What happens when the helium inside an enclosure leaks out?
Data centre goes squeaky.
Assuming that it is the same pressure as outside and air can't get in - then why exactly would it escape and create a lower pressure inside the drive?
^ Given heat cycling internally of the drive, nevermind atmospheric density changes, I'd imagine that it's almost never at the same pressure as the external surroundings.
Plus, helium is weird crap, and can go through stuff that you'd think would be utterly solid.
The point still remains though - if it all escaped and air couldn't get in you would end up with a vacuum.
If there is any avenue for helium to leak then some will be expelled as the inside of the disk gets hot. It will then float away in the denser air.
When the disk cools again the same avenue could be too small for air to be sucked back in. The pressure inside the disk will be lower - but will still be uncontaminated helium.
That process probably occurs at some factory initial burning in stage - so wouldn't be repeated in a user environment unless the disk gets hotter.
Whether it would eventually keep leaking due to Brownian motion - and keep reducing the internal pressure - is one for the physicists.
If the escape avenue also allows air to enter then the helium inside will be contaminated over time - even if only by Brownian motion.
Given the pressure differential (tiny) and the lack of thin permeable surfaces to percolate through, the short answer is that the drives may leak their helium but not on a meaningful timescale (ie, not within 5 years, possibly not within 10). It _will_ percolate through eventually, but eventually is a long time.
The interesting thing I found playing with He-Ne lasers (mostly running at a soft vacuum) is that what kills them isn't so much the helium leaking out, as much as nitrogen leaking IN (characterised by a violet discharge)
work it out....
you end up with a MASSIVE failure, since the space inside the drive is now way more turbulent and "sticky", head crashes, you name it...
Helium is this amazing stuff that is actually lighter than air....
using your assumption about air pressure, it should be possible to stick helium in an open glass and it would still be there 10 minutes later.
it would continually be trying to leak out and create a vacuum
The problem with your example is that the air can get into the glass to replace the helium. In the drive, air can't get in to replace the helium so any losses would create negative pressure inside - which would serve to keep the remaining helium in place.
It's the same concept as the trick you can do with a bottle of water - fill it up, put some small holes in (with the cap on) and the water will stay in the bottle.. remove the cap and the water will spurt out.
I wonder if it would be worth having double thickness drives. The lack of duplicated casing and circuitry would presumably allow more than double the number of platters?
Perhaps the size of a CD drive bay, as that's already a standard?
"Perhaps the size of a CD drive bay, as that's already a standard?"
That bay size already accounts for the "hot swap" container in which the drive is fitted. IIRC it was originally based on a 5.25 hard disk footprint.
The problem with the hypothesis is that this means a longer spindle and a taller spinning device which will raise the center of gravity and introduce additional stresses via torque. IOW, you increase the risk of wobbling.
"I wonder if it would be worth having double thickness drives. "
Nope. The bearings can't hold the stack rigidly enough
As for larger platters - remember Quantum Bigfoots. The platters are harder to make and any uneveness is exacerbated. It's been tried but is virtually impossible to keep reliable above about 3600rpm.
All these use cases are short term anyway. 16TB SSDs are already here (I have a couple installed) and prices keep coming down, whilst HDD prices and warranties are _STILL_ higher/lower respectively across the board than they were before the 2011 Thai Floods.
The extra money for SSD is made up for by greater reliability, longer life(*), lower access latencies and lower power consumption, so it's hard to make an apples-to-apples comparison between the two based on price alone.My feeling is that the break point for SSD adoption is about 3 times the HDD price and that's already been passed at sizes below 1TB (with SSDs now ruling the roost)
(*) When was the last time you saw a HDD with a ten year warranty?
> All these use cases are short term anyway. 16TB SSDs are already here
Sure, and it only costs as much as your house.
I would rather have 2 or 3 of the cheaper drives. I don't really need the extra speed and you can't trust any storage technology not to fail.
I was working in a PC build shop around the time these were in general availability. We had to RMA sooo many of them because they were unreliable garbage. The contemporary 3.5" models were way better.
I was talking about height rather than width - the platters could be the same size, just more of them.
And HDD's have their uses. My media library for instance - it doesn't need to be fast or last forever. My pretty much only concern when adding disks to it is GB per £.
I have no interest in spending thousands on a 16TB SSD when I could have the same capacity from HDD's for about £400.
I'd still much rather have an affordable 2Tb SSD.
As in MUCH rather.
That's the second article today on hard drives (Seagate, Toshiba) and I actually question why anyone is still pumping money into them, except to get the "last run" of hard disks out the door.
I do hope that these companies aren't spending all their time and money faffing about with helium.
I'm the other way around. There are plenty of packrats around who want to be sure what they have is still there in the event of a company going down or no Internet (remember, no Internet = no Cloud). Tape storage is out of reach for the consumer, so a way to economically store a lot of data (speed is not an issue, and pairing up helps guard against a catastrophic failure) is a boon in my book.
GIven the number of HDD failures I get to see where someone's tried to use them as archival storage, I don't trust 'em. If you want long term archival backup then use tape or ensure you regularly migrate your media - but bear in mind that HDDs will give you an undetected ECC error (corrupted sector read) about once every 45TB read, vs every 4500TB for tapes.
I'd suggest M-disk, but even the BD-R versions aren't really large enough.
If you're willing to part with the money, 2TB SSDs aren't that bad now. They cost about as much now as my 1TB drive did when I first got it. It hasn't been that long either.
They aren't dirt cheap but they aren't ridiculous either.
Some things are just utterly impractical (like tape)
Seagate seems to be the only real problem here. As long as you avoid them, you won't have much of a problem. Even if you do buy their stuff, it's only just a bit more of an inconvenience really.
Like I said, tape is out of reach in the consumer end. Has been since the QIC days.
Silent corruption I can deal with via error codes if necessary.
Like you said, BDs are too small for archival needs in the TB range.
And drive rotation is part of any archival plan. As long as it can be held to about one every 4-5 years, it's still within reason.
And large-capacity SSDs are still too expensive. They need to drop A LOT before they can fall into consumer range.