Slashdot Mirror
1TB In A Cubic Centimeter
rgetty writes "Inforworld posted this article describing the process used by a group of engineers and scientists from Kyoto University and Central Glass (Japan) to pack 1TB of data into a cubic centimeter of glass. Portable data warehousing is not too far off..."
The lab I work in uses nanosecond lasers to create BEC's. Down the hall though, is a faculty member who works on femtosecond spectroscopy. As someone guessed, the space required for all the equipment is very large.
It _does_ take up a room, or at least the optical bench does.
But more importantly, in order to generate femotosecond pulses, you need _enormous_ bandwidth (10^15 Hz). These are class 4 lasers that are extremely powerful and also extremely dangerous.
I can't imagine this technology being anything resembling "out of the box" anytime soon --- you'd need an entire support staff just to use it.
I doupt that. Even at a TB per cubic centimeter you aren't gonna walk out with all their data. Last month I was helping one company that was creating 3 TB of new data a day. Add up a few months of that, and you byond pocket size. This wasn't even a really big company, I've worked with some that do 20 TBs a day.
I won't even mention NASA and the like that can do a few 100 TB a day. (most of which they don't process). And speculations of what the NSA (echolin, however it is spelled) can get in a day aren't worth it. Though the latter is important to consider.
Clerk: Hi! Welcome to Fry's! Can I help you?
Shopper: Yes, I'd like one of those 8T holographic cubes.
Clerk: Here you are sir. That'll be $300 for the cube... oh, and $18 million for the giant femtosecond laser. You cleared out a room where you can store this?
Shopper: Yeah, I decided we don't really need a kitchen.
I am, of course, exaggerating. You can't really help at Fry's.
"Do you expect me to talk?" "No, Mr. Bond. I expect you to die!"
...but how the heck are we supposed to back it up?
;-)
--
The gift of death metal does not smile on the good looking.
Seriously, this would be a boon for NASA. Currently, they're pushing the limits of backup technology and it's expected to get worse.
A "cube library" (as opposed to tape) with a little shuttle to move the cubes around would be a godsend even if the laser to read them costs 300k.
Picture one of these; the laser might take up most of it, but the savings would be incredible.
-- "I am disrespectful to dirt. Can you not see that I am serious!"
Sure, ten or even five years ago, media size was really important, but as bandwidth increases the information bottleneck is the cost of data storage, and the speed of data transmission.
A terabyte ina sugarcube is terriffic, but not because I'll be able to put a box that can read it on my desktop in 10 years, it's because I'll be able to control a couple hundred gigs on a server somewhere, or even better, everywhere (like OceanStore), because the cost of the hardware is distributed, much like the internet compared to dialup BBSes of the '80s.
One of the supercool things about the net is that I'm using the latest expensive hardware every day when my packets are routed through gigabit routers and fiber-optic backbones. I don't have to pay for it like I did the long-distance copper wire when I called BBSes across the country.
Storage will continue to follow the same trend, where the terabyte and exabyte drive complexes will serve my storage needs, and not some primitive box I plug into my computer and have to upgrade every year or two.
Kevin Fox
--
Kevin Fox
But how large are the femtosecond lasers, manipulation for the store, sensors to detect the florescence, firmware, etc. ????
My rough guess would be room-sized, at current technology. Give it 10-20 years, and then you'll likely see a unit to fit in a home computing unit (whatever THEY will look like by then.....)
It's all very impressive, but frankly 1TB isn't all that big a deal anymore. I'm currently working on project prototyping a data store for a 5yr satellite mission. It is currently estimated that the system will use 2PB of storage (2000TB). Probably several hundred TB of that will have to be online and the rest nearline. All of it will be remotly accessable.
-- Any statement of the form "X is the one, true Y" is FALSE.
Subject says it all, size is great, but how long does it take to get the femtosecond laser to focus on a different part of the cube to read other data? (not to mention *finding* where this data is stored)
;)
Also, what would the transfer rate be? And how is the data encoded? (CRCs etc. you wouldn't want a speck of dust to prevent you from reading a couple of gigs from the cube...)
I wish there was a more technical article published somewhere (hopefully not in Japanese
-- the cake is a lie
Ok... the house I grew up in is about 110 years old. Some of the windows are the originals. Perhaps some are apt to forget this but, glass is a liquid (albeit an extremely viscous one). Over time, it pours.
Old windows are like that NOT because the glass has "flowed" down over years... but because the methods used to manufacture windows at that time created ones that were thicker on one end, and often slightly rippled. And of course it made more sense for stability to install them with the thicker part downward.
It's all urban legend stuff. So look through the alt.folklore.urban FAQ for details on this.
Besides - there are plenty of people who collect old bottles and the like. My mom does. And there are none of those signs of "flowing" glass, even in some of the REALLY old ones. (As in older than those windows)
I don't believe glass flows at room temperature. At all. And if it does, it's on a much, much longer timescale than what we need to worry about here.
---
"You know your god is man-made when he hates all the same people you do."
Then, you could fit the entire world's yearly production of information inside a cube that measured...
cube-root(1.5 million) ~= 115 cm
(OK, so how about we create a couple of these every year, and launch them into space, just in case something goes horribly awry with our planet?)
Yes, this post is offtopic.
/. reader since 1999,
And no, no user is going to moderate me offtopic.
This is because approximately 3 days ago, someone at Slashdot turned off moderation. This is probably obvious to those who pay close attention to the moderation system, but I hoped to bring this issue before a larger audience. The facts so far are:
There are far, far less moderations going on. The ones that are being applied are mainly negative. This leads any reasonable person to believe that the Slashdot editors are moderating the stories by hand (this post, for instance, will probably be buried in short order).
I am posting this message for two reasons:
1) I believe that we, the readers of Slashdot, deserve some sort of explanation, no matter how cryptic, of what is going on.
2) I would like to promote feedback from other readers.
How you can help:
This post will probably be either erased or marked -1 rather quickly. Please copy it and repost it as necessary.
- a dedicated
-mwalker.
--
What happens when you outlaw guns
I think its interesting how the avg user will probably not need larger and larger drives anyhow. As the power of your computer increases, and thus the ability to compress/decompress using more and more complex but space-efficient algorithms, so does its ability to 'recreate' on the fly things like music, computer graphics, etc. Remember when everyone said the CD-Rom was the future of gaming because of the size of the medium? For a few years, yes, we had all those 4 CD-Rom pre-rendered movies/scene games, but once the hardware caught up, practically every game now just renders on the fly; consequently, games are way smaller. Anyhow, it is very cool from a datawarehousing standpoint .. things like geographical analysis, centralized data stores (a la library of congress, I guess) will benifit. And the size of the read/write mechanism shouldn't matter as much, although clearly it will have to be reasonable.
"Old man yells at systemd"
This is nothing. As an EE student, I developed a Write Only drive for the Macintosh 128 in 1985 that could store over 1TB of data. Problem is, I could never figure out how to get the data back out....
Seriously, this is pretty cool stuff. Are we on the way to "isolinear chips" of Star Trek fame?
RD
I dunno. I've been hearing about the holographic storage medium mentioned in the article for the past 2 decades and hey never seemed to get past the major hurdle of actually reading the data off without destroying it. I realize that they are having greater success with this, but I still have a hard time beleiving this will make it into the mainstream market for at least a couple more decades due to some unforseen hurdle.
I am MuchTall
Japanese url.
Interesting -- they mention that the areas hit by the laser emit 680nm light, and are 400nm in diameter, and are separated by 100nm in all directions...
They also mention that this is about 2500 times as much data in one square centimeter, and that they extracted different data from different layers of the cube by varying the type of doping material, thereby varying the frequency of emitted light from each layer...
By the way, why does the lameness filter prevent me from posting the url link in japanese with unicode? That's pretty lame...
Holographic storage has been around a long time, and mentioned on /. quite often. However, the substrate life is terrible, and the precision of the lasers involed in writing and reading would be prohibitively expensive to mass produce.
The difference between this and that, is that they store information by writing it onto the same spot from different angles. This is volumetric storage - it's a 3 dimensional grid of points.
Any spoon would be too big.
Lapsing into reminiscence mode, I recall during my first exposure to the internet in the early '70s there was a project called the Terabyte Memory. It was going to be this huge (!) datastore tied to the net so that everyone who needed it could have someplace to keep large amounts of data (for a fee, of course). As I remember, it was going to be a warehouse-sized building with a whole bunch of mag tapes (the only affordable choice, since washing-machine-sized disk drives holding just a few tens of megabytes cost thousands of dollars in those days). The 'highspeed' links of the internet ran running at 50 kbps. If you ever catch a techie saying how great the old days were, you'll know they've definitely gone senile.
XFS on my beloved SGI at home does
Max Filesystem size: 18 million TB
Max File size: 9 million TB
That's according to their spec sheet, I could only dream I had 18 exabytes, course then I might need something bigger than an Indigo II, to get good use out of it.
...isn't this what a lot of people complain about with regards to Microsoft: software bloat was made possible by ungodly amounts of hard drive space (or alternately, very inexpensive drives)?
So while I think it's fantastic that these advances are being made, is it really that big of a deal?
Just think... if there was no such thing as MP3s, would your hard drive be bursting at the seams? This is just an example, but there are many things that many people can download now with their broadband connections to quickly fill their (even 75 GB!) drives... porn movies, DiVX, MP3s, you name it.
Now, I'm not really against larger hard drives, but there has to be a tradeoff somewhere. DiVX (and the like) are great, but now with this new technology (hopefully) we'll be able to carry around a credit card or small box with all of our DVDs. I don't really want more space, I want better quality stuff to be stored on that space... I'd rather have DVD quality than DiVX. But if I can get 1 TB in a cubic centimeter, I want something a hell of a lot better than DVD (at least for videos).
Let's just hope the transfer rates will be up to par when this tech finally hits the consumer markets.
In "Babylon 5", I'd always assumed the "data crystals" everyone plugs into their computers used holographic memory as mentioned in the last paragraph of this article. JMS probably had that in mind, too, but it's interesting to read this and find that laser-read data crystals might not be two centuries away after all.
C'Mon people! The most interesting bit...
Using a femtosecond laser, broadband communication technology that enables transmission of terabits of data per second is possible, he said, talking of another project he is working on.
Oh dear! I think I just got sexually aroused by a technical article. =(
Pinky: "What are we going to do tomorrow night Brain?"
Pinky: "What are we going to do tomorrow night Brain?"
Brain: "I would tell you Pinky but this 120 char limi
XFS on my beloved SGI at home does
Max Filesystem size: 18 million TB
Max File size: 9 million TB
BeOS's BFS, also a 64-bit filesystem, handles 18,000,000 TB hard drives and similarly huge file sizes (aka 18,000 petabytes).
"And like that
The same may be true of such a tiny medium for a large amount of data. Assume it contained all the personal information on people in New York City. At such a compact size it could pass security and change hands nearly undetected. Detection mechanisms would certainly put each of us under a finer scope when passing customs, thus erroding further the right to privacy.
That such a tiny medium with such a wealth of information may pass so easily, it could be very useful to a resourceful terrorist, or simply criminal nuisance (which is what, IMHO, most 'terrorists' are.)
Imagine, too, the further challenge this presents to the RIAA and MPAA, and moreover, the media companies. Duplication wasn't a problem until there was an inexpensive portable medium. Store the entire run of M*A*S*H on one of these and, provided duplication is inexpensive (eventually it probably will be) and everyone could have it, cheap. (Which is why the RIAA and MPAA fight so vigoursly any new technology.)
Don't get me wrong. Producers and performers should be encouraged to create new content, lest they sit on their butts and collect royalties for the rest of their lives, but distribution of content which contributes zero to the original producers and performers can be bad as well. Failing to recover the original investment or money or building a career, people will have less incentive to create the works we enjoy.
--
A feeling of having made the same mistake before: Deja Foobar
The artical dosn't spacifically say this, but it appears to be WORM.
Perhaps, someone will write a filesystem that assumes a huge amount of space, but that can only be written to once. Changes to files would be handled with diffs and versioning.
Special care would need to be taken to insure deleted files can be burned out, otherwise this is a law enforcement dream.
Keep in mind that they don't mention the size of the equipment needed to interface with this tiny chunk of glass. While it probably doesn't take up a room or anything, it would also need to bee minimized for any actual space savings.
Above 1 MB, with multiple processes interacting on the data, it would seem to me that the storage device would start to look like a massive collection of cassettes and tape reels. Perhaps the racks and index card concept could be used.
Look to the past, my friend. :)
--
All men are great
before declaring war
A government is a body of people notably ungoverned - AC
First, I can answer some questions regarding this technology. I, myself, am involved in the study of memory methods in ultra-short intermediates in photochromes, which also offer great promse to cheaply store TBs of information. We have been able to store over 500 GBytes in an area around the size of 20 mm^2, and also read it out as fast as we can switch to the pages using an AO modulator.
First, access times. For the samarium method, the access times are in femtoseconds (theoritical). However, this is limited by the steering or positioning device, which can be mechnical, such as a galvanometer, or it can be an array of VCSELs (if femtosecond VCSELs exist, that is), which offer access times as fast as they can be switched and run at the speed of the underlying circuit, and act just like transistors in terms of their function.
As far as a file system is concerned, these memories are arranged naturally in a page format, and this is quite superior to current linear methods. To get to a location anywhere in the memory, you specify the page an the xy position where the data begins.
The reason why this method can store so much data is because 1) they are storing bits as 400 and 100 nm dots, and 2) they are storing it in a volume. As I understand it, the luminous 'dots' are not diffraction limited because they are not coherent and do not interfere with each other. CDs are limited naturally by diffraction because they use very coherent light, in an active approach to read data (e.g. light -> disc). This is because coherent light is the only light which can be collimated to the
However, unlike the view of the article, I am not optimistic of immediate commerical applications. If you read it you will note this method requires the use of a femptosecond light source. Femptosecond lasers are only very recent inventions, from the 1980s, and are also very bulky. Most consist of exotic Ti:Sapphire rods along with sophisticated Optical Parrmeteric Ossicilation and amplification, which all is very expensive and can easily fill an 8'x10' table. Some newer techniques use Nd:YUV04 with a crystal Q-Switch of KTP* or LBO, but these are still slighly bulky although they can fit in a box (still very expensive).
There are no diodes which can generate pulses at this frequency, and the only possible cantidates are organic LEDs which do generate at several hundred picseconds; if these are suitable I do not know. Diodes will probably not reach these speeds for quite some time, and as long as synethic sapphire and other crystals remain expensive (==low yield), this might be a cantidate for heavy industrial use (NASA is a good idea, as someone mentioned), but as far as that sugar cube dream, probably not.
There are also technical considerations, like readout. I would be very interested in how they plan to focus and read 100 nm size dots. There is no CCD or Si sensor with that kind of resolution. How do they position accurate to +/-
These is yet another subset of storage technology which will not make it out of the lab due to the need for certain intristic (yet overlooked) fundamentals. Holographic technology was also limited similarly, both by position and medium (although in recent years very good medium with BER of less than 10-^17 have been found), there still remains the need for very high resolution sensors and positioning devices. Until then, commerical applications won't be practical.
"I'll just chip in a bit for RedHat: I actually have that installed on my university machine." - Linus, '95
You could probably just rely on WindowsXP to handle the storage for you. These files will go to "My Pictures", these to "My Music", these to "My Documents", these to "My Applications", and these to "My Secret Stash of Porn - Password Protected".
How long I've waited for that last one!
Dancin Santa
.. that losing your marbles could become a verys serious affair...
.. if only.
How would you design a filesystem for a storage device with 1 TB or more? It seems to me that the directory tree concept would become unwieldy, too much stuff would get lost.
Above 1 TB, with multiple processes interacting on the data, it would seem to me that the storage device would start to look like a mini-Internet. Perhaps the "domains" and "search engines" concept could be used. Or is there a better way to design such a filesystem from the ground up?
Toronto-area transit rider? Rate your ride.
The article referenced in this post is a bit short on information, but readers can get a more detailed view of the story from this article.
The technique involved is refered to as resonant hole burning. Rufus Cone and his optical group at MSU have been working on many applications of this technique for years, including optical storage and stabilization of diode lasers (how's 20Hz linewidth for stabilization of a diode laser?) highly accurate clocks, metrology and so forth. Cone has a link to a nice power-point presentaion on his web page.
Cone and his group have been using crystalline materials, while this Japanese group is using glass. The advantage of glass is that the storage medium can be tailored to a specific shape. This abstract, published by the Active Glass Project, indicates other interesting research, including the up-conversion of photons using glass.