June 30, 2003

Terabits per Square Inch...

Some awesome things are going on in the buildings just to the north:

Berkeley Engineering Lab Notes: Printer-friendly version: ...Moore's Law has been outpaced by the explosion in data storage. Every time you open a computer catalog, it seems, you can buy a bigger hard drive for less money. The price/performance curve of hard drives is now steeper than that of microprocessors. Most recently, the storage industry demonstrated hard disks that can pack 100 gigabits of data into one square inch of magnetic media. Within the next few years, manufacturers promise a whopping one terabit per square inch storage density. To help the industry reach that milestone, UC Berkeley professor of mechanical engineering Roberto Horowitz and his students are building microscopic actuators and sensors that enable drives to pack bits just nanometers apart. The actual disk in any PC's hard drive consists of an aluminum alloy plated with a magnetizable material. A motor spins the disk at up to 10,000 revolutions per minute. Another electromagnetic "voice coil" motor controls the position of the suspension system that holds the read/write head, not unlike a needle on the end of a record player's tone arm. As the platter spins, a cushion of air is created that lifts the head just above the platter so it literally flies across the media as it reads and writes data in a circular track. The closer the tracks are together, the more data can be stored in the disk.

Posted by DeLong at June 30, 2003 08:38 PM | TrackBack

Comments

While electromechanical controls are
crossing frontiers of their own, it's
the twin concerns of reliability and
density of bit storage that has made
huge gains. Researchers once believed
magnetic domains could not be reduced
beyond a certain minimum and some
forecast a theoretical "wall" for
storage densities. But while this led
to research in use of biological, not
magnetic materials for data encoding,
researchers at IBM and elsewhere made
critical advances which assured a
future for magnetics as a preferred
data medium.

Today, magnetic storage is found on
in many different formats-- from the
lowly tape cassette to ultra-high-
performance data cartridges of sizes
approaching a terabyte. But the data
storage choice for most systems where
extreme reliability is of paramount
importance continues to be one, or an
array of magnetic hard disks.

With manufacturing costs going down
and capacity soaring, most manufacturers
will stay with magnetics for a while.
They want the ability to handle the
convergence of PC and video devices,
a new format hungry for both high data
transfer rate and storage density.

Posted by: alpha on July 1, 2003 12:34 AM

What'll we do with all that storage space? It seems to me that hardware has outpaced software in the last few years. A word processor or a spreadsheet, for instance, do roughly what they did ten years ago, but take up a hundred times as much space. Games pack on lots of bells and whistles, but the gameplay isn't any better.

Posted by: PJ on July 1, 2003 01:50 AM

This reminds me of the loopy Julian Simon / Bjorn Lomborg theory that since a line can be infinitely divided, we'll never run out of resources. The theory doesn't really work with physical resources like copper or oil, much less air or water.

But "storage space" really is more like "information" or points on a line than it is like a physical substance that can be heaped in piles and sold by the pound. There's still some physical limitation, but it's not like we're even close to being able to some theoretical limit whereby each microgram of matter can only store "x google" bits of information. The points-on-a-line analogy works better there.

Posted by: zizka on July 1, 2003 07:38 AM

And you can already buy drives that operate at 15,000 RPM.

Sort of like owning your own Formula One racing engine in your computer.

Posted by: David Glynn on July 1, 2003 08:57 AM

Try Google(TM) on 'pixie dust' to get the latest news on the data density front.

Posted by: Matt on July 1, 2003 10:32 AM

Our 19 year old central air compressor fried itself a few days ago, and today I was looking through our collection of old warranties, manuals, etc. to see what size it was. While looking, I came across a sales slip from ComputerLand (remember them?) from 1986. It was for a 10 Megabyte HardCard, a fixed disk that was installed as an I/O card in one's IBM PC. It was one of the most liberating pieces of technology I ever bought. No more floppy disks for data storage! it was wonderful. It also cost over a thousand dollars. That's $100/MB. If a 80 gig drive is 80 bucks now (and I have no idea if that's actually the price, but it's probably close, the retail price of a MB of much faster storage is now $0.001. Such a deal!

Posted by: John Casey on July 1, 2003 12:33 PM

John, that's an excellent data point...

Storage per dollar has been doubling every year for as long as I've been following it. Divide the 100,000 dollars per gigabyte by 2 to the 17th power (for the 17 years) and it comes out to 76 cents per gig, which is exactly right.

Here's an example of 78 cents per gig (120gb for $94... and that includes shipping and inflation :-) http://www.newegg.com/app/viewproduct.asp?DEPA=&submit=Go&description=SpinPoint+120GB

Processing power per dollar has also been growing by 2x per year for as long as I can remember, except sometimes the raw power available does not get used efficiently.

I don't know if this article means to say storage will improve faster than this or processing power will improve slower, but 2x per year has almost been exact, and it is very convenient too.

Posted by: snsterling on July 1, 2003 04:09 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:11 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:14 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:14 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:14 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:17 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:17 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:17 PM

Michael Lesk's http://lesk.com/mlesk/ksg97/ksg.html "How much information is there in the World?" is worth reading.

Posted by: rdb on July 1, 2003 09:20 PM

Apologies for the multiple post - mozilla tries
hard!

Posted by: rdb on July 1, 2003 09:40 PM
Post a comment