HDD capacity stagnation

[Tom Del Rosso]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

 

[Rod Speed]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the longest time (a lot longer than we were
stuck at 528MB)

Thats very arguable indeed.

but when the gates opened we got 3TB, and not 8TB. Is there another obstacle?

Nope.

 

[Arno]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

The technology is at its limits. 8TB does not make ecconomic
sense, it would not fit standard HDD case sizes and it would
take forever due to slow interfaces. It would also be very
expensive. As to size progression, why would you expect
8TB? Sizes did never increase in tsucha large step.

Arno

 

[Franc Zabkar]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

At the moment we're at 1TB per platter. Therefore we should be able to
fit 4TB in a 3.5" form factor using current technology.

- Franc Zabkar

 

[Tom Del Rosso]

The technology is at its limits. 8TB does not make ecconomic
sense, it would not fit standard HDD case sizes and it would
take forever due to slow interfaces. It would also be very
expensive. As to size progression, why would you expect
8TB? Sizes did never increase in tsucha large step.

The perception of time varies a lot, so it's very subjective, but it seems
like the software limit that held us to 2TB lasted longer than the software
limit that held us to 528MB. When the 528MB limit was overcome we reached
2GB very soon as I recall.

The maximum size available in 1993 was 528MB and in 2003 it was 250GB. A
multiple of 500 in 10 years. Maybe that should be 1995, but then it would
have been only 8 years.

Since then it has changed by a factor of 12 in 8 years. So I asked if there
is a specific obstacle that is now in the way.

Case size, interface speed, and price have always been factors.

 

[Percival P. Cassidy]

At the moment we're at 1TB per platter. Therefore we should be able to
fit 4TB in a 3.5" form factor using current technology.

Such as the 4TB Seagate GoFlex Desk STAC4000100

But the internals go only up to 3TB

Perce

 

[DevilsPGD]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

Right now we're at some physics limits with regards to densities. We
were close to this point before, but perpendicular magnetic recording
suddenly becoming stable/reliable helped get through the last time.

That's not to say there won't be any capacity increases in the future,
but at this point we're waiting on a breakthrough rather than an
incremental increase. Until there is a breakthrough, performance or
reliability suffer greatly along with significant cost increases if you
try to ramp up density. Expect maybe 10% density increases, likely
mostly being lost to additional redundancy in the drive (in other words,
I'd expect to see more reliable drives before higher capacity drives)

More importantly, in the consumer space, few people really need more
than 2TB/3TB drives. In the high density storage business world,
smaller, more power efficient drives are all the rage.

Combining the current density limits with the fact that most of the
major hard drive manufacturers are rebuilding after floods will probably
leave us around the current range for a while. Once the manufacturers
get rebuilt, expect to see better and faster drives at current densities
(why build a factory to sell 500GB drives when it's the same factory
costs for 2TB drives and 2TB drives see for more?)

Given that we're up to 1TB per platter, 4TB drives should be possible in
typical form-factors, but my guess is that 3TB drives didn't sell well
enough to be worth investing in 4TB designs just yet.

 

[Rod Speed]

DevilsPGD wrote

Right now we're at some physics limits with regards to densities. We
were close to this point before, but perpendicular magnetic recording
suddenly becoming stable/reliable helped get through the last time.

That's not to say there won't be any capacity increases in the future,
but at this point we're waiting on a breakthrough rather than an
incremental increase. Until there is a breakthrough, performance or
reliability suffer greatly along with significant cost increases if
you try to ramp up density. Expect maybe 10% density increases, likely
mostly being lost to additional redundancy in the drive (in other words,
I'd expect to see more reliable drives before higher capacity drives)

More importantly, in the consumer space, few
people really need more than 2TB/3TB drives.

Thats just plain wrong, most obviously with PVRs.

In the high density storage business world, smaller,
more power efficient drives are all the rage.

Not with the operations that need massive amounts of storage like google.

Combining the current density limits with the fact that most of
the major hard drive manufacturers are rebuilding after floods
will probably leave us around the current range for a while.

They didnt all build in Thailand.

Once the manufacturers get rebuilt, expect to see better and faster drives
at current densities (why build a factory to sell 500GB drives when it's
the same factory costs for 2TB drives and 2TB drives see for more?)

Given that we're up to 1TB per platter, 4TB drives should be
possible in typical form-factors, but my guess is that 3TB drives
didn't sell well enough to be worth investing in 4TB designs just yet.

And how well they sell obviously depends on the price charged for them.

 

[DevilsPGD]

DevilsPGD wrote




Thats just plain wrong, most obviously with PVRs.

PVRs are one usage case, for the vanishingly small percentage of people
that 1) Have a PVR, 2) Overload the PVR, 3) Have the will to upgrade
their PVR, 4) Have a PVR that can be upgraded, 5) Have a PVR that can
handle larger than 2TB drives and 6) Have a PVR that can only handle a
single drive.

Like I said, "few people really need more than 2TB/3TB drives." Some do,
and pointing out specific narrow examples just makes my point.

Not with the operations that need massive amounts of storage like google.

Things work a little differently in the business world. Google needs a
lot of storage, but they also need very low latency, so having more
spindles can make sense.

Power efficiency:TB is probably a bigger factor than raw TB count, get
the power cost low enough and it makes sense to crank up the spindle
count for performance reasons.

They didnt all build in Thailand.

No, but a significant percentage of drives were manufactured in
Thailand, so the market will be affected for a while.

 

[Arno]

The perception of time varies a lot, so it's very subjective, but it seems
like the software limit that held us to 2TB lasted longer than the software
limit that held us to 528MB. When the 528MB limit was overcome we reached
2GB very soon as I recall.

The maximum size available in 1993 was 528MB and in 2003 it was 250GB. A
multiple of 500 in 10 years. Maybe that should be 1995, but then it would
have been only 8 years.

Since then it has changed by a factor of 12 in 8 years. So I asked if there
is a specific obstacle that is now in the way.

Ah, yes. There is: Data density per area. And it looks like this
time it may be a hard limit, i.e. one that cannot be overcome
anytime soon.

Arno

 

[Rod Speed]

DevilsPGD wrote

PVRs are one usage case,

Which blows your silly claim completely out of the water.

for the vanishingly small percentage of people

Even sillier.

that 1) Have a PVR,

Thats nothing like a vanishingly small percentage.

2) Overload the PVR,

Even sillier. You dont have to overload it to need more than 2/3TB.

3) Have the will to upgrade their PVR,

Even sillier when PVRs would be supplied with the larger
drives if they were available and reasonably priced.

4) Have a PVR that can be upgraded,

Or are buying a PVR.

5) Have a PVR that can handle larger than 2TB drives

Which they will certainly do when the larger
drives are available and are reasonably priced.

and 6) Have a PVR that can only handle a single drive.

Even sillier.

Like I said, "few people really need more than 2TB/3TB drives."

Repeating that stupid pig ignorant claim changes nothing.

Some do, and pointing out specific narrow examples just makes my point.

Taint a narrow example at all, fool.

Things work a little differently in the business world.

Not on the need for drives bigger than 2/3TB they dont except in the
sense that operations like google need immense amounts of storage.

Google needs a lot of storage,

And they aint alone in that.

but they also need very low latency, so having more spindles can make sense.

Like hell it does with the immense amount of storage they need.
They'd have plenty of spindles even with drives bigger than 2/3TB.

Power efficiency:TB is probably a bigger factor than raw TB count,

Wrong, as always.

get the power cost low enough and it makes sense
to crank up the spindle count for performance reasons.

Not with the immense amount of storage operations like
google need, they already have vast numbers of spindles.

No, but a significant percentage of drives were manufactured
in Thailand, so the market will be affected for a while.

Irrelevant to that stupid claim you made about MOST MANUFACTURERS.

 

[Joe Kotroczo]

Such as the 4TB Seagate GoFlex Desk STAC4000100

But the internals go only up to 3TB

Not anymore:

<http://www.hitachigst.com/internal-drives/desktop/deskstar/deskstar-7k4000>

 

[Lynn McGuire]

Right now we're at some physics limits with regards to densities. We
were close to this point before, but perpendicular magnetic recording
suddenly becoming stable/reliable helped get through the last time.

That's not to say there won't be any capacity increases in the future,
but at this point we're waiting on a breakthrough rather than an
incremental increase. Until there is a breakthrough, performance or
reliability suffer greatly along with significant cost increases if you
try to ramp up density. Expect maybe 10% density increases, likely
mostly being lost to additional redundancy in the drive (in other words,
I'd expect to see more reliable drives before higher capacity drives)

More importantly, in the consumer space, few people really need more
than 2TB/3TB drives. In the high density storage business world,
smaller, more power efficient drives are all the rage.

Combining the current density limits with the fact that most of the
major hard drive manufacturers are rebuilding after floods will probably
leave us around the current range for a while. Once the manufacturers
get rebuilt, expect to see better and faster drives at current densities
(why build a factory to sell 500GB drives when it's the same factory
costs for 2TB drives and 2TB drives see for more?)

Given that we're up to 1TB per platter, 4TB drives should be possible in
typical form-factors, but my guess is that 3TB drives didn't sell well
enough to be worth investing in 4TB designs just yet.

Don't forget that the cloud systems are the number
two client of hard drives after consumers right now
(saw that somewhere but dont remember where).
http://arstechnica.com/business/new...-how-the-giants-of-the-web-store-big-data.ars

The cloud systems would love to have 1 PB drives.

Lynn

 

[DevilsPGD]

Don't forget that the cloud systems are the number
two client of hard drives after consumers right now
(saw that somewhere but dont remember where).
http://arstechnica.com/business/new...-how-the-giants-of-the-web-store-big-data.ars

The cloud systems would love to have 1 PB drives.

I doubt they care. Rather, they care about the cost:efficiency ratio,
where cost factors in the original purchase price and energy consumption
over it's lifespan (including cooling) vs some combination of
performance and size. Physical size and formfactor is also a factor, of
course.

In other words, I agree, but it's more complicated than just saying "a 1
PB drive"

Sure, large drives would be great, but only once they beat out the price
point that modern drives achieve.

 

[Arno]

In the last episode of <[email protected]>, Lynn McGuire
<[email protected]> said:

I doubt they care. Rather, they care about the cost:efficiency ratio,
where cost factors in the original purchase price and energy consumption
over it's lifespan (including cooling) vs some combination of
performance and size. Physical size and formfactor is also a factor, of
course.

In other words, I agree, but it's more complicated than just saying "a 1
PB drive"

Sure, large drives would be great, but only once they beat out the price
point that modern drives achieve.

I think another issue is more problematic: I/O bandwidth to
capacity ratio. It has gotten worse and worse over time for
HDDs. Not a surprise as data-density increases on a plane,
while r/w is done on circles. Unless you have rare access
for most data, large disks are not very useful. Now, "the
cloud" (stupid term, it really is just client-server computing),
may turn out to be a gigantic data-heap, in which case the more
capacity, the better, but I somehow doubt that, as it clearly
is unreliable as a storage solution. These days, Megaupload
demonstrates the fundamental problem. It also demonstrates
thet "the cloud" is really a number of different special-purpose
clouds. For storage, large capacity may matter, but see the
lack of maturity of the whole idea. For computing, I do not
see this at all, and neither for presence (web-server, e.g.)
clouds.

Side note: The main selling-point for SSDs is that they
have a far saner I/O to capacity rating. This also explains
why they sell so well, despite being massively more expensive.

Arno

 

[Lynn McGuire]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

Amazon is selling a 4 TB Seagate for $425:
http://www.amazon.com/Seagate-Barracuda-ST4000DX000-SATA-Drive/dp/B005WX3NEU/

Lynn

 

[Rod Speed]

DevilsPGD wrote

I doubt they care.

Corse they do, 50 times less physical drives is a hell of a lot easier to handle.

Rather, they care about the cost:efficiency ratio,
Sure.

where cost factors in the original purchase price and
energy consumption over it's lifespan (including cooling)
vs some combination of performance and size. Physical
size and formfactor is also a factor, of course.
Waffle.

In other words, I agree, but it's more complicated than just saying "a 1 PB drive"

Sure, large drives would be great, but only once they
beat out the price point that modern drives achieve.

Duh.

But you stupidly claimed that there is no market for drives bigger than 2/3TB.

 

[Man-wai Chang]

Capacity seems to be progressing very slowly. We were stuck at 2TB for the
longest time (a lot longer than we were stuck at 528MB) but when the gates
opened we got 3TB, and not 8TB. Is there another obstacle?

Increasing the size of a hard disk is NOT just 1+1=2, 2+2=4... ! It's
more about physics.

--
@~@ You have the right to remain silence.
/ v \ Simplicity is Beauty! May the Force and farces be with you!
/( _ )\ (Fedora 15 i686) Linux 3.2.1
^ ^ 03:16:01 up 1 day 23:23 0 users load average: 0.01 0.03 0.08
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa

 

[Arno]

Increasing the size of a hard disk is NOT just 1+1=2, 2+2=4... ! It's
more about physics.

Whaaaat? I thought bits were _logical_ objects, not physical ones???

;-)=)

Arno

 

[Tom Del Rosso]

Whaaaat? I thought bits were _logical_ objects, not physical ones???

Of course I did ask for some kind of specifics. I don't know how it could
be ambiguous to anyone familiar with the history of the rate of growth in
capacity.