From this week's The Economist;
economist.com
SCIENCE AND TECHNOLOGY
A byte of the action
Today's magnetic-storage technologies are expected to reach their
capacity limits by 2004. The race is on to figure out what comes next.
PITY the poor engineers who build hard-disk drives for computers. For years
they have worked miracles to squeeze ever more data into ever smaller and
cheaper packages. Yet they are forever being upstaged by their more glamorous
colleagues in the chip-making business.
Perhaps that is understandable. Computer buyers, like car buyers, tend to be
more interested in speed than luggage space. But in many ways the
achievements of the magnetic-storage industry-a business which, according to
Disk/Trend, a consultancy based in Mountain View, California, is currently
worth $34 billion a year-are the more impressive. The amount of data that can
be squirrelled away on a square inch of disk surface (the industry clings
obstinately to imperial measurements) has increased by 60% every year since
1991. Over the same period, the cost of storage capacity has fallen from more
than $5 a megabyte to a mere five cents.
Earlier this month IBM announced the latest downward twist in this virtuous
spiral: "Microdrive", a disk that packs 340 megabytes of data (roughly
equivalent to 340 copies of the 445-page Starr report) into a cartridge smaller
than a book of matches. This is made possible by the latest "giant
magneto-resistive" (GMR) technology, laboratory versions of which can store as
much as 1,450 megabytes a square inch.
However, despite such continuing triumphs of storage wizardry, the end is now
in sight. That end is the so-called superparamagnetic limit. It is a consequence of
the way that data are recorded on a hard drive.
The zeros and ones of digital information are preserved as tiny spots organised
in concentric tracks on the surface of a rapidly spinning disk coated with a thin
film of magnetic material. The alignment of north and south magnetic poles in a
spot tells a reading head whether the spot is a zero or a one. Alternatively,
running an electric current through the head can alter the polarity of a spot, and
thus the message that it carries.
Over the years, engineers have increased storage densities by making these
magnetised spots smaller and then packing them more closely together. GMR
helps by detecting the effect of the spots' magnetism on the resistance of a tiny
wire in the reading head (previously it has been the current induced in the wire
by a magnetic spot moving underneath it that has been detected, a less sensitive
process).
Other improvements have included new, more powerful magnetic-disk coatings;
changes in the shapes of the spots; moving the head closer to the disk's surface
so as to see the newly miniaturised spots more clearly; and clever
signal-processing techniques to detect the spots' ever-tinier magnetic
fluctuations. Ultimately, however, these shifts are merely postponing the
inevitable-the point at which the spots become so small that they are
magnetically unstable, and thus liable to flip their north-south polarity at random
in response to heat-induced vibrations. This is the superparamagnetic limit. Pass
it, and any recorded data will rapidly be lost.
Bob Katzive, Disk/Trend's vice-president, believes the limit will be reached by
2003-04. Admittedly, disks should by then have achieved a capacity of some
8,000 megabytes (8 gigabytes) a square inch. But experience suggests that the
demand for storage will keep growing indefinitely. Overcoming the
superparamagnetic limit is therefore a matter of pressing concern.
Disk driven
One possible way forward is to borrow from the field of "magneto-optical"
(MO) storage. This works on a different principle from straight magnetic storage,
for although the information is recorded magnetically, it is written and read by a
laser.
MO disks are coated with a material whose polarity can be altered only at high
temperatures. Information is written on to the disk by using the laser to heat a
tiny spot on its surface and then applying a magnetic field. As the spot cools, it
adopts the field's north-south polarity. Retrieving the stored data involves
bouncing the laser beam (at a lower power) off the surface of the disk. The
magnetic polarity of a dot affects its physical characteristics enough for the
reflected laser light to differ, depending on whether it is bouncing off a north or a
south pole.
Because the dots are magnetically stable at room temperature, MO technology
should ultimately be capable of holding more information than conventional
magnetic storage. Packing the dots more closely then becomes a matter of
optics: focusing and guiding the laser beam.
One such scheme has been developed by TeraStor, a recently founded
company based in San Jose, California. Its main technology is called near-field
recording. In this, the head is fitted with a special lens, and is brought extremely
close to the surface of the disk. The result is a tightly focused beam, and thus a
tiny spot.
The first product to use this technology-a removable disk with a capacity of 20
gigabytes-should be available next year. After that, TeraStor believes its
technology will be able to keep up the 60% annual increase in storage density
that the industry has come to expect long after conventional hard disks hit the
superparamagnetic limit. According to Gordon Knight, the company's chief
technical officer, storage densities as great as 37.5 gigabytes a square inch
should be possible. That would allow a single disk to carry the magic figure of a
terabyte (1,000 gigabytes).
A different approach is being taken by Quinta, another San Jose startup. Its
Optically Enhanced Winchester scheme also involves MO. The vital twist is the
use of a tiny mirror which can be rotated with great precision by passing an
electric current through it. This mirror directs the laser on to the surface of the
disk.
Such fine control of the beam will, according to Quinta, allow the tracks of data
to be packed more tightly together on a disk than would be possible if a
conventional motor were used to position the laser. That should make storage
densities of up to 31 gigabytes a square inch possible.
The outside chance that one of these two companies has the answer for the next
generation of mass storage has already attracted backers from the established
hard-disk industry-despite the fact that neither has yet demonstrated that its
approach can significantly outperform traditional magnetics. TeraStor is backed
by Quantum, a leading disk-drive manufacturer. Last year Seagate, another
drive manufacturer, paid $325m for Quinta.
Not everyone, however, is convinced that either approach, snazzy though each
is, will prove profitable. One sceptic is Currie Munce, director of
storage-systems technology at IBM's Almaden research centre. He says that
IBM has already evaluated and rejected the near-field recording techniques now
being championed by TeraStor. He also suggests that conventional
magnetic-disk technology could hold out longer than expected, reaching storage
densities as great as 12.5 gigabytes a square inch. By that time totally new
technologies that IBM is working on-such as holographic cubes-may be
available.
But even IBM is not infallible. Its early support for Microsoft was ultimately
responsible for that company's massive "bloatware" operating systems, which
now run most personal computers, and which have helped to fuel demand for
huge disk capacities in the first place. Mr Katzive says that in the geek humour
of hard-disk engineers, the NT in Windows NT stands for "Needs a Terabyte".
He is only half-joking.
|
