density per kg or per liter is a meaningless metric for hard drives to improve. ssds are already far better in that. density per dollar is all hard drives have left and it hasnt really improved at all as far as I can tell in nearly 10 years. still 5x cheaper than ssd for cold storage but shrinking fast. will there be hard drives in 10 years ? my money is on NOOOOOO
Except SSD density per dollar is slowing down, isn't it. Performance per dollar is still improving fast but not the price. Flash node scaling stopped around 1x nm, so the industry moved to V-NAND. But with every new layer, it gets harder and harder to fab NAND. There will be a barrier to progress in how many layers of NAND can be stacked due to cost of fabrication at some time.
You can see this already in effect. The rate of improvement in the price per TB has been slowing down for years. Can SSDs go from 5x the price per TB of hard drives to say, 3x the price per TB? Probably. Will it keep narrowing the gap to 1.5x or lower the price per TB of hard drives? I really doubt it.
A new storage medium no doubt has the potential to overtake hard drives in cost per bit. I just don't think it will be NAND which is already reaching or has reached the end of several technological paths of progress in the near past and future.
"SSD density per dollar is slowing down" aprat from being incomprehensible its just rubbish nonsense.
ssd price per gigabyte is nearly halved in two years. u say its slowing down. even if it was its still improving. cost per tb for hdd not so much.
even u admitted they can get to 3x. at 3x hdd are dead. the more volume is lost for hdd, the more expensive they are because they can't amortize R&D.
at 3x price per GB. hard drives are down to below 50% their current volume which means probably means they are more expensive.
SSD is at 8 cents per GB now. they absolutely can and will get to 2 cents (which is lower than enterprise HDD cost)
at 4 cents per GB hard drives are dead and no1 will pay for new node. in fact this is already happening. the reason HAMR MAMR have taken half a decade is cuz no1 wants to pay to develop them. WD wants to re-invent itself as ssd maker
correction to this part: at 3x price per GB. hard drives are down to below 50% their current volume which means probably means they are more expensive.
at 3x price per GB. hard drives are down to below 50% their current volume which means price per GB has to rise to pay the companies fixed cost. meaning HDD will rise in price
>A new storage medium no doubt has the potential to overtake hard drives in cost per bit. I just don't think it will be NAND which is already reaching or has reached the end of several technological paths of progress in the near past and future.
Um? There is a very clear Roadmap with stacking, right now there is no reason why we cant do 1024 layers if not higher, the only issue is yield. There is also EUV NAND which is on the list of improvement, along with dozens of other technical improvement in the pipeline which will all improves Fab Capacity, Density and cost.
The TCO calculation also isn't just simply Cost / TB, you will need to factor in speed ( the amount of data increases while speed hasn't in HDD needs to more HDD require to balance I/O ) , power usage and space.
HDD wont die, but other than Cold storage, NAND is replacing HDD in more places than ever.
but they will, if they are only limited to cold storage and nothing else. do you think you can sustain the massive operations WD and Seagate have + plus many many part suppliers, because Amazon glacier is still interested ? not a chance.
ppl said optical won't die when thumb drive and wifi showed up, because "they still had some nice things about them"
And does it look like it's dead to you? Consoles? BluRay? Thumb drives have been around for ~2 decades already and while optical may not be a flourishing sector it's far from being dead.
Also when "ppl" say some technology will(not) die they are talking about a reasonably near future, not decades. Which makes "ppl" right. But if we're talking decades then everything you praise today will die too (eventually) so you too are wrong.
So whether "ppl" are right or you are wrong the conclusion is more or less the same. You're wrong. HDDs will be around for a long time, NAND will also die and be slowly (or quickly?) replaced at some point. What's your point?
"density per kg or per liter is a meaningless metric for hard drives to improve."
Except that it is tied directly to density per dollar. They aren't going to keep adding platters forever. Higher capacity hard drives lead to cheaper ($/TB) hard drives.
The resulting denser hard drives can be useful for customers, who can use less of them or get more capacity in a given rack.
right and nothing else is tied to per dollar right. For example no one cares about the R&D premium that got attached to the price to achieve that density/kg improvement.
If you can improve TB per dollar do it, if you can't stop pretending other metrics will matter at some unkonw point in the future.
If you can't improve TB per dollar and SSD do, the consumer have an easy choice to make. meanwhile in density per kg or per liter SSD are already miles ahead.
>density per dollar is all hard drives have left >will there be hard drives in 10 years ? my money is on NO
Flash's limited retention is a problem for long-term storage, maybe also medium-term.
We will find out more in the following years, as more and more people try to read data off their 1st or 2nd gen SSDs, after they were sitting in a drawer for a few years.
Pass on that stuff. I already have a proven storage solution. *eyes the stack of 100MB Zip disks and parallel port drive filled with porrrrrrrr...litical research material*
Biggest problem is that these always take forever to become available in reasonable numbers and with good service contracts. We install something like 3000 each year and HPE can never provide the latest sizes with the support level that we have. They just refuse to sell us the new models, because they can’t guarantee the quantity and replacement delivery schedules very early after starting production.
Call me crazy but why wouldn't these companies make a hard drives on a single platter in those capacities for the consumer market? It is my belief that it would help both life and performance of the drive to have less moving parts. 1 and 2 platters of 2TB each should cover 90% of the consumer market...
Largely because it wouldn't be cost competitive with current, mature technologies. These drives will be expensive per GB, and the platters only make up a very small portion of that price. The rest of it is in cost recovery for development, the mountain of reliability improvements that went into the platform, and likely the lower volume driving up fixed costs for them.
well... the long term storage solution is to hire 1,000 Chinese scribes to write your stuff down on papyrus. for your images, find a semi-trailer of Kodachrome (and a lab to process it).
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
41 Comments
Back to Article
MFinn3333 - Wednesday, September 18, 2019 - link
The 1st HDD was the IBM 350 which stores 5 million 6-bit characters (3.75 MB). It was leased at $3,200 a month in September 1957.quiksilvr - Wednesday, September 18, 2019 - link
September 1956. And that $3200 a month price included the IBM RAMAC 305 computer system as well.azfacea - Wednesday, September 18, 2019 - link
density per kg or per liter is a meaningless metric for hard drives to improve. ssds are already far better in that. density per dollar is all hard drives have left and it hasnt really improved at all as far as I can tell in nearly 10 years. still 5x cheaper than ssd for cold storage but shrinking fast. will there be hard drives in 10 years ? my money is on NOOOOOORevolution11 - Wednesday, September 18, 2019 - link
Except SSD density per dollar is slowing down, isn't it. Performance per dollar is still improving fast but not the price. Flash node scaling stopped around 1x nm, so the industry moved to V-NAND. But with every new layer, it gets harder and harder to fab NAND. There will be a barrier to progress in how many layers of NAND can be stacked due to cost of fabrication at some time.You can see this already in effect. The rate of improvement in the price per TB has been slowing down for years. Can SSDs go from 5x the price per TB of hard drives to say, 3x the price per TB? Probably. Will it keep narrowing the gap to 1.5x or lower the price per TB of hard drives? I really doubt it.
A new storage medium no doubt has the potential to overtake hard drives in cost per bit. I just don't think it will be NAND which is already reaching or has reached the end of several technological paths of progress in the near past and future.
azfacea - Thursday, September 19, 2019 - link
"SSD density per dollar is slowing down" aprat from being incomprehensible its just rubbish nonsense.ssd price per gigabyte is nearly halved in two years. u say its slowing down. even if it was its still improving. cost per tb for hdd not so much.
even u admitted they can get to 3x. at 3x hdd are dead. the more volume is lost for hdd, the more expensive they are because they can't amortize R&D.
at 3x price per GB. hard drives are down to below 50% their current volume which means probably means they are more expensive.
SSD is at 8 cents per GB now. they absolutely can and will get to 2 cents (which is lower than enterprise HDD cost)
at 4 cents per GB hard drives are dead and no1 will pay for new node. in fact this is already happening. the reason HAMR MAMR have taken half a decade is cuz no1 wants to pay to develop them. WD wants to re-invent itself as ssd maker
azfacea - Thursday, September 19, 2019 - link
correction to this part:at 3x price per GB. hard drives are down to below 50% their current volume which means probably means they are more expensive.
at 3x price per GB. hard drives are down to below 50% their current volume which means price per GB has to rise to pay the companies fixed cost. meaning HDD will rise in price
ksec - Thursday, September 19, 2019 - link
>A new storage medium no doubt has the potential to overtake hard drives in cost per bit. I just don't think it will be NAND which is already reaching or has reached the end of several technological paths of progress in the near past and future.Um? There is a very clear Roadmap with stacking, right now there is no reason why we cant do 1024 layers if not higher, the only issue is yield. There is also EUV NAND which is on the list of improvement, along with dozens of other technical improvement in the pipeline which will all improves Fab Capacity, Density and cost.
The TCO calculation also isn't just simply Cost / TB, you will need to factor in speed ( the amount of data increases while speed hasn't in HDD needs to more HDD require to balance I/O ) , power usage and space.
HDD wont die, but other than Cold storage, NAND is replacing HDD in more places than ever.
azfacea - Thursday, September 19, 2019 - link
"HDD wont die, but other than Cold storage, ...."but they will, if they are only limited to cold storage and nothing else. do you think you can sustain the massive operations WD and Seagate have + plus many many part suppliers, because Amazon glacier is still interested ? not a chance.
ppl said optical won't die when thumb drive and wifi showed up, because "they still had some nice things about them"
close - Thursday, September 19, 2019 - link
@azfacea "ppl said optical won't die"And does it look like it's dead to you? Consoles? BluRay? Thumb drives have been around for ~2 decades already and while optical may not be a flourishing sector it's far from being dead.
Also when "ppl" say some technology will(not) die they are talking about a reasonably near future, not decades. Which makes "ppl" right. But if we're talking decades then everything you praise today will die too (eventually) so you too are wrong.
So whether "ppl" are right or you are wrong the conclusion is more or less the same. You're wrong. HDDs will be around for a long time, NAND will also die and be slowly (or quickly?) replaced at some point. What's your point?
ikjadoon - Monday, September 23, 2019 - link
Consoles ARE moving away from discs. And Blu-ray players have lost mainstream appeal.It'll be slow, but I think optical will have lost any remaining marketshare in less than ten years.
nandnandnand - Thursday, September 19, 2019 - link
"density per kg or per liter is a meaningless metric for hard drives to improve."Except that it is tied directly to density per dollar. They aren't going to keep adding platters forever. Higher capacity hard drives lead to cheaper ($/TB) hard drives.
The resulting denser hard drives can be useful for customers, who can use less of them or get more capacity in a given rack.
azfacea - Thursday, September 19, 2019 - link
" tied directly to density per dollar "right and nothing else is tied to per dollar right. For example no one cares about the R&D premium that got attached to the price to achieve that density/kg improvement.
If you can improve TB per dollar do it, if you can't stop pretending other metrics will matter at some unkonw point in the future.
If you can't improve TB per dollar and SSD do, the consumer have an easy choice to make. meanwhile in density per kg or per liter SSD are already miles ahead.
sheh - Thursday, September 19, 2019 - link
>density per dollar is all hard drives have left>will there be hard drives in 10 years ? my money is on NO
Flash's limited retention is a problem for long-term storage, maybe also medium-term.
We will find out more in the following years, as more and more people try to read data off their 1st or 2nd gen SSDs, after they were sitting in a drawer for a few years.
PeachNCream - Wednesday, September 18, 2019 - link
Pass on that stuff. I already have a proven storage solution. *eyes the stack of 100MB Zip disks and parallel port drive filled with porrrrrrrr...litical research material*ballsystemlord - Wednesday, September 18, 2019 - link
Why porrrrrrrr...litical research material of all things?Ranger90125 - Thursday, September 19, 2019 - link
This comment is so out of character....and there I was thinking you were a stand up guy :)PeachNCream - Thursday, September 19, 2019 - link
Don't tell anyone, but I sit down a lot too.Lord of the Bored - Thursday, September 19, 2019 - link
That's disgusting!Zip disks, eww...
MDD1963 - Wednesday, September 18, 2019 - link
can't wait to grab 6-8 of these in a RAID 0! :) Endless fast storage!!!edzieba - Thursday, September 19, 2019 - link
Let's just go retro and call it magneto-optical.nandnandnand - Thursday, September 19, 2019 - link
Still waiting for 5D holographic superman crystals.zepi - Thursday, September 19, 2019 - link
Biggest problem is that these always take forever to become available in reasonable numbers and with good service contracts. We install something like 3000 each year and HPE can never provide the latest sizes with the support level that we have. They just refuse to sell us the new models, because they can’t guarantee the quantity and replacement delivery schedules very early after starting production.ads295 - Thursday, September 19, 2019 - link
Call me crazy but why wouldn't these companies make a hard drives on a single platter in those capacities for the consumer market? It is my belief that it would help both life and performance of the drive to have less moving parts.1 and 2 platters of 2TB each should cover 90% of the consumer market...
lightningz71 - Thursday, September 19, 2019 - link
Largely because it wouldn't be cost competitive with current, mature technologies. These drives will be expensive per GB, and the platters only make up a very small portion of that price. The rest of it is in cost recovery for development, the mountain of reliability improvements that went into the platform, and likely the lower volume driving up fixed costs for them.FunBunny2 - Thursday, September 19, 2019 - link
well... the long term storage solution is to hire 1,000 Chinese scribes to write your stuff down on papyrus. for your images, find a semi-trailer of Kodachrome (and a lab to process it).ballsystemlord - Thursday, September 19, 2019 - link
I'll also need a papyrus tree farm.......