Is Laser The Solution For Hard Drive I/O?
Page 1 of 1
Researchers at a Dutch university have submitted work on a laser-based hard drive head that they claim can run up to 100 times faster than current magnetic heads, but the viability of the technology is questionable.
The researchers from Radboud University Nijmegen have published a paper in the Physical Review Letters on a new laser light method to write data at a rate of 40 femtoseconds. A femtosecond is one-millionth of a nanosecond, and hard drive speeds are typically measured in milliseconds.
The technique works by using high speed pulses and reversing the polarity of the laser lights to produce the equivalent of a 0 or 1 to write the bits to the platter. As it does this, the laser heats up the area of the disc just enough to make changing its polarity easier for disk writes.
One drawback is that the laser pulse is much wider than the read/write head of a hard disk. The wider the head, the less data can be packed onto a disk, and Krishna Chander, senior analyst with iSuppli, said capacity is what's important.
"The industry is always going for capacity increases as opposed to performance increases," he told internetnews.com. "Not that they don't need performance increases, but the first problem they are always trying to solve is capacity."
Increased capacity via heat is what storage giant Seagate is working on as well with a project called Heat Assisted Magnetic Recording (HAMR, pronounced "Hammer").
HAMR focuses an area of the disc, heats it up, changes the properties of the platter so it's easier to write, then cools it down, so it becomes stable and data stays there. Seagate has been working on this for almost seven years, which should be some indication of how hard it is.
Heating and cooling metal platters on a hard drive may seem like a slow process, but HAMR can heat and cool these spots in pentoseconds, according to Mark Re, senior vice president of research at Seagate. Still, he said read/write time isn't the issue for performance.
"The limitation to speed of hard drives is the interface, not the read/write mechanism," he said. "[Lasers] are a nice way to do reader/writes because it removes process steps from the drive head, but the head isn't the main limiting factor for speed, the bus is."
Chander agrees. "This [laser read/write heads] is addressing the internal mechanism inside that drive, but the data that comes out still has the limitation of the interfaces, and they haven't overcome that," he said.
Both Re and Chander questioned the viability of the technology, not to mention its long development stage. According to Daniel Stanciu, co-author of the paper, he expects to see a working prototype within a decade. As it is, HAMR is still five years away. "At this point it's really cool physics," said Re.
Chander was a little more blunt in his assessment. Aside from the fact that capacity is more of an issue and not I/O, the Radboud project is good for disk writes, not reads, and he said on average, a disk spends one third of its time writing and two thirds of its time reading. So the project solves a minority problem.
"I don't know why they are trying to do this other than it may be someone's PhD project or something," he said. (Stanciu is a graduate student at Radboud.) "We're using blue laser systems that haven't improved in capacity or performance. Given laser has all that baggage, I don't know where these guys are heading."