New physics research out of the University of California Santa Cruz
(UC Santa Cruz) offers new insight into a poorly understood aspect
of
disc drive failures, according to industry experts, though they
say the information won't necessarily make disc drives more
reliable.
"There's been a misconception since this research was published
that this is why disc drives fail, or that this is a new reason for
drive failure unknown to the industry," said Josh Deutsch, who
co-authored a paper that appeared online in the 13 July issue of
Physical Review Letters. The research could theoretically
help disc drive manufacturers find more reliable materials for disc
drives, Deutsch said. He emphasized though, that disc drive
manufacturers have been using materials with good damping
capabilities for reliability all along -- it just wasn't until this
research that the reasons were fully understood.
He added, "this is far from the only reason drives fail -- it
shouldn't be misconstrued as such."
The research came from a chance meeting at a lecture several
years ago, between Deutsch, a theoretical condensed-matter
physicist and a professor at UC Santa Cruz, and Andreas Berger, at
the time an engineer for Hitachi Global Storage Technologies
(HGST). It was that chance meeting that steered Deutsch's research
in magnetics toward disc drives in particular.
According to a press release on the research, for each bit on a
disc, the magnetic disc head grazes a tiny patch of the drive,
forcing its polarity, or "spin," to align up or down, the magnetic
equivalent of a one or a zero. The patch's polarity in many
magnetic materials changes in a series of jumps that physicists
liken to an avalanche.
Deutsch and Berger discovered that such an ideal model
overlooked an effect, called spin precession that each magnetic
field exerts on its neighbors, causing bits to swing in circles
rather than flipping totally up or down, an effect that can cause a
chain reaction of "wobbly bits" that wipes out a sector of a disc
drive.
Better reliability vs. better density
However, while it's easy to equate this new insight with a
breakthrough that will address
recently uncovered discrepancies between
manufacturer ratings and disc drive reliability, one of the
authors of Carnegie Mellon University's drive reliability
research, Garth Gibson, an associate professor at Carnegie
Mellon, and chief technology officer (CTO) and founder of
Panasas Inc., (Ed. note: When does he sleep?), said that
since damping materials have always been used in disc drives,
though until now their precise mechanism was not as well
understood, the effect of these materials on drive reliability
has already been taken into account in current statistics.
Meanwhile, "the single most important property of a magnetic
disc drive is its cost," Gibson noted. While the new research into
magnetic avalanches could be used to make drives more reliable
through the use of better damping materials, Gibson predicted what
will actually happen is that drive designers will use it to figure
out how to get bits closer together with the same failure rate.
According to Gibson, disc drives are already a delicate balance
between reliability, performance and affordability. A significant
uptick in one of those categories has the potential to throw the
others off balance. "There's an analogy here with the auto
industry," Gibson said. "If we all drove around in tanks, fewer
people would be hurt in car accidents, but that, of course, isn't
practical." Better damping materials tend to be more expensive, and
while it's possible some consumers are willing to pay more for more
reliable disc drives, "historically, people want a low price more
than anything else." Moreover, Gibson said, expensive improvements
to just one relatively minor aspect of how disc drives operate
probably won't be judged cost effective when there are so many
other factors in disc drive failures.
Instead, he said, look for this research to come as a godsend to
disc drive engineers tasked with reaching certain thresholds of
density within set product roadmap time frames, who "often have no
idea how they're going to get there more than a generation
beforehand. Engineers are given an orderly process [for product
development], but somewhere along the line, someone still has to
make a [scientific] breakthrough."