all but eliminated.

+1.
Same thing happened with the processors.
And what they did is adding.
By then, we will have 2 x SSD running in RAID-9.

DrakeN2k said,

well you can say that but if your limited my matter and the laws of physics you cant get around them

silicon will end at something like 11-16nm you cant make chips smaller than that

It's not like disc based HDD are not limited by laws of physics either. I'm not a pro in the subject but my guess is there's a limit to the amont of data you can put in a disc based 3.5" HDD too.

When we will hit the limit we will do like CPU. With CPU we added core. With SSD we will just add more drive. SSD don't make any noise and don't produce much heat. At 2.5" (size of most of them) you can squeeze lot of them in a machine. By 2024 i guess the price will be down

Xenosion said,

There are already replacement technologies for silicon being refined. Silicon might be limited but are you honestly suggesting that progress will end with it?

At 6.5nm, things are so small that even getting power across the atoms without them bridging to anything else is impossible with any known conductive material we know about today.

When you think that a single atom of copper is 0.125nm and you need at least 7nm for conductivity to be at only 50% and thats without leaving space for layers, you can see we're already very close to the actual physics of how small we can go.

Currently, due to the conductivity (at 20% loss) we're already having with 25nm it still makes sense to have things smaller as we get a return in lower power needed and still make a gain in speeds etc.

However, if we bring it down to around 7nm with current technology we it'll simply not be worth the cost of production, hence why the research suggests that
"it's not going to be viable to go past 6.5nm ... 2024 is the end". AND that is including the use of nanotubes.

I'll be quite happy if we're stuck with 15nm, I'm sure someone will figure out a better way to make things work as storage when they have to. Else, I still wouldn't mind having 3.5" SSD with say 40TB on board.... and if the powers that be do push more for the cloud option, we wont be using too much of our own storage........

Xenosion said,
This is assuming that nothing changes between now and 2024... Come on...

I'm sure that as the technology fully matures, these issues will be all but eliminated.

Plus, they said the same things about many older HDD types at one time.

sagum said,

At 6.5nm, things are so small that even getting power across the atoms without them bridging to anything else is impossible with any known conductive material we know about today.

When you think that a single atom of copper is 0.125nm and you need at least 7nm for conductivity to be at only 50% and thats without leaving space for layers, you can see we're already very close to the actual physics of how small we can go.

Currently, due to the conductivity (at 20% loss) we're already having with 25nm it still makes sense to have things smaller as we get a return in lower power needed and still make a gain in speeds etc.

However, if we bring it down to around 7nm with current technology we it'll simply not be worth the cost of production, hence why the research suggests that
"it's not going to be viable to go past 6.5nm ... 2024 is the end". AND that is including the use of nanotubes.

I'll be quite happy if we're stuck with 15nm, I'm sure someone will figure out a better way to make things work as storage when they have to. Else, I still wouldn't mind having 3.5" SSD with say 40TB on board.... and if the powers that be do push more for the cloud option, we wont be using too much of our own storage........

This is still only considering the immediate concepts we have on creating storage media. There are dramatically different concepts such as holographic data and I believe when we learn more about quantum mechanics, that will be yet another massive leap forward in how we store data. Your examples are still in a typical smaller, better, faster, stronger (Daft Punk reference ) approach when what we need is another breakthrough. By 2024 there will be something replacing this and we will all look back at this discussion and laugh at how we all keep doubting the industry.