Slashdot Mirror
Better Holographic Data Storage
Pinlighter writes "Optical holographic data storage has the potential of providing better storage densities and access rates than the magnetic media used today. However, the technology has problems, mostly because the information tends to get a bit scrambled each time it is reread. According to the
link, a group of Japanese scientists have now developed a material which is stable for hours and across multiple rereadings. The material also allows easy erasure (by UV light) and rewriting."
Wow. If I wanted media that wasn't stable when you reread it, I'd go back to useing my 5 1/4" drive.
kwsNI
the information tends to get a bit scrambled each time it is reread
sounds great for AI purposes!
ok then your [sic] infringing on my copyright! Could you as [sic] me next time before STEALING my comments for your own?
I am no expert at this, but maybe the continuous, analog nature of holograms causes a problem here?
If you have a holographic picture, we usually have smooth surfaces; if you view a program as a picture, it looks (more or less) like randomly set pixels. Having a 3D representation is only going to make things worse.
Of course the goal is to increase the precision of the holograms, which will solve this problem; but unlike transistors, holograms do not have the kind of inherent "noise immunity" that silicon memory chips have. There is still a long way until we have data crystals à la Babylon 5.
I'm shocked that until now no one has said "Hey, Cool, we can have data storage like the Enterprise (NCC-1701D)!" And to think...you call yourselves geeks.
--
If there is a God, you are an authorized representative. - Kurt Vonnegut Jr.
No, it's an article about an actor. Don't think he's holographic, either, which is kind of a shame.
Hey, trolls - haven't you got something *better* to do with your time than sit on slashdot?
The materials they're talking about are hideously expensive to manufacture.. and I doubt their set-up is very fast at this point. I also have to wonder if they've started testing what normal electron flow, or thermal effects, to the material would cause in the way of intereference.
But they are starting to use angled beams to layer information, which should make the overall construct small, so possibly feasable to systems needing huge storage capacity.. at least short term capacity.. I also have to wonder if the beam disturbs areas that it's just 'passing through' and not trying to read.. in that case I hope the IRS adpots this technology in the near future.. I can just imagine what wholesale data degredation would cause some 'interesting' activity..
Until you get down to about 10^-7m you can consider everything analogue. Memories have a continually varying number of electrons held in a capacitor. Its the thresholding of the charge that defines a 0 or a 1.
.3 its a 0".
You can do the same with holograms, "if the reflected intensity is >0.7 (relative intensity) its a 1, if its
-Yarn - Rio Karma: Excellent
Apparently these new iron memories would change in one clock cycle, but would then hold the state. (I can't recall the exact details of how it works, it was too late when I read it last night.) I got the impression that it could be used like EEPROM or flash-EEPROM, except sufficiently fast and inexpensive to be used for main memory.
I think that this technology holds great promise for the industry and though it is not at the level needed for the consumer market there is no reason to put it down. For kwsNI you are contradicting yourself. In one reply you say that companies should put money into R&D and not lawsuits then you come in and say you would not use a new technology even before it is out of the R&D stage. So please make up your mind...Are you for or against R&D. If you are really for R&D then be supportive of a new technology and if you are not then don't suggest that companies spend money on it. If everyone looked at a new technology like you did in your reply then there would not even be a internal combustible engine.
So I am a little confused. It appears holographic storage merely uses a material that has optical memory properties (the electrons move to the light) to record a hologram on. So what is the benifit of holographic storage.
It would appear you could get just as high an information density using lasers not in any holographic projection system. So what is the benifit of holographic storage.
Marriage is the "pseudo-ethics" that cloaks the messy truth of sexuality in the raiment of propriety -- it's "Don't Ask,
Hmmm. Why not refresh it like dram?. Assuming you could refresh it reasonably fast and your computer doesn't crash.
Ryan
Remember: HAL's storage was Holographic. Holographic Algorithmic Whatsit...
. We've got computers, we're tapping phone lines, you know that ain't allowed - Talking Heads, "Life During Wartime"
So, for conceptual purposes, one can imagine the device functioning much like a CD with "stacked" tracks.
Humans are pretty damn ingenious.
The article doesnt seem to make any mention of how much space these babies can hold. I recall them being very good storage per physical space but the last time I had heard about them was a few years ago(?) and magnetic storage has come a long way since then. Anyone got some details?
God Fucking Damnit
I did undergraduate research on iron-doped lithium niobate. While LiNb03 *could* be used to store data holographically, organic polymers do a much better job at lower cost.
LiNb03 is fragile (it's a crystal, drop it and your data's dead) and very expensive to produce. The crystals are grown using the same? process as silicon but has a much more complex crystal structure and is much harder to produce consistent crystals of good quality). A crystal about 5 cm x 5 cm x 1 mm cost several hundred dollars.
Here's a project that some people are working on up here at Syracuse University.
:-) Okay, not exactly, but it's fun to call it that. A very small rectangular cube (about 0.5" by 0.5" by 1.5" or so) is filled with a protein substance suspended in a solution. It is primed by hitting the substance with a laser, and it denatures the protein. Then, a laser is shot through on an X plane, and a different laser hits on certain points on the Y plane. The protein is denatured where the two meet. This substance is a three dimensional memory system. You read back by a similar method, but with a lower power laser beam. The laser doesn't pass through the points that are denatured, and produces a grid of binary numbers essentially. Of course, the protein cube requires no power, so it's perserved when off. Imagine being able to take the ram out of one machine and put it in a different machine, without changing the contents.
:-)
Jello RAM.
These small cubes can hold about 4 gigabytes of data, and last price I heard was $20 (the cubes are practically nothing. They're cheap to make. It's the read/write equipment that costs a bunch). It has decent access speeds, cheap, and very small.
This is all very experimental lab stuff right now, so the size and speed can change. The goal is to make very cheap, small, random access memory. Might be good to replace tape drives. It's several years off though, but money keeps coming in and development continues. Should be nice stuff, keep on the lookout for it.
I saw development stuff in use a year ago when touring that laboratory, so I reserve the right to be incorrect or inaccurate with some of the statements in this post.
Mike DeMaria
It's the FBI, we're being raided!
Quick, break out the spoons! Eat the evidence!
for some jerk to come along and say " THIS STORY HAS BEEN POSTED BEFORE " and whine and whine about it... the people that do this are total assholes and should be anally raped and tortured..
:)
on the other hand, Holographic memory is quite fascinating.,, a few years back i collected all the holographic images that I could find, European artists/technoists seem to love holograms... theres always been a big market for them in Amsterdam/Berlin..
now if i could fsck that holographic nude Marylon Monroe, Id be much happy
From what I remember of my university photonics course the most pressing problem with lithium niobate memory is the slow read/write cycle.
doped lithium niobate can be "fixed" - made read only by baking for a few minutes, readback time is measured in seconds. The coolest thing about it is its ability to regenerate data from partial images, the classic demonstration is recording a line drawing of a cat. Feed in a picture of the cats tail, for example, and the whole image regenerates over the next few seconds.
According to the lecturer it was operating as an optical neural net.
memory structure is as follows:
light in
|
|
--#-----\
| |
\-----/
where the # is the crystal & the \/'s are 45 degree mirrors. Memory works by forming a diffraction pattern where the beam crosses itself in the crystal. An image is superimposed on the beam and recorded inside.
Lithium niobate is an electro optical material with a planar structure. Where there is a voltage across it its refractive index changes. The doping allows ions to sit in the material. These ions are disturbed by peaks in the diffraction pattern & get dislodged into other layers of the crystal, this provides the voltage necessary to modify the local refractive index & store the pattern.
Baking the crystal fixes them in position.
Too lazy to type any more.
Memory you can only write to once! YAYYYY no, really. If you are denaturing protiens and, say, an undenatured area is a 0 and a denatured area is a 1, since the denaturing process is generally a one-way street (and it'd take a lot to make be believe you can reliably put a protien back together with lasers), eventually all your data is going to be 11111111111111111. It's the same principle as burning a CD-R. & IANAS but I think the only group who would want to use this as a ROM storage device is the MPAA. It must be horribly difficult to get the protien stuff to last more than a few weeks before it starts to smell or degrade from exposure to light or what have you.
Memory you can only write to once! YAYYYY no, really. If you are denaturing protiens and, say, an undenatured area is a 0 and a denatured area is a 1, since the denaturing process is generally a one-way street (and it'd take a lot to make be believe you can reliably put a protien back together with lasers), eventually all your data is going to be 11111111111111111. It's the same principle as burning a CD-R.
& IANAS but I think the only group who would want to use this as a ROM storage device is the MPAA. It must be horribly difficult to get the protien stuff to last more than a few weeks before it starts to smell or degrade from exposure to light or what have you.
The act of re-reading does not scramble data nor does a few seconds mean decay. That is why this story is of any interest at all.
It says they "have now developed a material which is stable for hours and across multiple rereadings."
take a triptonica to subthunk
my grammar is horrible!
[wow, there are a lot of idiots posting today, not you, the ones that replyed to you]
I think the point is to make it stable without consuming power. Your solution is fine if you want RAM like memory, but not if you are trying to get a bigger hard drive.
Yeah, and you didn't put the apostrophe in the right place in your post, but do you see us wasting space complaining about it? Oh, I guess we do. Great discussion board we have here.
Your hard drive is on the lowest level an analog device - that's why you can read data that has been overwritten using several times if you have good enough equipment. This implies that the magnetic head is writing to the disk at a great redundancy in both area and magnitude.
There is no inherent "noise immunity" in the hard drive or your memory chips either, they have their own decay rates which are slow but existent.
You can do the same with any analog device: just store your thing with large enough pixels and volume and it will stay readable for a long period of time (provided the material won't decay).
Holograms (in general - not too sure about those used for data storage) store information as the Fourier transform of the scene they're recording. This has two consequences:
- A point of light is stored as a series of smooth fringes in the hologram
- the fringes for a given point of light extend over the whole of the hologram
So, the analog nature of a hologram isn't so much of a problem. Of course, the resolution of the hologram will limit the data storage capacity, as will the size of the hologram. However, if you lose half of your hologram, you don't lose half your data; your data is just twice as blurry (if that makes sense).Edric.
If the data density is as increased as everyone says it is, we can easily eliminate errors by compensating for them with hamming code. We already do it on CDs and tons of other types of data storage that has the possibility of loss.
Esperandi
Buy stock in vibrating mirrors!
I know what you're going to say, CDs are reliable ways to store data. And you're right. They're also unstable. So how do we take an unstable media and make it into a reliable data storage device? Hamming code. Complex error checking. Its a bit of overhead, but you don't have to worry about the instability and you still get some of the data density increase.
Esperandi
BTW, about CDs, if you scratch the "bottom", the CD will survive. If you scratch the label side, you're screwed, so quit laying them down label-side-down!
I think that write speed was also a big problem for holographic memory and that hasn't been addressed. The writing process is typically done using a photorefractive material and this means you shine light on it, set up an electric field and let the excited electrons diffuse trap states in the `dark' regions. This results in an index change that can be `read out' later. The problem is the diffusion time. I don't see how they can make that a very fast process, so while you'll get great read times, the writes are slow. Last time I, speed was more important to most people that size.
If this is true and the "decay" is predictable, this is easy to fix. You want to store 0xEF43? Pass it thru a filter that turns it into something that when corrupted turns back into 0xEF43... then after you read it once, refresh the store.
Of course if that doesn't work, error-checking code could easily make all of this possible.
Esperandi
Please tell me they're gonna make this stuff look COOL when they produce them? I mean, harddrive could look cool, but instead they chose to hide them in a metal box inside the computer (they coulda mounted it in plexiglass and had it mount in the front face). I'm thinking of a tray that you can eject like a CD-ROM tray but has the cube with the lasers around it sitting right there, maybe even add some sort of fluorescence to the process so you can see "glitter" when the laser hits a denatured spot (or a non-denatured spot, wouldn't make much difference)...
Oh, and make it an eerie green like the green used in The Tommyknockers movie..
Esperandi
WE 0WN J00!!!
That would be "they", not "we". You are a spammer, not a troll.
Trolls 0WN everyone, but they do so in an intelligent, creative fashion, not by mindlessly disrupting discussion. Trolls actually encourage discussion, in the form of luring those who think they are super-31337 into responding very strongly to an obviously fake post. It's an art form.
It's hard. It takes effort, and intelligence.
And this is why real trolls can't stand spammers: spammers keep good trolls, real efforts, incredibly funny, subtle masterpieces, from getting noticed.
So don't glorify yourself quite yet. Write something good, funny, flamewar-provoking first. Pose as an unbelieveably clueless ex-lawyer marketing department RWM whining about the Imminent Death of Slashdot (tm), to combine a few of the more prominent troll themes. Or you could take the outright funniness troll method, and post the next "Star (as in hot young actress) Wars".
But until then....
It is so awesome that they have the technology to do this. Imagine, backing up several hard drives to one disk about the size of a 3.5" floppy, screw Iomega's zip drive, if this is market'ed well, we may have a new super floppy/tape replacement/keep up with your proccessor/use it as more virtual memory that is faster than real memory/holy shit OC-12's are going to get the're bandwith's worth in downloading/etc...
I remember reading an article in the New York times around ten years ago, talking about how holographic storage was going go be the Next Big Thing. They talked about 1-cm cubes holding a gigabyte (GASP!) of storage. Of course, this was when a 65 MB 5-1/4" hard drive was large. Hopefully we'll eventually see some actual holo storage technology in common use... eventually. I'd hate for this technology to become yet another of those things we always hear about but never get to have.
"Destroy science and religion. Science would re-emerge exactly the same; but not religion." - Penn Jillette, paraphrased
I have recently written a paper on this area for a photonics course. The inherent problems of 3D data storage are numerous to say the least. The only way currently to write the data is by way of laser, which also creates a problem of cross talk between the layers of the lattice structure of the recording medium. There is currently research being performed to eliminate this problem by utilising a dual-laser writing technique that would create only constructive interference at the position of recording.
The other major problem is the type of laser used. A pulse laser writes and retrieves data faster, yet is known to damage the recording medium. The solution is to use a high-powered contiuous-wave laser, which is being looked into.
The real decider in whether holographic memory is whether or not a certain compression of data can be reached. Off the top of my head I think it is somewhere around 10 megabits of data per square cm. Last I knew, I believe it was somewhere near 1 megabit per square cm. This is a really facinating topic, and I encourage all to look into it for themselves.
Ciao.
nahtanoj
Volumetric storage, where data is stored all through the volume rather than just on the surface, is a obvious idea that's been tried a few times but has never really worked. DVDs have a little of this, with several layers accessed by adjustable focusing. The coincident-beam laser scheme is attractive in theory, but requires beam-steering, which is mechanically messy. Still, mechanically messy concepts have been turned into mass-market technology before; look at the innards of a VCR.
Stable, write-once volumetric storage might be useful as a backup and distribution medium. That may be a more promising direction than something that degrades with time.
Corning Glass once built a computer display device using a photochromic plate for image storage. The plate was written with a UV CRT, erased with a bank of IR lamps, and read with a green light. It looked like a microfilm viewer, with rear projection. A few units were built and the idea dropped. This is one of the few products, other than photosensitive sunglasses, ever to use photochromic technology.
That's awesome! Format your holo-drive and get a suntan.
Now I won't have to go outside to get rid of my pasty white complexion.
/will
About photopolymers: it is still impossible to make thick films (they are getting at 0.5 mm at present) and they are write once only (they need to be developed by UV light). You are interested in having bulk crystals instead of coated platters, because holograms can use the whole volume instead of some 2D layers as in HDs, DVDs and CDs. But photopolymers are still promising, the change of refractive index (i.e the efficincy) being much larger than in LiNbO3.
The paper addresses tecnical problems, because it is based on the scientific paper given in its references, which reports nothing revolutionary (the photochromic effect was already reported by Germans 6 months ago).
I think not more than the analog nature of electric, electromagnetic and light signal is, the basic approach could be the same.
If the lifetime of these memory blocks were limited to a small number of reads before needing to be replaced/refreshed, I think this could be the next candiate for the MPAA's replacement for the DVD!
The Other Nate
The Other Nate
But one thing you can do very efficiently with photorefractive crystals is comparing images (at some Terabit/s). This is because to calculate the correlation of two images (which tells you if two images are similar or where a smaller image is located in an image) you need fourier transform the images. Optically you can do a Fourier transformation very efficiently simply with a lens (in 10^-10 s = 100 ps): at the focus of the lens you have the Fourier transformation of the image and the time you need is just the time the light has to cover the distance of a few centimeters. A good image comparing system can be useful for military applications (guided missiles), security checking (fingerprints, face and voice recognition), person tracking, internet traffic filtering,... All dreams for the NSA, horror for me!
Realizing that you reserved the right to be incorrect or inaccurate, I think there's something missing here. If the read laser cannot pass through a denatured protein, how does is read the points behind a denatured point. Perhaps one laser will pass through a denatured protien unless it is being "activated" somehow by being hit with another laser?