Slashdot Mirror
Cringely's Shameless Self-Promotion
wild_berry writes "The latest edition of Bob Cringely's column at pbs.org, entitled Shameless Self-Promotion: Bob's Disk Drive is up. He's talking about replacing the glass or metal platters in present hard disk drives with foil platters in order to save energy." From the article: "The materials cost more but we use so much less of it (the disk is so incredibly thin) that the total material cost is substantially less. This 'floppy' material has the same kind of magnetic coatings used on standard disk drives and our drives live on the same technology growth curve as those others. The way we obtain greater storage density is simply by putting more platters in a drive (say 12-15 instead of 4-5 in an enterprise 3.5-inch drive) because they are much thinner and can be stacked closer together. The only parts of the drive that are significantly different are the platters and the heads and the heads vary only in having an extra slot."
Get out the tin foil...umm.. okay, it's alrealdy in there
I should read the FA, but what's to stop his platters from flopping all over the place?
The
And what do these thinner materials and more closely-spaced heads do for the MTBF and error rate in such drives?
Hey, I'm as much concerned about the environment as the next guy, but I would think foil platters, similar to floppies would make the disk much more vulnerable to head crashes. The disk would be cheaper and more env-friendly, but would crash fatally much more often...
That's quite a bold claim! If his claims are accurate, then we may be looking at the future of hard disk drives. Micro-disk drives would become the latest hotness, and Flash would disappear entirely from our memory. IF the technology works, that is.
Time and speculative investors will tell if it's really everything it's cracked up to be. I certainly hope it is, but extraordinary claims require extraordinary evidence.
Javascript + Nintendo DSi = DSiCade
Wouldn't that make it extremely prone to head crashes, worse yet if the media flexes even just a little bit it would drastically sharpen the impact angle and possibly even rip the foil right off the spindle. Sounds like a terrible idea for anything that isn't seismically secured.
Aluminum goes for about 60 cents a pound, glass much less. A typical disk drive is gonna use under a nickel of raw material in the platters.
"I should read the FA, but what's to stop his platters from flopping all over the place?"
Just keep 'em spinning all the time to keep them in a nice flat disc.
Where were you when the voynix came?
Ideally, the disks would be spinning so quickly that the outward force would keep them almost perfectly flat. Assuming the disks were very smooth and the internal atmosphere of the drive is gas-only (no dust - a safe assumption) there would hopefully be very little turbulence within the drive to cause fluctuations in the flatness of each platter.
In my lab we coat floppy materials (like plastic) in a spin coater at several thousand RPM. At that speed the disk may aswell be rigid.
At the UIUC Reflections|Projects ACM conference. It was actually a fairly interesting talk (http://www.acm.uiuc.edu/conference/2006/webcast.p hp) about the same topic, maybe a little more in-depth than the article. At least more pretty pictures than the article.
Slashdot Patriotism: We Support our Dupes!
Some guy shows up all paranoid about some problem, and wants to solve it using tinfoil. It's not like we've never seen this before.
I think he can use the same arguments to justify making automobiles out of tinfoil instead of all that nasty costs-too-much-energy-to-move heavy metal.
Where were you when the voynix came?
The only time you'd see a difference in energy consumption would be during spin-up. Not worth optimizing such a rare case (except possibly in a laptop drive where it does happen a lot and would contribute to battery life.)
Cringley must be old enough to remember Bernoulli disks. (They used a plastic film but same concept applies.)
Well at least I will have material close at hand for my hat when the aliens try to scan my thoughts.
How about a stack of "foil" platters read by a single head outside the stack, that can "focus" its read-sensitive probe electromagnetically inside the stack? Maybe they wouldn't even need an airgap, just some intervening film to help "address" the different layers. Perhaps a pair of heads reading a "stereoscopic" view. Maybe that could read a whole track at once. Multiple heads around the radii could read simultaneous tracks.
This kind of tech has a lot of problems in signal/noise, permissivity/permeability, etc. But the benefit could be a drive the size of a mini-CD-single case, with dozens of microlayer platters and octets of read heads, offering terabytes for milliwatts.
--
make install -not war
Speed control of the rotating disk is going to be harder if the disk has less mass. You basically loose a nice dampener that you had in the system.
The only real power savings would come during spin-up. Once the disk is spinning, there's no additional power used to rotate a heavy vs. a light flywheel. (Well, a little bit because of increased bearing friction, but it's probably negligible.)
Finally, if you lighten up the parts in a hard drive, most companies are just going to use the energy savings to drive the parts FASTER.
IANADDEBIAAME*
*I Am Not A Disk Drive Engineer But I Am A Mechanical Engineer
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
Okay, now with them spinning, rotate the drive around an axis other than the spin axis of the discs...
I seem to recall in the late 80's and through the 90's a removable cartridge drive system known as Bernoulli drives. They had "floppy" media (mylar, though, not foil), The drive would spin up the disk, then insert the heads, which were like hard drive heads - floating over the surface rather than the more standard pressed against the surface (a la Zip/Floppy drives).
Ah, Wikipedia - http://en.wikipedia.org/wiki/Bernoulli_drive
Basically, this drive is similar, just in a self-contained format rather than a removable cartridge solution?
Though, bumping the drive while spinning could do a lot of damage from precession of the platters causing the material to warp. Fast spinning disks are miniature gyroscopes.
"Okay, now with them spinning, rotate the drive around an axis other than the spin axis of the discs..."
And???? Due to the significantly decreased mass of the platters, isn't this much less of a problem now?
Where were you when the voynix came?
A little off topic, but where the hell is season 2 of NerdTV? It was promised at the beginning of the summer and it's almost the end of the year and still nothing? Not even an update as to what's going on with it :-(
i dont get it.
Datacenter disks spin 24/7. The energy used by the inital spin-up is null compared to the runtime.
Energy-wise it does not matter if if keep a 100gr or an 1 gr platter spinning. Its all just "wind" resistance. That gets higher the more platter you add.
So why will this 12-platter-design use less energy ?
...that having a heavy disk in the drive would make it more stable, easier to keep and a constant speed and generally a lot better for wear and tear on bearings and heads.
I also seem to remember that glass and ceramic platters don't expand as much as metals do during thermal change which happens a lot as drives are turned on and shuts down, so I'd wager that his idea of using tin foil, aluminium or any other metal is flawed. Seriously.
Adding more cheaper platters to reduce cost almost makes sense. That is until you consider that the read-write heads are one of the most expensive parts of a drive. Adding more of them would substantially increase the cost of a drive.
In particular, I'd like to see evidence for the following claims:
I'd sure like to see the assumptions and numbers underlying that equation. Gee, Cringe, which do you think costs more: The raw platters themselves, or the read/write heads? I would say the latter. So you're going to drop the costs of hard drives by doubling the most expensive component? Huh? Sorry, I'm not buying this at all. You don't think a non-cleanroom enclosure is going to result in data loss on the platters themselves? Even if you're not getting particles during the read/write phase itself, you're getting them on the platter. I'm not buying the logic here. Sorry, I'm not buying this at all. Until the advent of true Drexlarian nanotechnology, I doubt you're going to see a mechanical action (you still have to move the eread/write heads) beat an eletronic one (reading from Flash).I'm not saying that the technology Cringely talks about is impossible, I'm saying: A.) There seem to be a lot of unwarrented assumptions underlying his logic, and B.) Implementation always has unforeseen hurldes and obstacles that will make these drives seem like far less of a slam-dunk vs. current technology (or more specifically, where regular drive technology will be 18 months from now) than it appears.
Finally, once it is ready, I'd like to see real-world tests for speed/electrical consumption metrics with existing technology. There might indeed be some savings, but I seriously doubt they are as dramatic as Cringely claims.
Crow T. Trollbot
The platters have (per platter) a much smaller edge, so they're going to get less friction from the air. Less friction means less heat AND less power required to keep the disk spinning at the same velocity. The area isn't that big compared to the surface of the disk, but I'd guess (assuming the heads were at the outer tracks of the disk) that the air near the spindle spins with the disks and probably causes very little if any friction, so the majority source of air friction is going to be the edge (where it moves air around the "interesting" interior shape of the enclosure).
In his talk I referenced above, he specifically stated that they were using smaller/lower power motors because they didn't need as much power as a conventional disk. Also remember that conventional disk motors may have to be "overspecced" to be able to spin the disks up to speed in a reasonable amount of time, and that may make them less efficient when they're just trying to maintain speed rather than spin up. You'd have to ask an EE on that one though, 'cuz I'm not. Just another stab that occurs to me for why it may cost a lot less power.
He also referenced making higher-RPM drives than current methods. I want to say 30k sticks in my head, but I'm not sure on that you'd have to watch the talk to verify my tylenol cold muddled memory.
Slashdot Patriotism: We Support our Dupes!
I see this solution to be prone to data errors. Therefore, I suggest that the vendor puts their money where their mouth is. Put them in some dozen tablets at a college for students and test the fail rate. Stick them in a SAN for a large data center and test the mean-time to failure. Hell, run Slashdot on them. I want to see failure, power consumption, and shock test data.
In God we trust, all others require data.
...they have to wiggle the term "nano" in there. It doesn't matter if it is nano or not, "nano" is the tech buzzword now, everything new and shiny has to be nano to excite the vulture capitalists.
I for one welcome our new iPod mini overlords
the more they over-think the plumbing the easier it is to stop up the pipe
Hmmm, their website seems to be down.
I've always been curious - why don't modern drive have a spiral array of heads per suface, instead of the slower mechanical heads? It seems like track-seek speeds would disappear in such a design. Is the cost of a drive head than great, and how much of that cost is due to the movement mechanism itself?
I think he talks about how the foil disks are very flexible and therefore can avoid head collisions and return to their previous state very quickly. I assume this flexibility and very low mass would also take care of wobbling.
I'd like to know how resistant the disks are to stretching over time due to the very high RPMS.
It is by the juice of the coffee bean that thoughts acquire speed, the teeth acquire stains. The stains become a warning
I have enough problems with current hard drives (albeit cheap and old ones) dying when some dimwit in my office kicks the computer underneath the desk as he's trying to lean back. I can only imagine how fragile these drives will be compared to current technology.
... stuff I keep hearing about. Just how much drag do you think is incurred by a rotating cylinder of almost-perfect smoothness? Reducing platter thickness will make almost no difference at any achievable rate of spin.
Oh, and try spinning a foil platter at 30K RPM sometime and see how that works out for you. I don't normally think of confetti as a deadly weapon, but like you say, we're talking some interesting physics here.
Now imagine what happens if you tilt the drive.
The hub now has to transfer a force rectangular to the foil-plattern's surface - fast - to tilt the rotating plattern inside the drive.
But the foil-plattern want to stay where they are (think bicycle wheel)
A foil doesn't provide much resistance rectangular to it's surface. The process is called "folding" if done exactly or "crumpling and head crashing" if done in a foil-platter-drive. Maybe it would even be called "cringling" then?
Do I make any sense to you?
Coincidently the CAPTCHA for this posting was "weakness"
You just saved 300 watts, congrats.
How about we focus on using reusable materials for cases and PCBs? That will save our enviroment and wallets a lot more.
sure the floppy materials would be rigid while power is applied and the disk is spinning, but how will the manufacturer ship them to end users? Wouldn't the floppy platters just crash and bump and get all scratched up while in the hands of FedEx/UPS?
Sure traditional Hard Drive manufacturers may be in jeopardy if they don't license this technology but don't discard flash just yet.
First thing flash has over this technology is *proven* reliability. This new technology can't buy that for money nor love.
Second thing is that this technology has *nothing* over flash (except maybe extreme temperatures, but special flash chips exist too). Performance is not said to be better than flash (you can't beet nanoseconds to access data in flash).
The only thing it has over flash at the moment is a cheaper price. Have you seen flash price trends over the last two years? I would say that it roughly obeys an inverse Moore Law (where prices for a same capacity are halfed every 18 months).
Flash chips are nothing but plastic and silicium. If Sandisk our however started feeling some heat from this new technology they could *ALWAYS* lower the price, hoping to make it up in volume.
At the moment flash manufacturers are at max capacity and are structuring their prices to maximise profit IN THE CURRENT MARKET CONDITIONS. If a new competitor comes out with a ground breaking technology they will find a new price point to maximise their profit then.
Flash, inlike hard drives cost almost nothing to produce, their marginal cost is virually pennies, unlike tens of dollars for HDs. They currently support investment costs and high margins, but in a differnet market configuration they could outprice these new disks and ramp up production.
Flash is the future, its already here but the chip companies have no incentive to make it any more affordable than it currently is, they are milking us just like OPEC does with oil.
If somebody invents tomorrow a car that recharges in 3 mins and has 500 miles range and same performance and price as regular cars, the oil barrel will drop to $15 overnight, it's the same thing.
It's all about supply, demand and marginal costs.
Artificial intelligence is no match for natural stupidity
Its a joke, people.
The average disk drive consumes 6 watts. The average CPU and GPU are power-mad monsters. Intel and AMD have ramped up pwer consumption over the last 2 decades, while drives have come down by almost an order of magnitude.
Put your tin-foil hats on tighter, please.
I have a 5.25" 2/3 height (yes, 2/3 height, one-bay-and-a-half) removable SyQuest drive here with a couple of old disks. It no longer works, but it connects to any industry standard "MFM controller" (you remember, dual cables, a 14-pin data cable and a 34-pin control cable, really an "IBM PC style" hard drive adapter is probably more descriptive) and has a large glass window with a foam seal. The disks are 5MB each; to change them, you open the window pull the old one out, and put the new one in--with the PC off, of course.
I also have a truly massive dual-Bernoulli box here, with dual 5MB drives. These connect to a proprietary Iomega ISA controller over a large cable. The box is exactly the same size as the original IBM 5150 PC case, but the front is completely open and the two drives occupy its entirety. They have huge doors. The 5MB disks are the size of large paperback books or maybe slightly smaller than dinner plates. The thing weighs more than fort Knox.
I don't personally know which was first, but I was personally using both of these way back in the mid '80s as a BBS operator who needed backup and storage rotation media, and I had already got them secondhand from local universities, etc. at that time.
STOP . AMERICA . NOW
>The platters have (per platter) a much smaller edge, so they're going to get less friction from the air.
i f
Not failed physics... this is a flawed analysis.
There are a few different types of drag (I am an aerospace engineer). The relevant one in this case is caused by the surface of the platter, not the edge. Remember, the edge is really acting as if it were stationary - it's not moving the disc laterally thru the air, so the edge is irrelevant. Instead, the disk surface moving past the air drags some of the air with it - this is sometimes referred to as surface drag, or skin friction. No matter how smooth the surface is, moving it thru a "fluid" (such as air) causes shear in the fluid - the fluid closest to the surface is motionless, and the speed builds up as you move out away from the skin. This is called a "boundary layer". This layer can be smooth-flowing ("laminar flow") or rough (turbulent). Smoother skin means laminar flow. But there's STILL drag, no matter how smooth the surface.
Here's a good illustration: http://wright.nasa.gov/airplane/Images/boundlay.g
So making thinner disks and using more of them means MORE drag, not less.
Actually in this case, however, the motor can be sized DOWN, despite higher operating drag, because the largest power usage comes from spinning up the disc package - and a set of lighter disks will require less power to spin up. Very little power is actually used to keep it spinning, despite the drag.
--Brandon / Split Infinity Music
Reduce, reuse, cycle
There is no such thing as centrifugal force.
I think you'd best watch what you say on slashdot. Some engineer might hear you utter those words, and then horrible things could happen. One time I heard a story about a guy who said something about centrifugal force while he was having dinner in a restaurant; an engineer happened to hear him and he killed the whole town in a fit of rage!
*hides*
:(){
The disk is very floppy. The metal center is the only rigid part. The floppy plastic of which the disk is composed does not flop because it is too small, measuring only 1 3/16 of an inch from the metal hub.
We have always been at war with Eurasia!
And all those heads jammed in next to each other? A head, especially one that has to follow a 30kRPM wobblin' taco, requires a complicated gimbal suspension mechanism and arm. These things all take up space, which is in tight supply with 10 disks in a 1/4" high package. And putting R/W heads in close proximity can lead to crosstalk -- writting data on the platter "under" the head (the one you want to write to) can degrade the data already written to the platter over the head.
The Russians have won. They have made the world a cesspool of distrust, greed, fear and hate.
If anyone wants to learn more about the technology, it can be helpful to look at the patents themselves. A quick search on Antek and Anil (the company and one of the inventors' names respectively) turned up these three patents, which likely cover the technology in question.
May be a useful thing to read. {Prof. Jonathan Ezor, Touro Law Center}
I know, and I give my first-year students hell for doing what I just did. I sat there with that comment written for a couple minutes debating whether or not to say centrifugal or centripetal force. The problem is that traditionally the latter results in people saying something along the lines of "wtf you mean centrifugal n00b" and I opted for the more commonly used (if incorrect) term.
Damned if I do, damned if I don't. Oh well.
If I'm reading the article correctly, the claim seems to be that the lighter platters will save energy?
How?
With my primitive understanding of physics, the power required to keep something at constant velocity is basically the sum of the parasitic losses (in this case, aerodynamic and frictional losses). Changing the weight of the platter does not have much impact on energy consumption *except* for periods of acceleration (e.g. - the first couple of seconds during power-on).
Has my logic failed me here? How do the lighter platters save energy in a constant velocity system?
More
Question 2: I didn't see any reference at all to perpendicular recording in this article. Is that a feature of these drives?
Observation 1: If they're talking about us seeing these drives from major manufacturers in a year, then all this must already be running in the labs by now.
Observation 2: Combine this with those capacitor-batteries discussed a few months ago, and the whole landscape of personal electronics could look very different a year or two out. It could virtually make existing units obsolete overnight. Perhaps it's time to consider stock in the major manufactures again.
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."
I haven't watched the video but I did read the article and color me cynical but I have a hard believing any talk about a technology that is orders of magnitude better than an existing, established, technology and is cheaper. Plus this is based on 10 year old patents. Wouldn't somebody, somewhere within a hard drive company have seen these patents and worked on this technology if it were at all feasible in the market? There's any number of technology "breakthroughs" that look awesome in the lab but don't survive trying to mass-produce them. Making a few million of some complex piece of machinery is hard. That said, good luck to him though. I hope to be buying their drives soon.
I spent 15 years of my life designing disk drives and the chips for them, and a lot of them are still out there running in lots of machines.
Utterly stupid idea - As others have pointed out energy losses are due to rotational losses (friction and windage) in the drives after spinning up. Spindles in 3.5" drives with 1 or 2 platters (99.9% of the market) take about 1.2 amps of start current, and then speed maintenance on the order of 40ma to 200ma while running.
The article proposes adding multiple heads and more platters to the drive. With thinner platters, and proposes that thinner is cheaper. The cost in a disk is not the material it is made out of, but all of the processing associated with the preparation of that disk, machining, polishing, plating, etc etc. Not the price of the chunk of aluminum!
The most expensive thing in a drive is still the recording heads and the disks. The article goes on to propose adding more platters with more recording heads. More stupidity, pushing the cost of the drive through the roof for all the heads!
All of those heads add more frictional and windage losses, pushing the spindle current up.
Also, while in use the heaviest power consumption is in the servo. Those "seek times" that everyone likes to brag about take serious current to achieve. Add more heads and platters and this current goes up as well, and how thin the disks are will not effect this one bit.
So, higher cost for the drive, more current to maintain speed, and slower access time?
It's not April First by any chance is it?
Tin Foil Hat for this one.
www.effectiveelectrons.com "chips that work" Analog, RF, Mixed Signal
In light of The SHADOW's comment I retract my earlier statement about you making up the statements about the CAPTCHA.
Slashdot: Failed Car Analogies. Amateur Lawyering. Anecdote Battles.
Is to market or gear the storage device to PC's instead of the digital cameras? That's his idea? But why does he claim the "whole idea" is pretty much his?? What the heck.
No energy whatsoever is saved by this as the starting inertia is negligible if the drive is allowed to run for more than a few minutes.
While it might save energy in notebook drives which could then start and stop on an as-needed basis, but much better would be to design a drive which stored energy at shutdown in a supercap and used that to augment spin-up.
And a foil platter would NOT be shock resistant.
This idea is a non-starter.
Dog is my co-pilot.
Couple of problems here -
1. Heads are one of the most expensive parts of the drive.
2. You'd need over a thousand heads for each disk surface, so even if they were REALLY REALLY cheap, they'd easily bury the savings of not having an actuator. If a disk drive head cost $0.01, and an actuator was $10, you'd just about come out even on a single-sided, single platter drive (and the actual prices aren't anything like that low/high, respectively).
3. All modern disk drives are dynamically servoed, meaning that there are magnetic marks on the disk itself to allow the drive to figure out which track it's reading from. If you had fixed heads, you'd have to put the data tracks farther apart, so thermal expansion wouldn't cause the tracks to migrate out from under the heads when the drive heats up. That would dramatically reduce data density.
For what it's worth, there have been dual-actuator hard drives made, from time to time. They have one arm on each side of the disk, which can each seek independently. They have approximately half the average seek time of an equivalent single-arm drive, and get better transfer rates, as well. On the downside, they cost significantly more, use more power, and have a lower MTBF (twice as many heads to crash, after all).
-Mark
No real reason why the whole caboose couldn't be sealed and placed under vacuum? It doesn't have to be an extreme (-95kPag) vacuum to be worthwhile, either.
The heads in these kinds of disk drives are literally aerodynamically designed to rely on the air to fly them at the right height over the platter. The platter's spinning drags air with it, so creating enough wind for the heads.
The specs for disk drives specify a maximum operating altitude, something like 15,000 feet. They won't work in a vacuum.
Infuriate left and right
This is exactly what Iomega was selling about 10 years ago (before they cashed in with the Zip Drive). They were called Bernoulli cartridges, were the size of a small pizza box, and held about 100 megabytes of data.
They were famed for their reliability, but at the time, IBM's Winchester drive system was cheaper to license and deploy for Iomega, so their Bernoulli never got adapted to using metallic media.
Now times have caught up with the Bernoulli design. I think this is going to be huge.
Fundamentalism is a crime against humanity