In a curious bit of scheduling, HGST decided to roll out a bunch of new enterprise storage products this morning, during Intel's IDF keynote and Apple's iPhone 6 reveal. But the company had an ace up its sleeve: the world's first 10TB hard drive. The unnamed drive is designed for "cold storage" applications, according to HGST President Mike Cordano, and it's sampling to customers now.
HGST combined a couple of cutting-edge technologies to achieve this new capacity milestone in a standard 3.5" form factor. The most intriguing is shingled magnetic recording, otherwise known as SMR, which lays down tracks in an overlapping fashion. This approach increases the storage density of the media, albeit with a performance penalty attached to modifying data. That handicap may not be too much of a concern for enterprise applications that use the drive in more of an archival capacity.
Unlike typical air-breathers, HGST's 10TB monster is hermetically sealed and filled with helium. The lower-density gas cuts down on internal resistance and turbulence, allowing the platters to be stacked closer together. HGST has offered helium-filled HDDs since last year, but this is the first time SMR has joined the party.
Cordano didn't detail the 10TB drive's spindle speed, platter density, or other specifics. However, he did express a great deal of confidence in the firm's "HelioSeal" tech. Moving forward, all future HGST enterprise drives will be filled with helium. The sealed environment is suitable for heat-assisted magnetic recording, too, though Cordano doesn't expect HAMR-based drives to be ready until 2017 or later.
HGST also revealed a couple of new enterprise drives based on traditional recording technology. The largest of the two is the Ultrastar He8 8TB, which uses the second-gen incarnation of HelioSeal. This puppy has seven platters, each with an areal density of 664Gb/in². The spindle speed clocks in at 7,200 RPM, and the available interfaces include SATA 6Gbps and SAS 12Gbps.
The Ultrastar 7K6000 6TB will be HGST's last "in-air" enterprise product—no helium mojo here. It has only five platters, but the areal density is slightly higher, at 703Gb/in². Versions of the 7K6000 with 2TB-5TB of storage will be offered with lower-density platters. The 7,200-RPM spindle speed is the same regardless of the capacity, though. So is the choice of SATA and SAS interfaces.
Don’t get me wrong, I love how the hard drive manufacturers are finally stepping their games up and giving us larger capacity drives. But damn is this going to be expensive to keep up with. I’ve said it before that I always buy two of a large capacity drive cause you never know. Even if it’s not in the budget at the time, I’ll try to make that happen within 2 months of the first large capacity purchase.
Recently I had a WD Green 2TB die on me after a month. A MONTH. And of course it was right after I had moved all my data to it. Luckily, it was just a copy of my 1st WD Green 2T which is still running strong today.
10TB…wow, but I don’t think that will be in my budget for while…if EVER. 🙁
I would surmise that flutter is more of a problem for reading/writing without error than it is for motor current draw.
Either way, the power draw is significantly lower on He-filled drives and that makes me a happy camper. 🙂
I also remember 3dfx’s “100 billion calculations per second” commercial
[url<]http://www.youtube.com/watch?v=DmaYH1F6kho[/url<]
Actually a Hard Drive wouldn’t work in vacuum. The heads need the gas inside the case to float over the platter.
The spinning platter pull air(or in this case helium) along with them. this gas causes a cushion of air to form between the platter and the head and the heads float on this cushion.
I have an answer but I’m too busy laughing 🙂
SSDs are already effectively shingled. Common flash isn’t written sector-wise.
Don’t think it was 64MB of RAM?
Anyone here know when they’re gonna make hard drives filled with laughing gas?
The 8 TB Seagate also uses shingled recording, it’s just not filled with helium, so it has fewer platters. 😉
EDIT: I can’t find if the use of SMR in the Seagate has been officially confirmed, but it seems more than likely given the sudden jump in size/density.
Where are shingled SSDs now that I need them? Cheap, large, slow, cheap, 10 times rewritable, and cheap?
Transfer of heat should be very good due to forced convection, even if the conduction was poor.
When I was a kid I told my grandma there can be tens of thousand of transistors on a single chip. No wonder she’d never believe that.
My Amiga 1200 has a 40 meg hard drive in it (and 64 meg of ram) from ’96
What’s disturbing is the OS only takes about a 7 meg of that
That would seem to be suggested by [url=http://www.extremetech.com/computing/136257-the-squeak-of-innovation-western-digital-launching-helium-filled-hard-drives<] this. [/url<] On the other hand, I'm not sure that the chart tells enough of the story. Depending on how the helium was injected, that alone could explain the difference. [url=http://pureinfotech.com/wp-content/uploads/2013/11/hgst_6tb_ultrastar_he6_helium_filled_hdd.jpg<]This image [/url<] could provide better information, since a fluttering disc will bring problems. That said, flutter can be stopped in other ways. I'd really have to do some number crunching to see how much flutter we're talking and if it's enough to explain what is being seen. On the other hand, the answer could simply lie in the combination of moving to 7 thinner, more closely spaced platters and the helium. More closely spaced platters are more likely to keep the flow fully plugged (no turbulence), as would the hermetically sealed nature of the enclosure. The increased head convection of the helium may allow thinner platters, since cooling may become less of a concern (I don't know the material science stuff on the drives). The drives are reported to be much lighter than equivalent air-filled drives. Most of that isn't coming from the evacuation of the air, so I wouldn't be surprised if a fair amount of the savings comes from the reduction in inertial loading on the motor.
Based on the power numbers unless there’s some other voodoo they’ve kept entirely hidden…there’s less drag.
That I’ll grant, and it seems much more credible. It also helps explain why everyone always feels it necessary to mention hydrogen, which is 14 times the value of air (which makes the explosions that much more likely).
That would also go a long way towards explaining why we don’t push more for vacuum.
I would love to see a “Get Shingles” video follow up to the hit single “Get Perpendicular”… Or maybe a “Get HAMRed” video.
I WISH I COULD GO BACK IN TIME AND TELL HIM. HE WOULD HAVE BEEN SO HAPPY TO KNOW THAT WORKS
I know HGST claims it’s all about turbulence reduction, but I wonder if it’s actually about heat conduction. Gaseous helium conducts heat about 6 times as well as air does. If heat at the recording layer is a limiting factor, it would explain why switching to helium allows them to pack more platters in.
I find the minimal improvement in HDD capacities in the last few years worrying. More platters doesn’t sound like a promising idea, nor does shingled writing. If HAMR turns out not reliable, I wonder if there are any other avenues for progress.
So if I want to reroof a house, I just toss some of these up there? Now… What to do with petabytes of storage…
My 2001 desktop rig has a 32GB hard drive.
We now have SD cards that reach up to 512GB.
the key is, that you have to keep scratching at it.
One these would be perfect for a total system + NAS backup in a single drive. Stick it someplace off-premises or in a security deposit box, in case mother nature decides to ever make a housecall.
He’s probably referring to the tired old “Get Perpendicular” video.
I guess this drive could have a “Get Shingles” video, just to keep the embarrassing roots.
As I commented in a [url=https://techreport.com/forums/viewtopic.php?f=1&t=94411&start=30#p1217389<]forum post [/url<], I remain highly skeptical of claims of lower drag. The turbulence will be dependent on Reynolds number (Re = V*x/ν). Helium's kinematic viscosity is roughly 10 times greater than air's. For a 0.07mm gap (probably too big here) on a 3.5", 15,000 rpm drive, that means the comparable Reynolds numbers are: 35.5 (He) and 287.2 (air). These values are extremely small for Reynolds number, which means that we can reasonably assume that the flows are fully developed laminar flows. Which means my previous post holds. Taking the heads into account, you'll trip the flow and create a wake, but I'm still pretty skeptical that it's significant.
Isn’t the 8TB Seagate drive a normal air breather? So, to get to 10T, HGST needs He2 and shingled recording?
There’s got to be some other performance/behavior of these drives to make them useful.
I want to see a Backblaze study on these helium drives. I can’t imagine they are reliable once the helium leaks out in a couple years.
I had a friend with shingles. He said it was horrible.
[url<]https://www.youtube.com/watch?v=QzntZLHcYy0[/url<]
When I was a kid I gawked when a friend told me their computer can store 30 million bytes.
Needs a music video with a catchy jingle.