I was under the impression that helium tend to leak out of most things quite easily. How can they have a MTBF of 2.5M hours if the helium is a integral part of the disk?
https://www.hgst.com/products/hard-drives/ultrastar-he10Ultrastar He1010TB & 8TB
3.5-inch Helium Platform
Enterprise Hard Drive- Highest capacity – 10TB HDD, drop-in ready for mainstream enterprise
- Lowest Power – 56% more efficient than air-filled drives for lower TCO
- Most Reliable – Best-in-class reliability rating of 2.5M hours MTBF
You need clean, solid seals to do it, but it's not by any means impossible.
Trivia: helium is used for testing vacuum leaks. A mass spectrometer(!) is connected to the vacuum system, tuned to the helium signature. A helium wand (just a tank leaking gas out of a tube..) is waved around the vacuum chamber. When a signal is generated (helium concentration rises suddenly, then falls slowly), you know there's a leak around that location.
So if the vacuum system is rejecting helium from the outside very well, it will retain helium inside very well by the same fact. You just need atomically good sealing, that's all.
Tim
As far as I know, MTBF means nothing. They might as well make up the numbers, if they don't do already.
Do these drives cost less to ship?
I thought that helium like hydrogen will percolate through materials including glass and steel only a bit slower than hydrogen.
As far as I know, MTBF means nothing. They might as well make up the numbers, if they don't do already.
You are correct. In general, MTBF is complete frogshit for an entity like a hard disk.
What is important is
MTTF.
http://www.weibull.com/hotwire/issue94/relbasics94.htm
As far as I know, MTBF means nothing. They might as well make up the numbers, if they don't do already.
MTTF
d is the term normally used, Mean time to dangerous failure. This is usually calculated from the B10
d specification of a particular device or component. The B10
d is the number of cycles until a component fails dangerously and should be quoted by the manufacturer of the component or device. They get this mythical B10
d number by testing things until they fail (usually on the order of 20 million cycles). It's a statistical thing.
Just a ballpark guide to the reliability of a particular device or component.
As far as I know, helium will go through stainless steel unless the material is absolutely defect free (and I mean absolutely). Otherwise it can diffuse out through the grain boundaries, etc.
HGST
launched their first helium filled hard drive a little over two years ago, and there don't seem to be any reports of leakage from those.
Apparently a layer of graphene a single atom thick is completely impermeable to helium, so it's not impossible.
Do these drives cost less to ship?
Actually, they might. The link above reports that HGST's 6TB, seven-platter helium filled drive was 50g lighter than their 4TB, five platter air filled model.
So what effects would a user notice if the helium started to leak out?
So what effects would a user notice if the helium started to leak out?
Everyone in the data centre suddenly sounding like Mickey Mouse, maybe
Sorry, I'll get my coat
So what effects would a user notice if the helium started to leak out?
Overheating - AFAIUI one reason the helium is used because of its better thermal conductivity. I suspect it will also allow the heads to fly closer to the disk surface as well.
So what effects would a user notice if the helium started to leak out?
Overheating - AFAIUI one reason the helium is used because of its better thermal conductivity. I suspect it will also allow the heads to fly closer to the disk surface as well.
less friction. platters moving through air cause massive drag. in helium less.
don't forget this monster has 7 (SEVEN) platters and 14 heads )
First I thought that a (partial) vacuum would be both cheaper and more leakage-proof to make and reduce the air-drag/friction a lot, but then I realized that the heat transfer would suffer a lot... So there goes that good idea down the drain ;-)
First I thought that a (partial) vacuum would be both cheaper and more leakage-proof to make and reduce the air-drag/friction a lot, but then I realized that the heat transfer would suffer a lot... So there goes that good idea down the drain ;-)
That and I think it's the air (or whatever gas is in there) that 'floats' the heads above the surface of the disk as it spins. I'd expect no atmosphere would be bad for that as well.
-Pat
First I thought that a (partial) vacuum would be both cheaper and more leakage-proof to make and reduce the air-drag/friction a lot, but then I realized that the heat transfer would suffer a lot... So there goes that good idea down the drain ;-)
That and I think it's the air (or whatever gas is in there) that 'floats' the heads above the surface of the disk as it spins. I'd expect no atmosphere would be bad for that as well.
-Pat
That is a *very* valid point. It seems like I can't brain so good at midnight anymore - too old....
As far as I know, MTBF means nothing. They might as well make up the numbers, if they don't do already.
You are correct. In general, MTBF is complete frogshit for an entity like a hard disk.
What is important is MTTF.
http://www.weibull.com/hotwire/issue94/relbasics94.htm
MTBF for hard disk is equivalent to MTTF since a hard disk is not repairable. You could more accurately use MTBF on the computer system that the hard disk is installed in, but that is another discussion. In any case, MTBF is not a nonsense term but you need to understand what it means. It is the mean time between failures assuming that all disks are replaced at EOL (essentially end of warranty). MTBF of 2.5M hours does
not mean that the disk is expected to operate for 2.5 million hours (285 years) before failing. It means more something like if you have 1000 such disks running, then you will see one failure per 2500 hours, or 104 days. After the lifetime of the disks is used up (~5 years), you need to replace them all, otherwise the failure rate will be expected to shoot up. Really, the industry should use the term MTBF_SR (MTBF with schedule replacements). The internet archive seems to keep good records their hard disk failures and publishes lots of data about them for those interested.
First I thought that a (partial) vacuum would be both cheaper and more leakage-proof to make and reduce the air-drag/friction a lot, but then I realized that the heat transfer would suffer a lot... So there goes that good idea down the drain ;-)
That and I think it's the air (or whatever gas is in there) that 'floats' the heads above the surface of the disk as it spins. I'd expect no atmosphere would be bad for that as well.
-Pat
That is a *very* valid point. It seems like I can't brain so good at midnight anymore - too old....
I can't brain today - I have the dumb!
Seriously, as someone who's worked with vacuum systems in the past, evacuating the enclosure to reduce friction was my first thought, too. Then I remembered reading years ago that the head floats on a miniscule cushion of air that's dragged under it by the spinning platter, and that sank that idea right off the bat. I never thought of the heat transfer part, but it's very valid too. TANSTAAFL.
-Pat
Guys, look at the price!!!
It is way more expensive per GB than 3TB or 4TB drives.
And, a simple mechanical failure kills all 10TB data.
BTW, MTBF means nothing. I have never had any HDD survived to their MTBF. They usually die in 2 years at 24/7 operation, which translates to 17.5k hrs, only 1.46% of their rated 1.2M hrs MTBF.
Because MTBF != expected lifetime before failure. It means only failure percentage during some predefined time.
BTW, MTBF means nothing. I have never had any HDD survived to their MTBF. They usually die in 2 years at 24/7 operation, which translates to 17.5k hrs, only 1.46% of their rated 1.2M hrs MTBF.
Sounds like you need to take a hard look at your environment. I have had drives fail within 2 years, but certainly not all of them. Methinks you're abusing the hell out of your drives without knowing it. What kind of temperatures are they reporting? What is the vibration/shock environment like?
Everyone I talk to who complains about their mechanical drives consistently failing after 1-2 years is always either using a laptop (laptops should never use mechanical drives, it's just begging for failures) or they're using some horrible sealed enclosure that's cooking the drives at 55+ C 24/7.
If you keep a mechanical HDD below 35C 99% of the time and always below 45C, in a vibration/shock-free installation, it should last for many, many years. I have about 150 mechanical HDDs operating under my care, and I typically lose about 1-2 a year. That's a MTBF of about 1M hours, which falls roughly in line with the advertised values. Most of the drives are over 6 years old and working just fine. The oldest are over 10 years old.
G
BTW, MTBF means nothing. I have never had any HDD survived to their MTBF. They usually die in 2 years at 24/7 operation, which translates to 17.5k hrs, only 1.46% of their rated 1.2M hrs MTBF.
Suicidaleggrol beat me to it, just as I was replying to same quote .
Such a high failure rate, maybe unlucky. You have me worried now . I was considering a project where I need the HDD to be in constant use 24/7 writing multi TB a day and one of the things I was counting on at least a ~5 years lifetime (50% failure rate) to keep costs down.
I know manufacturers MTBF figures are not quoted for 24/7 continuous use, but I wouldn't have though that continuous use is going to be so much more stressful on HDD as to cause such a large discrepancy to manufacturers MTBF. need to research this further.
As far as I know, helium will go through stainless steel unless the material is absolutely defect free (and I mean absolutely). Otherwise it can diffuse out through the grain boundaries, etc.
Have you factored in the pressure inside the hard drive? How many defects could be reasonably expected? Calculated the diffusion rate? Calculated the resulting amount of time that it would take for the helium to diffuse out in a reasonable amount of time?
If the Helium is diffusing out, is that really a problem? Isn't it more the absence of oxygen/other stuff inside the HDD that is the goal here? If oxygen can't diffuse back into the HDD, what's the problem?
But no, you read once that helium at some unknown pressure will leak at some unknown rate through a grain boundary. I was going to make an LED blinky circuit, but then I remembered that leakage can occur on PCB surfaces. I probably shouldn't calculate whether those leakages will be significant right? I should just give up on the idea?
LED blinkies work at voltages < 5 V, so unless your negative and positive contacts are within 20nm of each other, you should be fine with a normal PCB surface.
Yes, it was a silly example on my part on how failing to factor in
quantities can lead to absurd leaps of logic. E.g.:
- "Current can leak over the surface of a PCB" --> "All PCB electronics is infeasible" (obviously ridiculous)
- "Helium can diffuse through stainless steel" --> "Helium-filled hard drives are infeasible" (implied by several people on this thread)
My cryocooler uses helium as its working fluid and has been sealed for a decade with no loss of performance. It still happily liquifies air when i turn it on. And this is helium undergoing constant over and under pressure waves. (it is a split-pair stirling engine configuration, the fun stuff you find on ebay.).
If it can be solved for that, it can be solved for a hard disk.