Storage Barns in a Thimble
In June, 1956, IBM shipped the world's first commercial magnetic hard-disk drive. As big as a garden shed, the unit used 50 magnetic disks spinning at 1,200 revolutions per minute. Together, the disks could hold five megabytes of data. IBM rented out the contraption for tens of thousands of dollars per year.
Digital storage has come a long way since. Magnetic disk drives consisting of a single coated disk now hold 100 gigabytes of data -- 20,000 times what the IBM garden shed did -- and spin 10 times faster, allowing personal-computer users to retrieve data in a flash. Modern drives no more than a few inches across fit inside the thinnest laptops and cost hundreds of dollars instead of thousands.
Stacked arrays of these disks housed in specialized storage computers can hold terabytes of data, a million-fold improvement over the original IBM machine. Such miniaturization is possible because over the past five years, storage capacity per square inch on magnetic drives has soared 100% to 150% annually. Since the original hard-disk drive, "we have increased the density of storage per given unit of area by 20 million times," says Currie Munce, a research director specializing in storage systems at IBM's Almaden Laboratory in San Jose, Calif.
THE BREAKING POINT.
Looming on the horizon, however, is a physical barrier that could halt the advances in magnetic-disk storage. Storage researchers have a fittingly large name for it: the superparamagnetic limit. They theorize that when disk drives become so small that the concentric magnetic tracks that hold data in positively and negatively charged sections are excessively compressed, the systems will cease to work. The energy required to maintain each tiny magnetic section -- just millionths of a millimeter wide -- will be equaled and canceled out by the surrounding heat energy inside the drive.
When that happens, the bits of magnetically charged data will start to flip randomly from one spot on the drive to another, destroying the whole system's integrity.
Originally, researchers thought a drive's top storage capacity was about 33 gigabits -- 33 billion bits, the smallest information unit on a computer -- per square inch of density. (In terms of density, disk-drive capacities are measured in bits, while overall storage capacities are measured in bytes, where 8 bits equal 1 byte). Commercial drives have already hit that level and are set to surpass it in the near future. Today, researchers peg the superparamagnetic limit north of 100 gigabits per square inch, a capacity that may show up in your PC within the next few years.
THE NEXT AGE.
This limit is for real, according to lab research. "The volume is small enough that we're seeing thermal fluctuation of the bits," says Robert White, director of the Data Storage Systems Center at Carnegie Mellon University in Pittsburgh.
Scientists aren't ready to concede, however. In fact, they're already developing new technologies for what might be called the post-superparamagnetic era. Most likely to hit the commercial mainstream first -- probably about 2004 or 2005 -- will be perpendicular recording. Current disk technology uses longitudinal recording, where the positive and negative charges of each piece of stored information are parallel to the disk's surface and exist in the same plane. In perpendicular recording, the charges will be perpendicular to the surface.
Perpendicular recording will allow the heads that read and write informattion onto and off the disks to handle more tightly packed data with less risk of thermal distortion. It will still ultimately be subject to the superparamagnetic limit. But flipping the recording orientation from horizontal to vertical could allow scientists to push that limit out to 1 terabit, or 10 times the currently presumed limit.
Researchers say perpendicular recording still has a ways to go from lab to fab, however. "It's a big transition for the disk-drive manufacturers. It will mean a change in head design because now you're reading the media not left to right, but up and down," says David Reinsel, a magnetic storage analyst with Framingham (Mass.) research firm IDC.
The next step researchers see beyond perpendicular recording is heat-assisted magnetic recording (HAMR). This will use a magnetic recording medium -- such as a carbon compound riddled with tiny cobalt formations -- that holds magnetic information more tightly and is more resistant to thermal distortion. That translates into the ability to pack data more closely together on a disk. Scientist say it could ultimately take storage to levels of 50 terabits per square inch or beyond.
The catch for now is that HAMR materials are so magnetically stable that existing mechanisms used for writing data to commercial hard-drive disks won't work. However, if the surface of HAMR disks are zapped with a laser at the same time that the head is recording information, the heat from the optical blast may allow data to be written. "If we heat it selectively where we want to write, the material becomes magnetically soft where we can change it. Then it cools back down to room temperature, which makes it very stable," says CMU's White. This would happen in tiny fractions of a second each time someone stored or called up information on a PC.
While perpendicular recording is largely a known quantity that could be mass-produced, HAMR developers still face the question of how to manufacture their technology. It's hard to keep the heat from the laser concentrated in a small enough area to prevent the thermal energy from disrupting other sectors of the disk. And the added time required to heat up the disk surface could slow retrieval and storage.
STORAGE IN 3-D.
Disk-drive maker Seagate, based in Scotts Valley, Calif., has already shown prototype HAMR devices, but nothing is ready for the factory yet. Experts are betting that HAMR will be commercially available within the next five years.
Researchers are also doing blue-sky thinking about technologies that might step to the fore once magnetic recording has reached its limit. One is holographic storage. IBM and a handful of startups are working on storage systems that use three-dimensional light pictures created by shining a tightly focused laser on a specific point on a piece of polymer. The theory is that the pixels of the three-dimensional image would represent digital "1s" and "0s" in the same way that magnetic charges do on a hard disk.
Adding this third dimension would make a huge difference. The laser beams used in holograms can be finely tuned to recover different data images from slightly different depths within the polymer media, meaning data images can sit on top of each other, dramatically increasing potential capacity.
Also, holographic data retrieval can occur simultaneously instead of sequentially, just as the human eye can take in an entire image at once. By comparison, standard magnetic disk drives can see only thin strips of data. With holograms, "we just don't use the surface of the material. We use the whole volume of the material," says Nelson Diaz, the CEO of InPhase Technologies, a holographic-medium and -device startup based in Longmont, Colo., that was spun out of Bell Labs two years ago.
The idea of holographic storage has gotten a big lift lately from advancements in digital cameras. The same technologies used to capture images and then process them into digital information also work well in holographic storage. The main obstacle here, however, is that existing polymer materials can't be written over multiple times. That makes them best-suited for permanent archiving, which is needed by videographers, hospitals, and banks, and that's where holographic storage will likely catch on first.
Another possible use for this technology is in read-only cartridges for video, software, or games. InPhase is set to release its first product, a removable read-only holographic drive system that can hold greater than 100 gigabytes of data. Each holographic disk will cost $40 or $50. "A consumer holographic product [for those purposes] could be available in the 2004 time frame," says Diaz. "I know people who are far enough in development to do that."
PROBING FOR DATA.
Just when a new form of rewritable media for holographs might arrive remains anyone's guess, though. "The progress has been limited, and the issue of this material's limitations has been the same for 20 or 25 years," says IBM's Munce. The holographic devices that'll start to appear in the next few years won't outperform existing storage equipment enough to replace it, at least not in the short run.
On the far, far horizon is one other approach -- dubbed probe technology -- with the potential to revolutionize storage. This involves using tiny actuator arms, thinner than a human hair, to read and write peaks and valleys on molecule-size disks. The big plus of probe storage is that it's done on a molecular scale, allowing for the storage of huge amounts of data in extremely small spaces.
The downside is that tiny machines based on such technology might be far more susceptible to electrical and thermal disruptions than today's disk drives. And most of the equipment that's now used to move around molecules requires lots of hand-holding. Also, it'll be difficult to shrink the technology required to move molecules enough to make it appropriate for commercial use. Another question is: How to build systems that can translate data that's stored on a molecular level into pictures on a computer screen? For these reasons and more, probe technologies may never take off.
You might wonder why such feverish research is being done in this field, considering that demand for magnetic recording devices has slowed lately. After all, today's hard drives are so spacious that companies and individuals seldom fill them up. Indeed, for the immediate term storage makers are focusing on delivering faster playback rather than enhanced capacity. And that could slow the pace of innovation for the near future.
Still, demand for advances in storage technology could ratchet up quickly if new consumer technologies such as personal videorecorders and digital video editing take off in the next few years. Once that happens, the superparamagnetic limit will seem no more daunting than the sound barrier did after Chuck Yeager proved it was just another milestone on the way to something far bigger.
By Alex Salkever, Technology editor for BusinessWeek Online