Slashdot Mirror
Western Digital Announces World's First 10TB Helium-Filled Hard Drive (techgage.com)
Deathspawner writes: Western Digital today announced a new, helium-filled enterprise HDD that allows for 10TB capacities without using the SMR method, sticking to industry standard PMR. SMR, or Shingled Magnetic Recording drives, can not typically be used natively by the OS or disk controllers, and instead often require extra software and/or firmware updates. This makes their broad adoption limited, since the drives are not drop-in replacements for the far more ubiquitous Perpendicular Magnetic Recording (PMR). WD's latest enterprise drive, sold as the HGST Ultrastar He10, uses the PMR storage method, and as such is a full drop-in replacement for any standard hard drive.
I have an SMR 8TB Seagate drive which is marked as an archival drive. It's worked well so far. The problem with Shingle Magnetic Recording drives which I have noticed is that occasionally the drive will "stall" while it rearranges data. This is probably extremely bad for some raid systems as the paranoid ones might think the drive has prematurely died. Still this drive was inexpensive for its size and stores a LOT of data which is handy for backing up my actual RAID NAS system. Just don't use a drive like this in your Raid or you might run into serious problems.
I worry about these helium drives leaking their helium eventually and dying. They claim to have a sealed unit where the seal will last for years which is hopefully the case but you never know...
"The benefit of using helium is that it’s less dense than air, putting less strain on the motor. End result is a five or six platter drive that can spin up to 7200 RPM on less power, while improving reliability of the drive (2.5 million hours MTFB)"
So im guessing if the helium did leak out you would probably just see a somewhat lower drive life.
I agree that Hydrogen sounds like a better approach.
1) orders of magnitudes cheaper
2) even less dense
3) much larger molecules, so leakage should be far less
Not inert, true, but that should be possible to deal with.
Nice thought on #2, but incorrect. Hydrogen diffuses through steel quite quickly.
Hydrogen is the bane of ultra-high vacuum (UHV) systems, which have stainless steel walls 1/2 to 1 inch in thickness. New system? Hydrogen comes out of the steel itself, as it contains some. But with time, that might be depleted, but no —more comes in from the atmosphere, or from any replaced components or new seals.
For UHV systems, helium is quite useful for finding leaks. Microscopic or even nanoscopic pathways for the helium atoms to make their way in. One frequently has poor base vacuum, and must hunt around blowing helium on suspected parts. These could be anything: micro-crack in a weld, stress-crack in a feed-through, an improperly bolted seal, a loose bolt. It can get very fiddly.
MTBF is NOT the same as expected/rated life. To equate the two is the oldest, most naive misconception in the book. MTBF gives the number of drive-hours between failures, IN A LARGE POPULATION of FRESH drives. For a 2.5 million hour MTBF, if you have 10,000 drives operating, then you expect one failure every 2500 hours. That's 3-1/2 months. Right from the beginning. It does not account for wear. In fact, the failure rate function will not be a straight line. It will rise as the drives age. When the failure rate read off that line becomes very large, you have reached the limit on expected life.
Expected/rated life is determined by analyzing wear factors. They usually don't spec this figure to the user, but it is well known to be on the order of 5 years for a good quality drive that is not probing some kind of new territory in terms of design/technology.
Just for one example of how/why the drive wears out, consider the spindle bearings. They are prelubricated. That lubrication does not last forever, and there are no "oil here" stickers. Helium leakage is just another, new factor to add to all the other wear factors that drives are subject to.