Seagate Wants to Ship 100TB HDDs by 2025

[Guest]

Seagate plans to use its Heat Assisted Magnetic Recording (HAMR) technology, along with bit-patterned media, to offer 100TB hard drives by 2025.

Seagate Wants to Ship 100TB HDDs by 2025 : Read more

 

[jaber2]

And yet I can only have 100MB on my outlook at work

 

[jimmysmitty]

I don't even want to imagine how long it would take to clone a 100TB drive on SATA 3.

 

[bloodroses]

I don't even want to imagine how long it would take to clone a 100TB drive on SATA 3.

I know formatting 2 8TB hard drives in raid 1 took 3 days on my NAS. I shudder to think the amount of time for 100TB...

 

[billymainor3]

of course right now the barracuda st1000dm001/st2000dm001 hard drivers are having a massive failling by the firmware :/

 

[stdragon]

Seagate plans to use its Heat Assisted Magnetic Recording (HAMR) technology, along with bit-patterned media, to offer 100TB hard drives by 2025.

Seagate Wants to Ship 100TB HDDs by 2025 : Read more

Assuming a full SATA3 link saturation @ 600 MBp/s (read from source, write to destination), it would take 37 hours!

 

[elbert]

My brother has a 10 year old WD 750GB Black HD he paid $79 for that still works. He would like to upgrade to higher capacity but the 1TB Black is $79 now so he see little advantage over his decade old drive. If HD makers dont actually start trying to sale hard drives to consumers they may have no market in 2025. After 2012 they just started trying to up the price for higher capacity. Before 2012 highest prices were $200~$300 and now they are $400~$500. Point is if they dont wake up there only customers will be enterprise. Everyone that has wanted a 2, 3, or 4TB drive has had a decade to buy one and capacity is kind of CPU performance when it comes to upgrading at a given price point.

 

[jimmysmitty]

My brother has a 10 year old WD 750GB Black HD he paid $79 for that still works. He would like to upgrade to higher capacity but the 1TB Black is $79 now so he see little advantage over his decade old drive. If HD makers dont actually start trying to sale hard drives to consumers they may have no market in 2025. After 2012 they just started trying to up the price for higher capacity. Before 2012 highest prices were $200~$300 and now they are $400~$500. Point is if they dont wake up there only customers will be enterprise. Everyone that has wanted a 2, 3, or 4TB drive has had a decade to buy one and capacity is kind of CPU performance when it comes to upgrading at a given price point.

In 10 years though there have been performance upgrades. However one thing I would say is that SSDs is the only avenue I can see 100TB from thats viable.

 

[elbert]

My brother has a 10 year old WD 750GB Black HD he paid $79 for that still works. He would like to upgrade to higher capacity but the 1TB Black is $79 now so he see little advantage over his decade old drive. If HD makers dont actually start trying to sale hard drives to consumers they may have no market in 2025. After 2012 they just started trying to up the price for higher capacity. Before 2012 highest prices were $200~$300 and now they are $400~$500. Point is if they dont wake up there only customers will be enterprise. Everyone that has wanted a 2, 3, or 4TB drive has had a decade to buy one and capacity is kind of CPU performance when it comes to upgrading at a given price point.

In 10 years though there have been performance upgrades. However one thing I would say is that SSDs is the only avenue I can see 100TB from thats viable.

Performance means nothing give the SSD. The hard drives capacity and reliability are what most matters to consumers. I would get a small SSD and the most reliable large capacity 6TB+ 5400RPM hard drive. Also I dont see how they are making a performance upgrade given the last 10000RPM drive design was 2012. Sure they add a bit of cache to call it a new design to show small improvements for data centers efficiency. Tell you the truth tho HD makers better watch out for thumb drives more than SSD's. When those hit affordable TB capacity's HD's are done.

 

[bit_user]

Assuming a full SATA3 link saturation @ 600 MBp/s (read from source, write to destination), it would take 37 hours!

I get 46.3 hours.

More likely, these would use at least 12 Gbps SAS or even NVMe.

 

[bit_user]

My brother has a 10 year old WD 750GB Black HD he paid $79 for that still works. He would like to upgrade to higher capacity but the 1TB Black is $79 now so he see little advantage over his decade old drive.

Two problems with this picture. First, HDDs have a price floor. There's a minimum amount it costs just for the materials and assembly. This price floor shifts with time, probably not decreasing much, if any, as HDD technology and manufacturing is incredibly mature. To get the benefits of density improvement (i.e. more GB/$), he needs to spend more and get into the mid-range market segment.

The second issue is probably somewhat lack of competition in the consumer HDD market. Probably the healthiest consumer HDD market segment is for NAS HDDs. That's probably where you'll see the best improvement in GB/$, over time.

 

[bit_user]

In 10 years though there have been performance upgrades. However one thing I would say is that SSDs is the only avenue I can see 100TB from thats viable.

In what sense do you even mean?

People forget that NAND flash is a charge storage device. Try filling a QLC SSD and letting it sit on the shelf for several years. Think you'll still get your data back, after that? Guess again.

This also applies to your USB thumb drives, SD cards, etc. They're all getting less reliable, as density increases.

But HDDs, magnetic tape, and optical are a different story.

 

[stdragon]

People forget that NAND flash is a charge storage device. Try filling a QLC SSD and letting it sit on the shelf for several years. Think you'll still get your data back, after that? Guess again.

This also applies to your USB thumb drives, SD cards, etc. They're all getting less reliable, as density increases.

But HDDs, magnetic tape, and optical are a different story.

Bit-rot, digital dark-age...etc.

It's quite frightening when you think about it. So the average consumer SSD could hold data between 1 to 2 years if unplugged and shelved at room temperature. Magnetic media degrades (and I hate tape, lots of bad experience with DLT/LTO media). And standard CDR media oxidizes within a span of 10 years and incurs too many CRC errors to recover from.

If anyone is looking to leave data for future generations in storage, at least look at M-DISK which is optical media that has a life span of a thousand years. Though I reckon even in 40 years finding a drive to read them might be a little hard to come by. In fact, I bet the drive itself degrades in storage way before the media. Store it in pure nitrogen at 1 atmosphere pressure??

Anyways, it's all good until the music stops. Hopefully in the future the best data recall is in paper form via some future historian. Then again, if having to resort back to books for lost knowledge, I suspect we didn't have a very good future to begin with. But I digress.

 

[none12345]

Dont need 100TB yet....but with games now taking about 100GB a piece, it wont be all that long until you see games at the 1TB level. I would not be suprised to see games at 1TB by 2025. At that point a 20TB harddrive wont feel very large. 100TB might be excessive, but 20 will certainly be needed, maybe 50.

Hell i had a game update today that took 90GB of hard drive space to update the game. And thats on top of the space the game was already using before the update.

--------------

Interesting watching hard drive space grow. My first family computer with a hard drive had a 50 meg hard drive. I remember longing for 1GB in that era, never having enough space. Eventually i had hundreds of megabytes, still too small, then a gigabyte, then multiple gigabytes, 10s of gigabytes, hundreds of gigabytes, finally a terabyte, then 3TB, and now im at 5TB and running out of space again. Im currently a factor of 100,000 compared to that first hard drive, and still running out of space.

I wonder if ill live long enough to see exabyte scale consumer drives. I'll almost certainly live to see petabytes in home computers.

 

[USAFRet]

"For consumers, we still think the big hard drives are mainly for niche applications,"
Jan 4, 2007, talking about the first 1TB HDD.
https://www.pcworld.com/article/128400/article.html

 

[elbert]

In 10 years though there have been performance upgrades. However one thing I would say is that SSDs is the only avenue I can see 100TB from thats viable.

In what sense do you even mean?

People forget that NAND flash is a charge storage device. Try filling a QLC SSD and letting it sit on the shelf for several years. Think you'll still get your data back, after that? Guess again.

This also applies to your USB thumb drives, SD cards, etc. They're all getting less reliable, as density increases.

But HDDs, magnetic tape, and optical are a different story.

Actually it fades for HDDs and the others as they all lose data over time if not rewritten. It may last a bit longer on HDDs but 7 years your going to see some lose as magnetic charge fades. Not only that but you also have mechanical failure on HDDs. Even a user burned CD degrades after about 5 years were only the pressed manufactured versions last very long if they remain unscratched. I have very old Quantex Pentium 60 with the math error that worked last time I used it but doesn't boot now due to lose of data. I would have to reinstall an OS and I checked its out with a live boot cd to find a bit a data still on the drive but much is corrupt.

 

[bit_user]

In 10 years though there have been performance upgrades. However one thing I would say is that SSDs is the only avenue I can see 100TB from thats viable.

In what sense do you even mean?

People forget that NAND flash is a charge storage device. Try filling a QLC SSD and letting it sit on the shelf for several years. Think you'll still get your data back, after that? Guess again.

This also applies to your USB thumb drives, SD cards, etc. They're all getting less reliable, as density increases.

But HDDs, magnetic tape, and optical are a different story.

Actually it fades for HDDs and the others as they all lose data over time if not rewritten. It may last a bit longer on HDDs but 7 years your going to see some lose as magnetic charge fades.

I certainly didn't mean to imply they're immune from degradation. Just not as fast as flash is becoming.

Not only that but you also have mechanical failure on HDDs.

This is why anyone seriously using HDDs employs RAID (or duplication). But, even with RAID, the underlying media still needs to have adequate longevity. If you had a RAID of SSDs sitting unplugged, they'll all degrade at about the same rate - RAID won't really help with the longevity problem.

 

[cryoburner]

My brother has a 10 year old WD 750GB Black HD he paid $79 for that still works. He would like to upgrade to higher capacity but the 1TB Black is $79 now so he see little advantage over his decade old drive...
...Also I dont see how they are making a performance upgrade given the last 10000RPM drive design was 2012. Sure they add a bit of cache to call it a new design to show small improvements for data centers efficiency.

As the platter density increases, more data can fit in each track, so with each rotation of the platter, the drive can read more data. Your brother's 10-year old WD Black drive likely only holds 250 GB on each of three platters, whereas a modern drive could hold a TB or more per platter. A 7200 RPM desktop drive from 10 years ago won't even be able to hit 100 MB/s sequential read/write speed at the fastest part of the drive, dropping to under 50 MB/s near the end of the drive. A current generation drive, on the other hand, should be able to get around 200 MB/s at the start of the drive for sequential transfers, dropping to around 100 MB/s toward the end. So for large file accesses, the current 7200 RPM drives can be up to twice as fast. That is for sequential transfers though, since increases in platter density won't do much for random access times.

Of course, that's what SSDs are for, and why 10000 RPM drives are no longer a thing. 10000 RPM drives improved random access times a bit, but not in any way comparable to an SSD, which made those drives more or less obsolete. To improve access times while retaining the high capacities of hard drives, there are now hybrid SSHDs, hard drives with an SSD cache, which can automatically store some gigabytes of regularly-accessed files that see lots of random accesses to an SSD portion of the drive, effectively improving the random access times of those files far more than what a faster spindle speed would provide. Or a separate SSD cache can be used with standard drives in a similar way.

As for pricing, as bit_user said, hard drives tend to have a price floor of around $45 or so for a modern, single-platter drive. There's a lot of materials and precision mechanical components in such a drive, so you shouldn't expect prices of the lowest-capacity models to fall much below that. And since desktop drive platters have been at 1TB (or a bit higher) for a number of years now, that is what a 1TB drive costs. It doesn't cost that much more to add additional platters though, which is why you can find 2TB drives for as little as $60, a doubling of capacity for about a 35% increase in price, or 3TB for as little as $80, if not less in sales. As a result, 1TB drives tend to have the worst cost per GB among modern drives.

You could also pay more for a more expensive "Black" drive or whatever, but don't expect it to perform significantly better or last any longer than other 7200 RPM desktop drives. If you want to pay more for performance, that's what SSDs are for, or perhaps a hybrid drive like a Firecuda if you want something in-between. 500GB SSDs that beat a hard drive in every way in terms of performance are already widely available in the sub-$100 price range, and it shouldn't be long before 1TB SSDs get there as well, so HDDs will likely soon be relegated strictly to bulk data storage. For anyone not storing lots of data like video, hard drives may become a thing of the past within the next few years, and will likely become more of a niche product that's not installed on most new computers.

Dont need 100TB yet....but with games now taking about 100GB a piece, it wont be all that long until you see games at the 1TB level. I would not be suprised to see games at 1TB by 2025.

Games don't average 100GB though, only a small minority of them manage that. Most recent AAA games are still under 50GB. Just look at some of the popular game releases on Steam in 2018 and compare their listed install sizes...

Pillars of Eternity II: 45GB
Monster Hunter World: 20GB
Assassin's Creed Odyssey: 46GB
Final Fantasy XV: 100GB
Valkyria Chronicles 4: 36GB
Frostpunk: 8GB
Final Fantasy XII The Zodiac Age: 50GB
F1 2018: 50GB
Warhammer Vermintide 2: 45GB
Ni no Kuni II: 40GB
SoulCalibur VI: 20GB
Far Cry 5: 40GB
Shadow of the Tomb Raider: 40GB
Kingdom Come Deliverance: 40GB
Vampyr: 20GB
Metal Gear Survive: 20GB

Install sizes much above 50GB tend to be relatively uncommon, with Final Fantasy XV being the only 2018 release on Steam that I can immediately think of (along with a few others not on Steam). And for those larger games, it's often simply a case of their data being stored in an inefficient, uncompressed format. With thread counts on processors increasing lately, there should be even less reason to not compress game files though. And perhaps increased use of procedurally generated content could help keep install sizes in check as well.

Also consider that while hard drive density can increase, performance tends to increase at a slower rate, so you can't just keep doubling the size of games and expect them to load in a reasonable amount of time off a standard hard drive. Plus, it's typically rather easy for most of those with modern broadband connections to re-download games, so there often isn't actually a need to keep one's entire game library installed. Once you've played through a single-player game, unless it offers significant replayability, you probably won't be coming back to it for a while.

Do I think a 1TB game by 2025 could be possible? Sure, but that probably won't be the norm. And by then, we might even have 10TB SSDs for not much more than $100, and you probably wouldn't want to wait through the ten-minute load times of that game on a traditional hard drive anyway. : P

 

[photonboy]

SSD prices will continue to drop relatively faster than HDD's.
Also, since they comprise a controller and the memory chips primarily, then price is mainly dictated by the price of the memory CHIPS.

An HDD has a lot of overhead with the case, arms, motors etc whereas an SSD not counting the memory is just a plastic case, simple circuit board and controller. Not much cost there before the memory chips.

As SSD memory prices drop they'll push HDD's out of the lower capacity spaces so in a few short years a 1TB HDD will not make any sense. The SSD only needs to cost slightly MORE per GB (since it's faster) to displace the HDD.

Put another way if an HDD cost $40 and an SSD cost $50 which one would you buy?

Remember when an 8MB (not GB) USB stick was like $30?

 

[bit_user]

SSD prices will continue to drop relatively faster than HDD's.

Really? Do you really have that much faith that lithography improvements will continue, unabated? Are you sure you're not just extrapolating from previous trends?

If not, what are the developments and technologies that will surpass the gains leading to 100 TB HDDs in just 6 years? I only did a quick search, but I don't see any flash memory roadmaps going out that far.

 

[stdragon]

SSD prices will continue to drop relatively faster than HDD's.

Really? Do you really have that much faith that lithography improvements will continue, unabated? Are you sure you're not just extrapolating from previous trends?

If not, what are the developments and technologies that will surpass the gains leading to 100 TB HDDs in just 6 years? I only did a quick search, but I don't see any flash memory roadmaps going out that far.

It's my limited understanding that the biggest cost in SSD technology is the initial R&D. But once done, it's all about volume and mass production. With HDDs, there's a lot of R&D, but I would have though by now SSDs would be cheaper. I mean, it's quite amazing how what otherwise is an exceedingly complicated electro-mechanical device that contains all sorts of rare earth materials with a complicated production process and mechanical tooling. With SSDs, the production of chips scales out rapidly and serves all sorts of other markets from the foundry. So yeah, i would expect at some point HDD tech hits a wall in terms of cost per GB in value. Apparently we haven't reached it yet. That in of it self blows me away!

 

[USAFRet]

It's my limited understanding that the biggest cost in SSD technology is the initial R&D. But once done, it's all about volume and mass production. With HDDs, there's a lot of R&D, but I would have though by now SSDs would be cheaper. I mean, it's quite amazing how what otherwise is an exceedingly complicated electro-mechanical device that contains all sorts of rare earth materials with a complicated production process and mechanical tooling. With SSDs, the production of chips scales out rapidly and serves all sorts of other markets from the foundry. So yeah, i would expect at some point HDD tech hits a wall in terms of cost per GB in value. Apparently we haven't reached it yet. That in of it self blows me away!

With both type of drives, it is also cramming more and more GB into a particular amount of sq in.
Be it a platter, or multiple platters, that is a hard limit on square inches. Only way to increase capacity is to increase the areal density, and still be readable at XX00 RPM.

In my pile of crap, I have a 130MB laptop drive, from mid 90's. Data in current drive of the same size platter is 10,000 times more dense. When 1TB drives first came out, predictions were that we couldn't get any larger. 1TB was a hard physical limit. And then quickly shattered.

With SSD's, how much can we cram into each chip, because we can only make the chips so big, and still fit in the same size package. 2.5" box or m.2 2280. This is where 3D NAND came into play. NAND cells in vertical orientation, and how to access.
And then quality control in production of that ever increasing chip density.

 

[photonboy]

SSD prices will continue to drop relatively faster than HDD's.

Really? Do you really have that much faith that lithography improvements will continue, unabated? Are you sure you're not just extrapolating from previous trends?

If not, what are the developments and technologies that will surpass the gains leading to 100 TB HDDs in just 6 years? I only did a quick search, but I don't see any flash memory roadmaps going out that far.

Please don't misunderstand my post.
I did NOT say SSD's would soon replace all HDD's at all price points. I said they would replace them at the LOWER capacities and slowly work their way up. I thought I was pretty clear in fact.

I said nothing about 100TB SSD's.

And by "relatively" I mean for example that a 2TB SSD in two years is probably going to drop in price more than a 2TB HDD will relative to their current prices.

 

[bit_user]

SSD prices will continue to drop relatively faster than HDD's.

Really? Do you really have that much faith that lithography improvements will continue, unabated? Are you sure you're not just extrapolating from previous trends?

If not, what are the developments and technologies that will surpass the gains leading to 100 TB HDDs in just 6 years? I only did a quick search, but I don't see any flash memory roadmaps going out that far.

Please don't misunderstand my post.
I did NOT say SSD's would soon replace all HDD's at all price points.

I was reacting to the specific statement I quoted, as it implies $/GB for big SSDs dropping faster than for big HDDs. I don't know if that's possible, even without regard to whether big SSDs actually catch big HDDs.

As for the cost advantage of small SSDs, I see that as another consequence of the HDD price floor we already discussed. So, no disagreement there.

 

[photonboy]

SSD prices will continue to drop relatively faster than HDD's.

Really? Do you really have that much faith that lithography improvements will continue, unabated? Are you sure you're not just extrapolating from previous trends?

If not, what are the developments and technologies that will surpass the gains leading to 100 TB HDDs in just 6 years? I only did a quick search, but I don't see any flash memory roadmaps going out that far.

Please don't misunderstand my post.
I did NOT say SSD's would soon replace all HDD's at all price points.

I was reacting to the specific statement I quoted, as it implies $/GB for big SSDs dropping faster than for big HDDs. I don't know if that's possible, even without regard to whether big SSDs actually catch big HDDs.

As for the cost advantage of small SSDs, I see that as another consequence of the HDD price floor we already discussed. So, no disagreement there.

That's what happens when you take one sentence out of context. You say it implies... for "big SSDs dropping faster than for big HDD's.."?

But in context I said nothing of the sort. Here's another sentence from the same post:
"As SSD memory prices drop they'll push HDD's out of the lower capacity spaces..."

That's not the same thing at all. I said "lower capacity spaces"; seems straight forward to me... but since you thought that implied larger capacities too then let's think about that. Go look up the cost/GB for SSD's and HDD's and see what's dropped the most RELATIVELY... I'd be surprised if SSD's were not dropping quicker.

https://www.backblaze.com/blog/hard-drive-cost-per-gigabyte/
"the rate of change has slowed dramatically as observed in the chart below which represents our average quarterly cost per gigabyte over time."

*but to be fair, it's not so clear though IMO the info below is more short-term, not long-term:
https://www.theregister.co.uk/2018/03/23/disk_price_flash_analysis/

My opinion is that SSD's will replace HDD's at the lower capacities and slowly work their way up. I think short-term boosts in density due to HAMR or MAMR will help HDD's but FLASH memory IMO will continue to get cheaper and at some point it may only be SERVERS that benefit from HDD's.

Oh, I don't mean the next few years but it's hard to imagine many HDD's being purchased in the consumer desktop space past say 2030. Some I'm sure, but not many outside of servers and those with needs of say more than 2TB. IMO.