IBM introduced the hard drive (RAMAC) in 1956; it stored 5 MB in an enclosure the size of a refrigerator, and cost $150 thousand. In the next 45 years, hundreds of companies entered the industry and went tits up or were bought out as new formats drove out the old; by 2000 only a handful of companies were in the business. IBM soldiered on, responsible for almost every significant breakthrough (including giant magnetoresistance, for which Stuart Parkin was unjustly denied the Nobel in physics in 2007. But it was losing money on every unit an industry with razor-thin margins and huge capital costs. I was long gone from the Valley by the time Hitachi bought it out, and I don’t know why they thought they’d could make money at it. Turns out, they couldn’t.
So, today you can get a 1 TB drive for $300 . Per MB, adjusted for inflation, that’s about a billionfold improvement in value—and that’s not accounting for huge improvements in data-transfer rates, power consumption and, umm, portability. (That’s an IBM-developed hard drive in your 80 GB iPod you listen to on the plane; here’s a RAMAC being put onto a plane.)
The “end of the hard drive” (when data density reaches its physical limits) has been five years away, for the last fifteen years. It still is. At IBM Almaden, while I was basically a grease monkey working on disks and motors, my data-storage colleagues were working on holographic materials and atomic-force microscopy. The HDD guys are gone, but the mad scientists are still there, and I’ll bet that one day you’ll have them to thank when you have the MGM Films library, in fully-immersive 3-D, on your 1 PB iPod.
« The logic of collective action | Main | How to lie (to yourself) with statistics » The Reality-Based Community: Byte me26 Feb 2008 05:56 am Comments (4)Comments on this entry have been closed. |
Today's Headlines From The Atlantic |
Copyright © 2009 by The Atlantic Monthly Group. All rights reserved.
It's not just the disk drives. When I started at a certain Fortune 500 company in 1977, they had just booked very good results for the previous year, so they wrote two checks for $5 million each.
One bought 5 Megabytes of RAM for their IBM 370 mainframe. This wasn't the genuine IBM memory, which was more expensive yet, but was OEM memory from Control Data Corporation.
The know nothings spouted off about how stupid the programmers were back in the 1970s and 1980s not to record the entire 4 digits of the year in every record. They have no comprehension of the effect of technological improvement on the IT industry. Anyone who wasted all that space would have been fired for incompetence.
The other check bought a business jet. I'm not sure if it was the Gulfstream or the Lear.
I don't know where the author gets his "we've always been five years away from the physical limits" schtick. It was initially put much further away and we have been approaching it monotonically - the usual 'physical limit' in this case is no more a bit or so per atom.
Still a lot of room and a lot of hardware to play with, but enough with the rah-rah already.
Megan,
I think IBM is not leading anymore but following.
http://colossalstorage.net
The author here. I get my "physical limits schtick" from having been an HDD R&D engineer. There are all sorts of physical limits well short of one bit per atom. In particular, the superparamagnetic limit, at which magnetic grains spontaneously flip polarity at room temperature. That limit wasn't known exactly, and has been pushed lower for years now. See tinyurl.com/2h875w
As for colossalstorage.net, care to buy some prime land, just 100 miles west of Miami?