Disappointing hard drive value (was: raw files are HUGE)

[Allen]

Ken Lucke wrote:

Besides, what's the problem? Storage has never been cheaper - I just
picked up [yet another] Maxtor 300 GB Firewire/USB drive from Costco
for $149, including cables... that's 50 cents a gig. That makes 3

However using USB or firewire greately limits the speed. Now you can
install an external SATA 3 that will run as fast as the internal
drives.

I paid $10 USD less than that each for 4 500 Gig SATA 3 drives.
$139 X 4 = $556 for 2 Terabytes.

Flash memory hasn't come close to being able to match hard drives for
capacity. Yes, I can purchase a 1 Gig CF for about $20 to $30
compared to the $70 at discount I paid for them when first available.
Stores were selling them for $129 at the time.

But at any rate, when you compare the changes in HDs from the many
thousands of dollars for a 10 Meg Winchester drive of the 80's to the
1 Terabyte USB external drives for $400 give or take about $50 those
HDs certainly have kept pace with drive capacity for a reasonable
price about trippeling in the last year.

You've been going to the wrong stores. A few years ago HDs were
expensive and running in the 120 Gig range. I have one left that is
that small. I have over 6 terabytes...make that 7 between 5
computers.
Those 5 500 Gig SATA 3 drives with 16 meg caches cost about 1.5 times
the price of the 120 Gig drive I mentioned. You can now purchase 80 to
120 Gig drives for around $50 pt $60 with a bit of shopping. Actually
Best Buy had 250s for $89 a while back.

I just looked at New Egg and they have a 250 Gig SATA 3 WD (8 meg
cache)for $70, a 320 Seagate SATA 3 for $90 and a 500 Gig SATA 3 WD
for $139. The last two have 16 Meg caches.

BTW they have a 1 Terabyte external with 32 Meg Cache set up for both
USB2 and Firewire and fan cooled for $379 after rebate. ($9 shipping)

<edited, for brevity>

Hello, timeOday:

Hard disks have been around far longer, than flash memory. Hence, it's
unrealistic to expect the same kinds of rapid price/performance gains,
from such mature technology.

What we are seeing is still a rapid increase in capacity and prices
*still* coming down. One Terabyte external (with fan) for about the
same price as a 400 Gig just 6 months ago.

Cordially,
John Turco <[email protected]>

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com

I just read the Fry's Electronics ad in this morning's paper and saw
that they have 500Gb Maxtor SATA drive for $130, which further confirms
Roger's price. I remember buying many 30 MEGAbyte drives back in the
1980s for more than that; I had unwillingly become my company's PC guru
and installed those same drives. I also remember the 7.5 Mb packs (six
discs, 10 usable surfaces) to use on mainframe drives in the late 1960s,
at around $650 each. Please, don't anyone tell me that drive prices
are static.
Allen

 

[Dave Martindale]

Pretty much, drive speeds have gone up in step with drive capacities.
You get greater capacity either with more platters or a higher data density.
Both translate to greater throughput given the same rotational speed.

No, drive speeds go up with about the square root of drive capacity.
Capacity increases with higher track density on a platter, more
platters, and higher data density along the track. Only the latter
increases data rate given the same rotational speed. The other two
factors only increase the total number of tracks.

Dave

 

[AZ Nomad]

No, drive speeds go up with about the square root of drive capacity.
Capacity increases with higher track density on a platter, more
platters, and higher data density along the track. Only the latter

The number of platters hasn't increased over the years. The top capacity
drives have more platters than entry level drives, but the number of
platters used hasn't increased over the last twenty years.
You're not going to find a drive with eighty platters.

 

[J. Clarke]

timeOday wrote:
Ken Lucke wrote:

Besides, what's the problem? Storage has never been cheaper - I
just picked up [yet another] Maxtor 300 GB Firewire/USB drive
from Costco for $149, including cables... that's 50 cents a gig.
That makes 3

However using USB or firewire greately limits the speed. Now you can
install an external SATA 3 that will run as fast as the internal
drives.

250GB, 1 400GB, and 2 300 GB drives attached to my system.

I paid $10 USD less than that each for 4 500 Gig SATA 3 drives.
$139 X 4 = $556 for 2 Terabytes.

I have surely appreciated the crazy explosion in flash memory
capacity,

Flash memory hasn't come close to being able to match hard drives for
capacity. Yes, I can purchase a 1 Gig CF for about $20 to $30
compared to the $70 at discount I paid for them when first available.
Stores were selling them for $129 at the time.

But at any rate, when you compare the changes in HDs from the many
thousands of dollars for a 10 Meg Winchester drive of the 80's to the
1 Terabyte USB external drives for $400 give or take about $50 those
HDs certainly have kept pace with drive capacity for a reasonable
price about trippeling in the last year.

but hard drives are not keeping pace. I was just looking for a
drive today and was disappointed that hard drive prices haven't
fallen more since I bought my last drive a few years ago.

You've been going to the wrong stores. A few years ago HDs were
expensive and running in the 120 Gig range. I have one left that is
that small. I have over 6 terabytes...make that 7 between 5
computers.
Those 5 500 Gig SATA 3 drives with 16 meg caches cost about 1.5 times
the price of the 120 Gig drive I mentioned. You can now purchase 80
to 120 Gig drives for around $50 pt $60 with a bit of shopping.
Actually Best Buy had 250s for $89 a while back.

I just looked at New Egg and they have a 250 Gig SATA 3 WD (8 meg
cache)for $70, a 320 Seagate SATA 3 for $90 and a 500 Gig SATA 3 WD
for $139. The last two have 16 Meg caches.

BTW they have a 1 Terabyte external with 32 Meg Cache set up for both
USB2 and Firewire and fan cooled for $379 after rebate. ($9 shipping)

<edited, for brevity>

Hello, timeOday:

Hard disks have been around far longer, than flash memory. Hence,
it's unrealistic to expect the same kinds of rapid
price/performance gains, from such mature technology.

What we are seeing is still a rapid increase in capacity and prices
*still* coming down. One Terabyte external (with fan) for about the
same price as a 400 Gig just 6 months ago.

Cordially,
John Turco <[email protected]>

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com

I just read the Fry's Electronics ad in this morning's paper and saw
that they have 500Gb Maxtor SATA drive for $130, which further
confirms Roger's price. I remember buying many 30 MEGAbyte drives
back in the 1980s for more than that; I had unwillingly become my
company's PC guru and installed those same drives. I also remember
the 7.5 Mb packs (six discs, 10 usable surfaces) to use on mainframe
drives in the late 1960s, at around $650 each. Please, don't anyone
tell me that drive prices
are static.

The pace of progress in the disk industry seems to have slowed. It used
to be that new drives came in around $400 or so and dropped out around
$40 or so. But the 500s are approaching the $100 mark and the 750s are
under $200 with no next generation yet on the market--the terabyte
drives are supposed to be out this month I understand at around
$400--they may very well hit the sub $100 range before the next
generation ships.

 

[Folkert Rienstra]

The number of platters hasn't increased over the years.

It fluctuates.

The top capacity drives have more platters than entry level drives,

But they didn't use the maximum number of platters possible for a long time.

but the number of platters used hasn't increased over the last twenty years.

Oh, yes it did.

You're not going to find a drive with eighty platters.

But you may find them with the maximum number that can fit within a standard
half height drive. But then, with perpendicular recording now, the need for that
may be over for quite a while again.

 

[Folkert Rienstra]

No, drive speeds go up with about the square root of drive capacity.

Platter capacity.

 

[AZ Nomad]

No, drive speeds go up with about the square root of drive capacity.
Capacity increases with higher track density on a platter, more
platters, and higher data density along the track. Only the latter
increases data rate given the same rotational speed. The other two
factors only increase the total number of tracks.

You're right about performance going up w/ the square root of capacity...
Think of tracks*data_density*platters as data in three dimensions.
The platters is pretty much fixed; track density and data density have both
improved. Only data density improves performance. You can only
read one track at a time.

 

[ASAAR]

You're right about performance going up w/ the square root of capacity...
Think of tracks*data_density*platters as data in three dimensions.
The platters is pretty much fixed; track density and data density have both
improved. Only data density improves performance. You can only
read one track at a time.

One head can only read one track at a time, but I recall reading
about drives that had two or three read/write heads. This might
have been 15 or more years ago, so I don't recall whether the heads
were constrained to the same track, but I don't think so. IIRC,
there were independent actuators. I'm sure that these drives were
out of my price range, nor were they likely to be the MFM and RLL
drives that I used back then.

 

[Bob Willard]

One head can only read one track at a time, but I recall reading
about drives that had two or three read/write heads. This might
have been 15 or more years ago, so I don't recall whether the heads
were constrained to the same track, but I don't think so. IIRC,
there were independent actuators. I'm sure that these drives were
out of my price range, nor were they likely to be the MFM and RLL
drives that I used back then.

Standard practice is multiple heads per cylinder (one head per surface),
but only one active R/W head at a time, so only one set of R/W channel
electronics is needed.

In the 80's, there was at least one rather expensive HD (ISIS) with parallel
R/W channels, but only one shared actuator. Great for STR, but no help
for seek time.

In the 60's, there was at least one HD (IBM 1301)which offered multiple
actuators, which potentially gave improved STR as well as improved seek time.
But IIRC, the max. seek time on that huge beast was still seven seconds:
the time to recirculate the oil in the hydraulic adder (sometimes needed
to remove oil bubbles).

In the 50's (the dawn of HD time), the IBM 350 had one actuator and two
(IIRC) heads for its 25-platter assembly. For a short seek, the actuator
moved in or out along the radius of the platter stack; for a long seek,
the actuator moved out beyond the platter stack, then up or down parallel
to the central shaft, then in to the desired platter and cylinder. Seek
times were rather long, but still beat the snot out of the competition,
which was mag. tape.

 

[ASAAR]

Standard practice is multiple heads per cylinder (one head per surface),
but only one active R/W head at a time, so only one set of R/W channel
electronics is needed.

Well of course I've never heard of a drive that didn't have a head
per platter surface. Seek time between surfaces would really suffer
if a disk drive resembled a one head per cylinder carousel. <g>
The ones I heard of with multiple sets of heads probably had only a
single set of R/W channel electronics per stack of heads, but a set
for every stack of heads. Although they could have been completely
independent, I imagine that when the channels were active, they were
all servicing the same track on the same platter surface, buffering
the entire track and of course allowing the complete track to be
read in a fraction of the normal time.

 

[Folkert Rienstra]

You're right about performance going up w/ the square root of capacity...

Platter capacity.

Think of tracks*data_density*platters as data in three dimensions.

The platters is pretty much fixed;
track density and data density have both improved.

Only data density improves performance.

Right, so much for your:
"You're right about performance going up w/ the square root of capacity"

 

[Rod Speed]

Standard practice is multiple heads per cylinder (one head per surface),
but only one active R/W head at a time, so only one set of R/W channel
electronics is needed.

In the 80's, there was at least one rather expensive HD (ISIS) with parallel
R/W channels, but only one shared actuator. Great for STR, but no help
for seek time.

In the 60's, there was at least one HD (IBM 1301)which offered multiple
actuators, which potentially gave improved STR as well as improved seek time.
But IIRC, the max. seek time on that huge beast was still seven seconds:
the time to recirculate the oil in the hydraulic adder (sometimes needed
to remove oil bubbles).

In the 50's (the dawn of HD time), the IBM 350 had one actuator and two
(IIRC) heads for its 25-platter assembly. For a short seek, the actuator
moved in or out along the radius of the platter stack; for a long seek,
the actuator moved out beyond the platter stack, then up or down parallel
to the central shaft, then in to the desired platter and cylinder. Seek
times were rather long, but still beat the snot out of the competition,
which was mag. tape.

And the DEC RS09 had fixed heads, one per track, literally.

 

[Roger]

Consider, also, that HD capacity increases are subject to big leaps
when a new technology is introduced, followed by a plateau until the

Such as the recent change to using vertical magnetic domains instead
of horizontal. I think all the new SATA drives are already using
this.

next leap. Flash-memory capacity, on the other hand, inexorably
increases according to Moore's Law.

Roger Halstead (K8RI & ARRL life member)
(N833R, S# CD-2 Worlds oldest Debonair)
www.rogerhalstead.com

 

[Dave Martindale]

Platter capacity.

That's the most correct answer, but you can't tell the platter capacity
from the outside of the drive, just the overall capacity.

On the other hand, as long as you compare drives of the same physical
size, which are the highest capacity in their "family", the number of
platters remains about the same, and so platter capacity is roughly
proportional to drive capacity.

On the other hand, if you compare today's 80 MB drive to an 80 MB drive
from 5 years ago, the latter probably has 3 or 4X as many surfaces as
the former.

Dave

 

[CJT]

That's the most correct answer, but you can't tell the platter capacity
from the outside of the drive, just the overall capacity.

On the other hand, as long as you compare drives of the same physical
size, which are the highest capacity in their "family", the number of
platters remains about the same, and so platter capacity is roughly
proportional to drive capacity.

On the other hand, if you compare today's 80 MB drive to an 80 MB drive
from 5 years ago, the latter probably has 3 or 4X as many surfaces as
the former.

Dave

I doubt you can find an example of "today's 80 MB drive" to compare with.

 

[Folkert Rienstra]

"Folkert Rienstra" <[email protected]> writes:

You are posting the Reply-to address instead of the From address.

That's the most correct answer,

On rethink, that should have read 'surface capacity'.
The smallest drive in the range may even use only one
surface or even only a partial surface.

but you can't tell the platter capacity from the
outside of the drive, just the overall capacity.

Which is no reason for making a wrong statement.
And I think that the square root means that a capacity increase is derived
from equal sector and track density increase. That may not be true either.

On the other hand, as long as you compare drives of the same physical
size, which are the highest capacity in their "family",

That sentence makes absolutely no sense. Presumably you mean
"same physical make-up", like same number of platters, heads, rpm etc.

the number of platters remains about the same, and so platter capacity
is roughly proportional to drive capacity.

Until you change to Perpendicular Recording and all bets are off again.