Helium: Can it prevent the onset of Shingles? Filling disk drives with helium gas delays the time when shingled magnetic recording has to be adopted. That's because helium gas-filled drives can cram more platters into a disk drive enclosure, thus increasing capacity without having to alter the data recording method. The current PMR (Perpendicular Magnetic Recording) …
I don't see why using helium in a drive (in itself) enables more platters to be used. I suspect that the drag factor means that more platters can be used and not cause excessive drag leading to unacceptable drive power requirements. Has anyone got a clearer idea?
In itself, filling with helium should have no effect on the manufacturer's ability to pack platters in or what kind of recording they use. (If filled with air, there may be an effect due to oxygen reacting with the surface of a read head or disc surface, but I doubt it.)
Also interested to hear more. The article wasn't super-informative on that front.
I assume they need to change the shape of the disk heads to maintain the correct float distance in helium, too - presumably by making them bigger, or by giving them teeny-tiny wings (shaped for confidence and comfort, natch).
I don't get it either - just looked-up the viscosity of Helium gas, which I had expected to be smaller due to its lesser degrees of freedom, but its the same more or less as Nitrogen. So I can't see how the platter friction will have been reduced.
My aerodynamics is limited, there are more factors in there than pure viscosity, but to me it looks mostly like a pure shear load of head plane against disk plane - which is viscosity by definition.
I'm sure they won't have spent years developing stuff that doesn't even work on paper, so what is the magic factor they are able to manipulate in this technology then?
I don't think that's the answer, the drag heating will be the same, fuelled as it is by viscosity - and note that it heats the gas not the platter or head.
The thermal conductivity of Helium is indeed 5x better but this is a forced-air cooling as the gas will be whipped around - the heat will be conducted away by bulk mass flow not by diffusion. OK there is diffusion across the boundary layer to the casework, but this should be a small term.
So, thanks for the suggestion, but the question remains open as far as I can see it.
While shrinking the hell out of disk drives may be good for portable devices I can see no purpose in having a marvel of miniaturisation in a basically empty pc chassis.
Even server units cpu's seem to get agoraphobia these days.
In fact I've just looked over the back of my monitor and there's room for several thousand platters even if it was pushed up against a wall.
Eight inch platters????????
http://en.wikipedia.org/wiki/Storage_Module_Device
I worked during one summer (1985) with machines using these drives. These were the washing-machine sized things you see in films from the early 80s and late 70s. The (removable) platters were 14 inches in diameter, stacked five high (80MB drive).
I used to have a (broken) widget taken from the spindle of one of these drives - it was the electromagnetic brake that enabled the controller to stop the platters in a timescale shorter than weeks, and in this case it had failed "off". Five 14-inch disks of millimetre-thickness aluminium have a substantial amount of angular momentum and the interior of such a drive causes very little drag.
Helium being what it is, the only way I can see such a drive working is if the drive's platter chamber is properly hermetically sealed. This would require a chamber similar in construction to that of a light globe or radio valve which has a fully sealed envelope. Electrical connections would enter the envelope and connect to the electrodes in the same manner.
Such a connection through the glass requires a special seal where the glass must properly wet the surface of the metal wire (this can be a manufacturing problem). Perhaps also connections could be done by electrical (capacitive/magnetic) coupling through the glass or whatever non-conducting material is used for the chamber. For instance, the motor stator could be on the outside of the chamber and the rotor on the inside.
Frankly, it'll be interesting to see what they come up with, as from my experience, keeping helium in any container is somewhat of a problem.
All I hope is that these drives will be more reliable than the current batch of failures that I've had over the last couple of years.
BTW, the shingle recording method reminds me vaguely of the recording system used in VCRs where the hi-fi stereo channel is laid down under the video tracks.
Or, you could just start pricing SSD's sensibly and then putting your research into SSD instead, and then everyone would be happy.
I'm honestly holding onto money from a year, eighteen months ago to replace one of the two 1Tb drives in my laptop with an SSD. Given that I can't get past 256Gb without going into the realm of more than my laptop is worth, I've continued to hold off.
If you're really worried about speed, you'll be using SSD's by now anyway. If you're not, you're holding off for SSD's to become more affordable. All the stop-gap and major research required to get to the next stage of magnetic disks isn't really worth it unless you have an urgent need for it. Just make a cheaper SSD. A 512Gb SSD should cost no more than 1.5-2 times a 256Gb SSD, especially if you have models that ALL use the same board but just extra chips (or different capacity chips) on them. Fact is, that's just not true at the moment. As soon as it is, it'll mean that 1Tb SSD's will become viable and then we'll ALL have one.
I started with a 20Mb hard disk back on a 386 (and that, we paid a fortune to upgrade from the stock 10Mb disk). I'm not stranger to good-old-spinning-metal. But the fact is that I'd be loathe to buy anything with SSD's in at the current prices, but - like 99.99% of all people and networks - don't need to go mad on hard disks when a cheap one will do. If this raises the costs of hard disks for a speed improvement inferior to an SSD upgrade - and SSD doesn't change, nobody really wins. But if they could just forget about magnetic recording entirely (like audio cassettes, video tape, professional audio, and a whole host of other sticklers for magnetic recording have already done, the only real exception being backup tape that can take DAYS to do its job) and instead make SSD's a bit cheaper, everyone wins.
To join onto my post, I actually liken it to CPU's a few years back.
Sure, it was possible that we could all eek out another 500MHz out of the top-end chips and do things that way. But, actually, most people never hit that kind of bottleneck. And, sure, we could invest in hyperthreading and get more done that way for certain compromises, but most serious computers just switched it off. But you could just slap another CPU core down on the die at the more ordinary clocks and it's actually MUCH more useful, and much more valuable to the average user. And now dual-core is seen as a minimum, quad-core is pretty ordinary, and even 8-core machines no longer raise eyebrows. Meanwhile, all the hyperthreading and single-core-but-3.5GHz guys are stuck in very specialised niches where they can be.
All that research is going to be wasted in just the same way, unless you can provide real VALUE for doing this. A slight increase in disk speed at the cost of writes? Few are going to take that payoff over just using SSD which improves almost everything.
A quick check on Crucial shows their M4 SSD at £148.82 for 256GB, or £296.84 for 512GB, so pretty much exactly twice the cost. The premium for 512GB over 256GB has been more or less a factor of two for years, certainly since I bought my first one getting on for three years ago now.
GJC
"Or, you could just start pricing SSD's sensibly and then putting your research into SSD instead, and then everyone would be happy."
The pricing hierarchy is due to the usual stuff - lower demand & volume for higher priced stuff, lower yield on more advanced tech, higher "bin" parts going into the fancy versions. I've seen press coverage claiming new SSD's at $600 for a 1TB unit, but even if that's true then 2TB units will become the ones that cost 4x something of half the capacity - for example, because in a 2.5" package you'd need to use 64GB NAND chips, where everything currently is centred around 32GB, and you'd probably need a new controller design.
Your suggestion that there's no point in putting money into anything except SSD ignores the current limitations on SSD's, primarily that the driver of lower price is process size reduction (apparently focused particularly on consumer grade MLC NAND), and that impinges on various aspects of performance, in particular endurance. With over provisioning and other techniques you can make the problem disappear for desktop users, but in an enterprise application that doesn't wash, and we're talking about much more expensive SLC NAND, in which case the balance of performance and cost indicates that there's quite a bit more life left in spinning rust yet.
From a "professional" perspective, we also don't have any track record for how well SSD storage copes with age in real time (and to an extent the ageing of early NAND is not much of a guide to how data written to upcoming 14nm chips will last). When you can get and believe a good long SSD warranty then that will be a sign of maturity that might herald the end of HDDs. At the moment I'm happy to use an SSD in a properly backed up client machine, but wouldn't dream of using one for enterprise storage or in the data centre (not that I'm faced with making such choices, I might add).
What happens when the helium runs out, thanks to that bastion of the free market, the USA, dumping the world's supply at rock-bottom prices?
"This opens the way to having more platters in a standard drive enclosure, seven instead of four in a present day 4TB, 3.5-inch form factor drive, giving you 75 per cent more capacity and no diminution in write speed"
Even better than no diminution (great word, aiming to use it in the real world today) in write speed, if managed correctly it can lead to an increase in read/write speed too.
There exists the risk of other manufacturers going straight from the actual technology to Phase Change Memory or similar methods, making HDDs obsolete and preventing Hitachi from recouping their investment. Of course this is also true regarding SMR or any other technology that improves 'mechanical' disk drives. The closer current technology is to its physical limits, the higher the stakes and the bigger incentive for other manufacturers to develop non-magnetic storage. The next years will be interesting -in the Chinese curse kind of way- for HDD makers.
Yes it would be a terrible sight....
Ones beloved big beige box, plummeting in flames from the ceiling towards the floor.....
Crashing in a burning wreck near the power point it was once tethered......
"Oh the huge manatee......" - You can wail, as the moths and spiders jump and run for their lives....
In fact you had probably better run for your own life... the combustion temperature of helium is not that far below the fusion temperature....
And if your Discus Flambe, goes nuclear fusion.... "Ooooooooo Dear.... - you had better be a long way away when that happens - like in the next solar system."
So I am calling for an immediate ban on the development and use of helium filled drives, simply because of the fact that all it takes is for one drive to go critical and it's the extermination of not only all life on this planet, it's the ending of the planet.
Satan - even she is doomed.
Rather than faffing around with helium, surely they could just pump out all the air and leave a vacuum in drive housing - there would be no resistance at all then. Must be possible. I have one of those radiometer things (http://en.wikipedia.org/wiki/Crookes_radiometer) on my windowsill and the lack of gas in the globe certainly lets the vanes spin pretty damn fast.
The heads in your hard disk "float" a very small distance from the surface of the platter, kept aloft by aerodynamic forces.
If they touch the drive surface, that's a "head crash", which usually rips the heads off and gouges furrows in the platter surface. That's generally considered a Bad Thing.
So going to vacuum would need some other way to float the heads.
Presumably something about the aerodynamics of helium gas makes more, thinner platters possible.
You cannot operate a hard disk drive in a vacuum as they require the read/write head to "fly" over the surface of the platter using aerodynamic effects. Even if it was possible to engineer a lightweight head and arm assembly accurately enough to skim just a few microns above the platter surface (which is extremely unlikely), it would be impossible to provide the sort of cushioning effect against mechanical shocks that a thin film of gas gives.
Why don't people do just a little research before posting comments like this? I've lost count of the number of times this particular suggestion has been made at various times.
This post has been deleted by its author
Yes, there is a problem with helium, it leaks through lots of things (its molecule being quite small). Everyone knows this, and putting helium inside a drive can be a bit problematical. That being said, there IS a solution, replenish the gas that leaks out. Pretty simple if you ask me. There are lots of things that have alpha decay (which is essentially helium nuclei in disguise) and just pop a little bit of a nice isotope in a corner of the drive. Alpha particles don't need much shielding and many have long half-lives so there shouldn't be any problem.
Or as the saying goes: "How hard can it be?"
Indeed there are some radioactive consumer devices, but the rate of activity is tiny. To put this in perspective, there will be of the order of 10^21 helium atoms in a 3.5" disk drive enclosure. Replenish just 1% of those a year, and that's about 10^19 alpha particles required. If we use Americium 241 as a source - as commonly used in smoke detectors - then each devcay yields 5 alpha particles. So that means approximately 2x10^18 nuclei will have to decay in a year. That's about 0.5gm of decayed Americium per year. With a half life of over 400 years, it will require over a hundred grams of the element. In comparison, an average smoke detector only has about 0.3 micrograms.
So the equivalent radioactivity of several hundred million smoke detectors would be required to replenish just 1% of the helium required for one of these drives in a year. It would also generate about 10 watts of radioactive power. Released into the environment, that could rather spoil somebody's day...
Ionising radition made up of high energy alpha particles in a sealed box with highly sensitive semi-conductor heads and minute magnetic regions. What could possibly go wrong?
nb. it would take quite a lot of something highly radioactive to produce alpha ions in any appreciable quantities. It's a non-starter, but a nice bit of lateral thinking...
...of certain posters (or maybe they just omitted the sarcasm tag).
But just in case someone doesn't get it, there has to be a gas of some sort inside a hard drive to "fly" the heads above the platters instead of letting them actually touch, and it's the friction between the platters and the gas that drags that gas along with the platter surface that creates the "wind" that lifts the heads.
thinking it was about the medical condition often referred to as shingles, but more officially called herpes zoster. I've had it. thus the 'eagerly' up there. Nasty crap. If you had chicken pox as a kid, the older you get, the better chances you have for the virus to flare up again as it never goes away. It just becomes dormant. LIke I say, nasty crap and you WILL be going to the hospital for the pain alone.
Gaseous hydrogen is even harder to contain than helium so the seals in the hard disk enclosures would have to be improved at great expense. The use of hydrogen in power station generating sets for cooling is based on good ventilation and gas detection systems that report buildups of dangerous levels in the turbine halls since they are not totally sealed and leakage is expected.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
