Slashdot Mirror
The Story Of GMR Heads
lopati writes "The story of GMR heads, "the breakthrough that boosted the capacity of hard-drives from a few gigabytes to 100 gigabytes and more--came from chance observation, basic research and a vast, painstaking search for the right materials." Check out the helpful infographic." Background: This is a story, essentially, about how hard drives broke through some of the space limitations at the beginning of the 1990s - pretty cool background.
Comment removed based on user account deletion
IBM's major problem was that, although they were able to scale down the GMR head very easily, they had large stocks of old media that was not certified for use on GMR drives. (Incidentally, most of that media is in an enormous warehouse in Hungary, which is where most of their drives are produced now.) They designed a recertification process that was supposed to allow them to separate the media that would be suitable for the 75GXPs from the media that wasn't suitable, but that process was deeply flawed and this resulted in the high failure rates of their drives.
You may find it a bit odd to be hearing this from a former Maxtor employee. Well, the dirty little secret of storage companies is that reverse engineering is rampant. My colleagues at Maxtor probed, disassembled, and tested the IBM drives; indeed, they might have known what the bug was even before IBM did.
So, the obvious RISK of GMR technology is: do not use platters that are not certified for use with the new heads. Those who disregard this creed are certain to meet with a nasty public relations disaster in due time.
freebsd guy
"the breakthrough that boosted the capacity of hard-drives from a few gigabytes to 100 gigabytes and more--came from chance observation, basic research and a vast, painstaking search for the right materials."
;)
In summary, the guys at IBM ran out of HD space for their um, 'special files'?
One of my physics profs, Yumi Ijiri, moved to my school after doing a few years of research for NIST and IBM regarding GMR technology. Basically, noone could figure out why GMR worked, or how to systematically improve upon the concept. IBM found a neat combination of thin films created these extremely sensitive magnetic sectors, and instead of finding out why/how it works, they empirically tried some 22000 or so combinations until they progressively found better and better arrangements. After the fact, they hired Yumi to figure all the physics out, but her research was also inconclusive. It kinda scares me that there's stuff in my harddrive that IBM and NIST couldn't figure out after 4 years of research.
This is where I get my recommended daily allowance of "Foot in Mouth."
I suppose this means that GMR technology is "sufficiently advanced."
Yeah....there will come a time, probably within our lives (maybe 20 years), when a $200 hard drive will be able to hold every movie, song, and book ever created. How do you fill that one up? Well, when they get that big, there might not be enough of a market for that much storage so the price will go up. But...
(For the pedantic, my argument rests on the fact that in 1992, a 100 megabyte HD cost about $200, and today, a 100 gigabyte HD costs about the same (give or take). At the same rate, we'll have 100,000 gigabyte in ten years, and 100,000,000 gigabyte in 20. Physics blah blah blah.)
At DVD-size of ~2.5 gigabytes per movie, uncompressed music at about 40 mb/song, and books at (generously) 20mb/pdf-book, this makes (in 20 years):
400,000 movies, -or-
2.5 billion (10^9) uncompressed songs
5 billion books.
(Please don't flame me if my math is wrong--just correct me politely). Unfortunately, I wouldn't be surprised if in 10 years, most people are still using 56k dialup and 4 gb? DVDs. Again, I ask you, how are you gonna fill up that disk?
But, I'm not good at predicting the future of hard drive storage. In 1989, I had a big argument with a buddy about hard drives. My contention was that nobody would be able to use more than 30 (well, maybe 40) megabytes of hard drive space.
Here's a link to one of my posts on the Spintronics slashdot article a few weeks ago. I think I posted it a few hours too late for most people (moderators included) to notice it.
Explains basics of GMR, which is based on magnetoelectronics, or it's catchier nickname Spintronics. Also related to GMR are the non-volatile RAM's commercially available now.
Cool part is that GMR devices were commercially available only a few years after discovery in the lab. That's an accomplishment usually reserved for potentially ground-breaking devices (ie, transistors). T'will be very interesting to see how this field progresses in the future.
make world, not war
With all this great technology, I wonder why these larger capacties are only available on IDE drives.
It seems to me, SCSI drive capacities used to outstrip IDE by quite a bit, and the price penalty wasn't all that much (~$200). Lately, all I see in the catalogs for SCSI is 18GB or 36GB, while IDE is at 80GB, 120GB, and even 160GB.
Is there something about this technology that isn't compatible with SCSI, or does SCSI not scale well, or what?
In the USA, we like stuff watered down, like beer, television, and freedom.
My family's been in MR tech (well, magnetic storage) for over 30 years now. I worked 3 years in IBM's MR head manufacturing facility in San Jose (Cottle Rd.). It used to be that the substrate people (the ones who made the actual disks) didn't have much to do because the MR heads could not write small enough to pressure them.
GMR heads caused quite the stir because they could write smaller than the substrate had resolution.
Now IBM's "pixie dust" has swung the pendulum the other way, as the head is once again the bottleneck.
An interesting tid-bit is how many production managers were hired away from IBM soon after GMR heads were released.
I'd rather have someone respond than be modded up.
We were going through a surplus catalog that was trying to unload some 100M drive units. We would go on about how to cool those things in his room and just what kind of kick-butt BBS we could have with THAT much space. Assuming it was even possible to get the C=64 to talk to the HD unit.
You start combining the elements of course. 100x99x98/6 is about 160,000, which is the number of different combinations of 3 elements you can have. But then you can also continually adjust their relative concentrations - A B_x C_y allowing x and y to be any number between 0 and infinity - in practice you might sample at 10 different points in x and y to get a rough idea: that's another factor of 100, so about 10 million ways to combine three elements just in terms of chemistry. Go look at alloy phase diagram books for a sample of the complexity you can get combining three metallic elements into alloys. And why stop at 3 elements? The high T_c superconductors take 4 or 5 or more to work.
:-) And so difficult :-(
But this isn't just chemistry either - the material is nonuniform, layered. Each layer can be composed of some different magnetic or non-magnetic alloy, and each layer can have a different thickness, and the number of layers is itself a variable. The combinatorial possibilities are in the billions! Obviously they narrowed it down considerably to find what they needed in just 30,000 samples - but there may be something even more spectacular out there among the billions of other possibilities, just waiting to be found.
That's what makes science these days so interesting
Energy: time to change the picture.