6TB helium-filled hard drives bump capacity, decrease power use | Hacker News Reader

6TB helium-filled hard drives bump capacity, decrease power use

computerworld.com

107 pointsShivetya82 comments

I would like to understand why they are using helium at all instead of a several other plausible alternatives. The article offers the following rationale for using helium in this capacity;

>>>At one-seventh the density of air, helium produces less drag on the moving components of a drive - the spinning disk platters and actuator arms -- which translates into less friction and lower operating temperatures.<<<

This appears to be a slight misunderstanding as the density of any gas is variable and we have found that it is directly proportional to the unit weight of the substance, the pressure and inversely proportional to temperature. It is not fixed. They could have easily achieved the result in multiple ways. For instance, to reduce density it might have been easier to simply reduce pressure and then create a hermetically sealed chamber with a reasonably inert gas with low atomic or molecular mass such as nitrogen. This should have saved them a lot of trouble as helium is actually pretty hard to contain. An alternative nitrogen based solution would also have been good enough for a long enough period of time and it could have been achieved in a much less expensive manner. (they say in the article that successfully trapping helium for a long enough period of time required a decade of research. This is a relatively solved problem for gases such as nitrogen)

I am sure that the people at Western Digital can see something that I don't - I'm just curious as to what that insight is. What makes helium worth all of the trouble?

I'm guessing the fact that it's an inert gas is important as it won't be a risk (in case of fire or toxicity) in case the enclosing is broken.

Also I think the density was presumed to be at room temperature, reasonable pressure. Under these circumstances He might provide the least drag. Having a pressurized enclosure might make the drive more expensive as stronger materials need to be used, and it would also make it more dangerous in case of mishandling, so I think there was a limited range of pressure they could maintain in the enclosure.

Nitrogen is also inert enough for such use, it's easily contained and it doesn't require such measures.

>>>Having a pressurized enclosure might make the drive more expensive as stronger materials need to be used, and it would also make it more dangerous in case of mishandling<<<

Actually containing He at all would require these measures anyway, as He diffuses through solids much faster than air and as such cannot be easily contained. To quote the wikipedia article;

>>>One industrial application for helium is leak detection. Because helium diffuses through solids three times faster than air, it is used as a tracer gas to detect leaks in high-vacuum equipment (such as cryogenic tanks) and high-pressure containers. The tested object is placed in a chamber, which is then evacuated and filled with helium. The helium that escapes through the leaks is detected by a sensitive device (helium mass spectrometer), even at the leak rates as small as 10−9 mbar·L/s (10−10 Pa·m3/s). The measurement procedure is normally automatic and is called helium integral test. A simpler procedure is to fill the tested object with helium and to manually search for leaks with a hand-held device.

Helium leaks through cracks should not be confused with gas permeation through a bulk material. While helium has documented permeation constants (thus a calculable permeation rate) through glasses, ceramics, and synthetic materials, inert gases such as helium will not permeate most bulk metals.<<<

Further, high pressure containers are dangerous as they might have an outward failure. That is unlikely to be the case with a drive that is at say .5 atm as the pressure is directed inwards, so the worst case over is most probably a drive filled with air or a slightly crumpled drive. (as a thought experiment think about how submarines implode instead of exploding when the hull encounters pressures beyond what it can handle)

For what is worth, we're breathing more N than O - so toxicity is not an issue when comparing He to N.

The idea of using helium in a hard drive is old, goes back at least 20 years.

As I recall, the main reason for using helium is that it conducts heat better. Heat? Yes, the spinning drive acts like a centrifugal pump which moves the gas and heats it.

Helium gets you low density at standard pressure. I am guessing that they prefer not to construct hard drives as pressure vessels. Probably that makes thermal conductivity, among other things, a problem.

The density of helium is less than 1/7 of the density of nitrogen, which is a good approximation for air.

Two ways that this is helpful come to mind:

1. Aerodynamic effects like drag and vorticity are largely governed by the Reynolds Number, which holding all else constant, varies linearly with density. Changing the density of the fluid can drastically change the aerodynamic situation. See http://www.grc.nasa.gov/WWW/k-12/airplane/dragsphere.html for an interesting and accessible example.

2. "Flow-induced vibration" is caused by the swirling gases crashing like stormy seas, making platters vibrate and heads flutter. Switching to helium cuts the momentum in these turbulent flows to a seventh of what they would be with air.

... to reduce density it might have been easier to simply reduce pressure and then create a hermetically sealed chamber...

Disk heads "fly" above the surface of the disk on a cushion of gas that's pulled under the head by the spinning disk. If there isn't enough gas pressure inside the disk enclosure, this method of locating the heads won't work. Wikipedia actually says:

If the air density is too low, then there is not enough lift for the flying head, so the head gets too close to the disk, and there is a risk of head crashes and data loss. Specially manufactured sealed and pressurized disks are needed for reliable high-altitude operation, above about 3,000 m (9,800 ft).

http://en.wikipedia.org/wiki/Hard_disk_drive#Integrity_and_f...

Atmospheric pressure on the top and bottom faces of a 3.5" drive case (~22 square inches) is in the neighborhood of 300 pounds each, and there's still more on the front, back, and sides. A square box of thin aluminum is probably not a great starting point for building a pressure vessel. If the box is filled with a gas at close to atmospheric pressure, you don't have to worry about it.

Helium's specific heat is five times that of air, so it probably offers better heat conduction than air at 1/7th atmospheric pressure would.

Shot in the dark, but might there be advantages simply because there are more molecules flying around than with air? 10nm ground effect flying probably isn't trivial.

I'm sold, but I have an additional question though. Can you please explain why hard-drives follow a flying head approach? Why don't they just perpendicularly fix the reader above the disk, rotate the disk and then use a linear actuator to move the reader back and forth - exactly like optical drives?

I'm guessing that the status quo exists and that my suggestion is inherently flawed because of the precision required to maintain the head at 10 nm above the surface and as such letting fluid dynamics wrestle with gravity might do the trick for you (given that you can make the spindles in a precise manner that the forces cancel out in a way that ensures that the head is exactly at the desired distance above the drive)

That leads to another thought, would the following conjecture be in the realm of possibility? Imagine that you have coated the disk with mystery thing X - which is an etched semiconductor that does something very magical. When an incident beam of light strikes a point on the coating, it releases electrons, which are then controlled through magical mechanism Y etched on X to create a magnetic field that flips the bit. If you do this then you can remove all of this stuff and it might lead to some gains.

"Can you please explain why hard-drives follow a flying head approach? Why don't they just perpendicularly fix the reader above the disk, rotate the disk and then use a linear actuator to move the reader back and forth - exactly like optical drives?"

The head is only "flying" relative to the platter. A hard drive works pretty much like you describe -- a short stack of fast-spinning platters and a read/write head on a swing arm that pivots back and forth over the platters.

A pic: http://msuweb.montclair.edu/~maciakl/blog/img/hd.jpg

They are child's play to take apart (http://turmon.org/Images/celeste-disk.jpg)

I think you should be careful about taking the notion of 'flying' the head too literally. Think of the air as a 'spring' which is holding the head at a precise distance above the platter, the head is pushing "down" and the spring is pushing "up", so if the head is too far away its down force is stronger and it moves closer, if the head is too close the air's "up" force is stronger so it moves away. This is simply a very precise way of placing the head over the platter, which is necessary because the shape of the magnetic field is a sphere and the 'circle' of its intersection with the platter is determined by the distance from the platter by the head.

In terms of the mystery thing 'x', most of what you might think of as candidate materials have been tried. The challenge is to reliably (well at least reasonably so) flip the state of a bit on the substrate in about a microsecond. And do that as cheaply as possible.

It is exceptionally challenging engineering.

Your description is a much more precise way of looking at it. I was trying to reach for the same concept when I talked about fluid dynamics and the forces acting on the head cancelling out so that it rests at exactly the right distance, but I lacked the clarity to state it so neatly. Thanks!

>>> It is exceptionally challenging engineering. <<<

It's also very beautiful... When you look at something like that it's a bit like looking at a work of art - I don't know how to explain it but it is awe inspiring. I wonder how people end up working on such things... How do you get to the point where you can raise your sleeves and create a system as beautiful as this?

I don't think any one person can create a piece of engineering as complex as a modern day hard drive. There are just too many disciplines involved: Physical, electrical and embedded software engineering, applied physics, information theory for encoding the data and recovering from errors. But certainly a single person or group in those fields can be responsible for specific breakthroughs.

Can you please explain why hard-drives follow a flying head approach? Why don't they just perpendicularly fix the reader above the disk, rotate the disk and then use a linear actuator to move the reader back and forth - exactly like optical drives?

The read/write arm needs speed and extremely fine-tuned movement that can't be had from a mechanical actuator. See: https://www.youtube.com/watch?v=Wiy_eHdj8kg

Re: the last part about etched semiconductors:

Keep in mind that the size of a magnetic grain (which stores 1 bit, roughly) is about 8nm. Compare that to the current cutting edge lithography scale of 14nm, which is the size of the semiconductor 'wires'.

Thus, a bit is smaller than the size even of the wires in any semiconductor circuit we can build, which means a system as you describe would have much lower storage density.

Helium is nearly 7 times better at conducting heat than both air and nitrogen [0], lower density [1] and about the same viscosity [2].

[0] http://www.engineeringtoolbox.com/thermal-conductivity-d_429...

[1] http://www.engineeringtoolbox.com/gas-density-d_158.html

[2] http://hyperphysics.phy-astr.gsu.edu/hbase/tables/viscosity....

Honest question: if air friction reduction is so important, then why not push for zero pressure drives?

Some pressure is required to keep the head flying above the surface of the drive. Removing all gas from the drive gives you a very nasty z-axis control problem.

I wonder if hard drives were part of Elon Musk's inspiration for the Hyperloop skids.

You would need much better seals to resist the pressure difference.

Air pressure is 1kg/cm^2 ... so you will need to make a hdd that can operate under 400 kg of external load. Getting to the point of vacuum is expensive too. And hdd already require enormously precise manufacturing process.

Same reason the US flag was 'flapping' in the 'wind' on the Moon. ;)

Without any air, there's nothing to stop the head from travelling without running into control issues. The gas molecules actually act as a buffer preventing too much movement. The head will flap about with the slightest motion like a flag in a vacuum.

Why not hydrogen? We are running out of helium. And it is in such minuscule amounts the fire safety should not be a concern, and for most of the stuff inside the drive hydrogen is chemically inert enough.

Because you at the very least you are going to have problems keeping the hydrogen inside the drive.

And there are probably even worse potential issues that my absolute lack of chemistry and materials knowledge can not predict.

Hydrogen is notoriously good at diffusing through seals, and even through metal. Negative internal pressure does make this harder, but it might be a problem. Handling hydrogen is generally a hassle, so it was probably chosen to make life easier/cheaper at the factory.

>Hydrogen is notoriously good at diffusing through seals

So is helium. That's why you can't cash in on the impending global helium shortage by buying up helium tanks.

We could always just Kickstarter a fund to buy the US government storage facility.

> Hydrogen is notoriously good at diffusing through seals, and even through metal.

Exactly this. Efficient hydrogen storage is a huge open research question. Even keeping the helium sealed can be difficult enough.

Can't people transmute some helium?

They can, it's just very, very, very expensive.

Nah... It should give you even more energy than you put on it, how can it be expensive, instead of a wealth source? Maybe we should should put some more money into helium creation research.

Perhaps due to hydrogen embrittlement of the metals?

I don't think that would be a big problem. Hydrogen embrittlement lowers the strength of the material, which I don't think is a major problem with hard drives. They're really never in any load-bearing conditions and you try not to drop them to begin with. A random thought that I had may be that the hydrogen interacts with the various metallic alloys in the platter itself, potentially forming some sort of precipitate that negatively alters the magnetic properties of the platter itself.

> hydrogen is chemically inert enough

In addition to just keeping it stored (as others have mentioned), hydrogen can have huge structural effects. It doesn't matter if it's chemically inert if it ends up diffusing through your metal and forming voids and the like:

http://en.wikipedia.org/wiki/Hydrogen_embrittlement

We're not running out of Helium, that's a common misconception. The price just doesn't reflect the extraction cost at the moment and it's not being extracted from the natural gas resources it is found in.

Do you have a good read for this? I'm always telling people around me that we're running out of helium.

The wikipedia article is a good start, follow through to the sources for in depth information:

http://en.wikipedia.org/wiki/Helium#Modern_extraction_and_di...

"According to helium conservationists like Robert Coleman Richardson, the free market price of helium has contributed to "wasteful" usage (e.g. for helium balloons). Prices in the 2000s have been lowered by U.S. Congress' decision to sell off the country's large helium stockpile by 2015."

I'm a bit skeptical of the quoted numbers, but even if they're a magnitude off there's still plenty left to be extracted. The US alone apparently still has enough proven helium in natural gas reserves for the next 40 years of worldwide consumption, and estimated unproven reserves for the next 40.000 years:

"Diffusion of crude natural gas through special semipermeable membranes and other barriers is another method to recover and purify helium.[84] In 1996, the U.S. had proven helium reserves, in such gas well complexes, of about 147 billion standard cubic feet (4.2 billion SCM).[85] At rates of use at that time (72 million SCM per year in the U.S.; see pie chart below) this is enough helium for about 58 years of U.S. use, and less than this (perhaps 80% of the time) at world use rates, although factors in saving and processing impact effective reserve numbers. It is estimated that the resource base for yet-unproven helium in natural gas in the U.S. is 31–53 trillion SCM, about 1000 times the proven reserves."

what a relief, thank you!

What? No. You're completely misreading this. The situation with helium wouldn't be as dire if we were recovering it properly (which we're not, because of the economics of boneheaded political decisions as wikipedia describes). But once those deposits run out, the nearest economically viable source of helium is Jupiter. Seriously, no joke. This is a crisis.

We could theoretically harvest the solar wind for alpha radiation. It will be much more efficient near mercury and you could use all that energy to vaporize rock and climb the gravity well with the cargo.

This seems like a Simon-Ehrlich wager case. We don't know of an immediate way to stop the crisis, but does that mean we really will run out of helium before someone discovers a better method of production?

I am not an economist and I'm not going to bet $10k on it or anything, but I think it's reasonable to expect something to change the situation.

There is no method of production. None, zero, zilch.

Unless you mean, say, production of ionizing alpha radiation, capture and containment, in which case you're using up another scarce resource of radioactive source products, generating far more radioactive waste then existing dirty nuclear plants, all for an amount of helium that you need a laboratory to measure.

Or maybe you mean by fusion? In which case you'd need a fusion reaction which yields stable helium isotopes which can be extracted without including radioactive byproducts. The difficulties here make interplanetary trade seem like nothing.

Economics is not a magic bullet. You actually need viable choices first, which we don't have here.

I suppose if we get better at fusion we could electrolyze the oceans and fuse the hydrogen into helium.

Sorry could you explain how I'm misreading it?

As far as I can tell, all natural gas resources has a percentage of Helium. Since we won't be running out of natural gas anytime soon, we won't be running out of extractable Helium. It just isn't extracted right now because the US is dumping their stockpile. By 2015 this should be done, prices should spike because production was stopped and will take some time to restart, and then it'll go back to "normal".

Hydrogen damages metal structures.

I am looking for a term - a good word for the feeling I got reading this article. It's a bit like the Iain Banks term "swim" - the lizard brain feeling for a human on a distant planet suddenly wondering how it got here

I am looking for a term describing the realisation that your toolset is ridiculously, hopelessly inadequate. I have happily taken hard disks apart and pulled at the heads and platters and thought "I have seen therefore I understand" - and it's articles like this that give me $TERM - the feeling that I have seen, I have had the Press Release explanation, but I am not even close to seeing or understanding

Ahh, thanks for reminding me of that term; I've desperately had it on the tip of my mind for a few days now!

Wait for 60TB HAMR

http://storageeffect.media.seagate.com/files/2012/03/perpham...

Netflix, which uses HGST high-capacity hard drives in its data centers

I thought Netflix uses Amazon AWS ?

probably talking about the openconnect CDN, which I don't believe is AWS based. One of the options for openconnect is for the machines to be hosted at the ISP, so that part definitely is not AWS.

Part of their CDN uses custom boxes based off Backblaze pods. http://gigaom.com/2012/06/06/why-netflixs-cdn-should-scare-t...

Are HAMR drives commercially available yet? Wiki makes it sound like 60TB HAMR drives will be readily available at low cost for consumers by 2016 or so, do you think that's likely the reality of things?

Seagate is suggesting the first HAMR drive will be a 3.5 inch Barracuda of 7.5TB

I think they are taking it gradually.

The problem with HAMR is evaporation of the surface film.

Amazing technology in due in 3-5 years? Probably not.

For computing power, it would be ridiculously expensive to try and use AWS to stream all their video.

Netflix has CDN's they put at peering points, you can read about it here @ https://www.nanog.org/sites/default/files/tue.lightning1.gen...

But you can also check out the Netflix open connect site @ https://signup.netflix.com/openconnect/hardware

IIRC the latest version of the CDN is 100% flash, though.

Yeah, they talk about that on slide #9.

Their newest boxes are 100% flash, but they're still rolling out the older spinning-rust designs. The all-flash boxes are for the most frequently requested movies, in the locations where there's enough traffic to justify them.

Hopefully this doesn't accelerate the likelihood of a helium shortage. Helium, as a abundant as it is in the universe, it is a finite and important resource on the earth.

A lot of scientific and medical equipment depends on it.

Vaguely related - super conduction at warm temperatures (haha, it's still pretty chilly around those windings but liquid nitrogen isn't too expensive) isn't too far off according to MRI manufactures and the various articles I have seen here in recent times. At $100k ish per quench, the service contracts sure looks like a good option when you have 3 or 4 of them in a year!

The thing that jumped to my attention was that this drive has 7 platters. I've heard that as platters go up reliability goes down. Can anyone who knows more about disks comment on this?

Seagate are going to ship SMR HDD in the next 6 months. With BPM is still way off on any RoadMap.( Correct me if i am wrong ) The other alternative is HAMR. But do HAMR works with Helium Filled HDD? My imagination is that the laser in HAMR would set off Rampant fire.

And to be Honest, even with that theoretical 50 - 60TB HDD from HAMR HDD isn't enough to fit our Cloud Platform needs. The amount of data we create everyday and Backups we need.

Do we have anything in research for an HDD that is capable of 100- 1000TB?

You are mistaking helium (inert,fire suppresant) for hydrogen (flammable).

I suppose that if they are sure they can hold the gas in the chamber for at least 3 years (typical depreciation rate for drives) they will do fine.

It will give a challenge to the data recovery guys, now having to run the drives in a helium sealed working chamber.

It will also reduce the lifetime of disks, making the best case near the average.

That may not be much of a problem for medium and large business, but small business and consumers won't like the trend.

It won't affect them. The point is to reduce power consumption and put the platters closer together.

Neither of those are a problem for data recovery.

This is pretty cool, but I assume the increase in productional complexity means consumers won't benefit from this for quite some time.

It's either cheaper or it isn't. I don't understand this "but you make it up in volume" logic.

It's cheaper to make a greater volume of a product since you pay once for tools and your factory, which can make any amount of a product.

Like everybody who uses Netflix?

What was the major issue with sealing He in a HD that it took 10 years to figure out?

It's very hard to keep helium in anything. Hard drives are not sealed, they have (filtered) holes to allow for expansion / contraction of the internal atmosphere caused by heating / cooling cycles.

Helium is not a renewable resource, and we're running out.

I keep on thinking that we've reached the limits of traditional drives, and hoping that we dump a lot of investment into SSDs and new SSD architecture. It just feels weird to have something so mechanical in a computer.

Theoretically we're running out of Helium, just like we're theoretically running out of oil, but actually we've still got plenty left. Because of the US dumping their stocks the price has been artificially low making it unprofitable to extract from natural gas which contains a lot, and it has been just released in the atmosphere because of that instead of contained. As soon as prices start to reflect the extraction price there won't be a problem anymore.

This makes me wonder if these new hard drives will have a limited shelf life. I know they have spent so much research on keeping the He in, but it's more likely that they have just slowed down the leaking, so will this hard drive still be spinning in ten years? Some hard drives are expected to keep running for at least that long.

Also, I worry about air pressure changes. If a particularly nasty weather system came along with a very low pressure centre, and the hard drive is completely sealed, then it will have to either flex to accommodate the He expansion or be very rigid indeed. If it flexes, does that de-calibrate the head alignment? Is the case strong enough to deal with 200N force pulling it apart?

I would imagine they include a bladder of some sort to handle expansion and contraction due to temperature changes while the drive is operating, in order to keep the internal pressure fairly constant. That's probably a wider range of pressure change handling than weather systems would require. (There was a time when I knew how to do the math for that, but I haven't used my MechEng degree in a long, long time.)

"the main issue being finding a way to reliably keep the helium from leaking out of the devices"

http://gizmodo.com/5943238/helium+filled-hard-disks-less-dra...

I don't think it took them 10 years to figure it out. It's probably only now that the cost is justifiable when compared to other techniques that increase capacity.

The problem with Helium, that it is small: therefore even glass acts as a porous sponge from that point of view.

Helium will inevitably diffuse and leak through over time.

So you think this isn't a viable step or you have doubts about the long term?

Notice on the packaging says: If you start talking with a high pitched voice, replace the unit immediately.

EDIT: It's a joke, lighten up people..., cause you will with this drive :)

EDIT2: And in the case the drive fails, just open it and get a guaranteed 1 minute of fun.

Ah, I was going to make a joke about whether these drives would sound all squeaky and weird, but I think you indirectly beat me to it :-)

If I would want a comment thread full of jokes I would go to reddit.

Reply on news.ycombinator.com