100x Faster Hard Drive In Lab

Gary lets us know about research out of the Netherlands that has succeeded in reading and writing a hard disk using polarized laser light. The researchers claim this offers a 100-times speedup over reading/writing using magnets. People have been trying for years to write data using polarized light; the secret of the current work's success lies in its disk's materials — gadolinium, iron, and cobalt. Working prototype drives should be available within a decade.

Comments

A decade?

[ScrewMaster]

Working prototype drives should be available within a decade.

Spare me. I've been hearing about incredibly dense optical storage for thirty years now. I have yet to see it.

Re:A decade?

[janrinok]

Haven't you seen the developments in CDs and DVDs during that last 30 years? Everybody else has! A DVD is an incredible amount of storage when compared to the 5MB (yes MB!) hard drive or even my cassette tapes that I was using in the late 70s.

Re:A decade?

[ScrewMaster]

I don't know why you think 1.6 TB makes a particularly dense storage medium. It doesn't, and I hope that they're working on a serious contender because our civilization's needs for information storage are increasing at a decidedly non-linear rate. Most of the PC-using population doesn't have much use for more processing power right now, but we can all use a bigger hard drive. Hell, the various up-and-coming surveillance societies alone are going to drive the need for more bits as time goes on, and if digital content delivery really takes off 1.6 TB is going to seem puny.

In any event, there have been many promising technologies in the past few decades that have fallen by the wayside. These guys are saying we'll see their ideas commercialized within ten years? YEARS? That means they have, at best, a proof of concept and have a whole lot of research and engineering ahead of them before any products actually ship. Odds are, their approach won't prove commercially viable for one reason or another, and this will just be another scientific footnote.

As it happens, I was referring to the so-called "holostore", which was being billed as the next great thing in optical storage in the early 80's. Well, that was when I first read about it anyway (Scientific American, as I recall) and it was supposed to store data in a three-dimensional crystalline format that would be read and written by multiple laser beams. They were promising that several Libraries of Congress could fit on an inch-high cube and that it would be available "in a few years." Well, it's been over a quarter of a century now. I understand that it's a work in progress ... but I still can't buy one.

So, like I said, I'm still waiting. I want to see cheap desktop storage that's a couple orders of magnitude more than we have now. Then I'll say they've fulfilled their promise. Avery Brooks is waiting for his flying cars ... he can have them. I'm still waiting for some real storage. I have a lot of videos I want to keep.

Do we even have to say it?

[Dirtside]

I think this story is a dupe from, like, 1993.

Seriously, I can't think of an otherwise plausible tech that's been vaporware longer than light- or holography-based data storage. I know there have been working examples for years, and I think there's even a (really, really expensive, very specialized) production version or two, but come on! How hard can this be?

Stupid hype

[imsabbel]

Actually, this couldnt have less to do with data storage (you cannot really focus your femto-second laser down to spotsizes lower than what we currently have in HDs, plus there is no real way for a femtosecond source that not bulky, wastefull and expensive).

On the other hand is the switching of magnetic domains by the polarity of a circular pulse an archivement in itself. But of course fundamental research doesnt interest anybody, so they have to create a stupid "next storage medium" out of it.

No, it isn't really 100x faster

[geophile]

The researchers managed to transfer data at intervals of about 40 femtoseconds, or quadrillionths of a second, about 100 times faster than conventional magnetic transfers

That optimizes a tiny part of the problem. There are two mechanical issues, 1) waiting for the right part of the disk to rotate under the read/write head, and 2) arm motion. Without eliminating one or both of these delays, I don't see how this leads to faster secondary storage access in practice.

Re:No, it isn't really 100x faster

[janrinok]

Which is probably why they said that it will take a decade to produce usable devices. However, that doesn't detract from the discovery or achievement. It is another hurdle passed which will let someone else concentrate on solving the other problems.

Re:No, it isn't really 100x faster

[scotch]

You don't even need a mechanical mirror mover - you can direct and refocus light very quickly using solid state tricks with LCDs that modify their refractive index locally.

Yeah, this is great, but....

[Jamey]

Except that it really doesn't help that much!

Hard drives have gotten bigger, and bigger, and *BIGGER* over the last 20-30 years. But they don't *FEEL* that much faster. They've become wonders at streaming huge blobs of contiguous data out - so why do databases need huge steaming bloody chunks of RAM cache? Because the random access times *SUCK* and really haven't gotten that much better!

Capacity has gone from 5MB to 1TB, but spindle speeds have gone from 3600RPM - up to a max of??? 15K RPM for some really expensive drives? Track-to-Track seek hasn't gone up much. Neither has real nor manufacture's claimed throughput rates.

RAM hasn't nearly kept up with CPUs, either, but the disparity is nothing compared to the hold you get when you have to go after some data from the hard drive that isn't in the cache.

It's so bad, I strongly considered putting 3 4GB FLASH modules with IDE adapters (RAID5 - but I didn't study this to see if 2 8GB with RAID1 might be better, or other variations) into my new machine on the PATA header to act as the root drive, holding everything but /home, /var, and /tmp.

Sequential read speed is kinda nice, but I *do* need to do random accesses sometimes! I listen to my nice little 2TB RAID array all the time, as the heads move back and forth singing their little song.

Bah HD speed

[silas_moeckel]

I can tipple the transfer rate and reduce the average seek time by about the same by using 3 sets of heads. Oh you wanted something thats cost effective please move along. Really though I do not know why they could not use multiple servo motors to at least split the heads already on server class drives, any hardware geeks want to chime in? It seems there is a big push for 2.5 inch SAS drives I cant see why you could not stack some of those platters in a 3.5 and add extra heads and controlling gear? Sure your not speeding up single transfers but your cutting the rotational latency in half and allowing multiple operations at once great for servers.

Where's my flying car?

[binaryspiral]

Where's my flying car? Damn it - it's still in the labs.