6TB Helium drives

With the global supply of helium dwindling to the point of the element becoming extinct, I reeeeeally don't want to see the price tag of these.
 
With the global supply of helium dwindling to the point of the element becoming extinct, I reeeeeally don't want to see the price tag of these.
That's never happening. It's constantly being produced through natural radioactive decay (uranium 238 for example). Alpha particles are essentially just a helium nucleus.
 
That's gonna be a long RAID sync when one of those suckers dies...
 
That's never happening. It's constantly being produced through natural radioactive decay (uranium 238 for example). Alpha particles are essentially just a helium nucleus.

well, technically yes, but from an industrial standpoint helium is going to get very difficult to come by
 
That's gonna be a long RAID sync when one of those suckers dies...

I expect a full day. Although I would not use 5XXX RPM drives in my work raid servers. That is unless the raid was just a backup.
 
That's never happening. It's constantly being produced through natural radioactive decay (uranium 238 for example). Alpha particles are essentially just a helium nucleus.

Sure, but at the rate of consumption vs production we're going to see a massive scarcity of the element within a few years time with accompanying massive price tag.
 
everyone start farting in a sealed bucket and harness the power to fuel your everyday lives
 
I expect a full day. Although I would not use 5XXX RPM drives in my work raid servers. That is unless the raid was just a backup.

Unless the drive is "wounded" and goes to 35 MB/sec as some of my 3 TB seagates did on recovery :) That would be 45 hours.

Did anyone find out what the rpm actually is for these?
 
Unless the drive is "wounded" and goes to 35 MB/sec as some of my 3 TB seagates did on recovery

Drive doing a background recovery of UREs? I have seen that. Weekly raid checks have reduced most of this type of bad behavior on my small sample of around 200 drives.
 
Drive doing a background recovery of UREs? I have seen that. Weekly raid checks have reduced most of this type of bad behavior on my small sample of around 200 drives.

That sounds about right. What exactly does the drive attempt to do about the read error?
 
That sounds about right. What exactly does the drive attempt to do about the read error?

guessing something akin to spinrite or ddrecover; come at the problem block from different directions, at different speeds etc. hoping to get one last good read that passes parity
 
The article says it runs cooler than a 7k4000 (7200 RPM 4TB drive)

6Gbit/s SAS or SATA interface
50 per cent more capacity than the Ultrastat 7K4000
A 23 per cent reduction in power use
At 640g it is lighter than the 7K4000
It runs 4-5⁰C cooler
 
Just a bit of info on these drives that is public, they are still under nda for specifics.. They are actually 6-7 degrees cooler idle and 3-4 load than the 4TB 7200rpm US drives. I can't verify the actual power decrease % but it makes sense based on the temps. The drives are noticeably but not not significantly lighter than a standard 4TB US (No Helium jokes please). The external speeds are at least on par with their 4TB brethren for random and better for sequential. Another interesting point is they can be completely submerged in certain liquids for liquid cooling which you will hear more about coming soon.
 
Just a bit of info on these drives that is public, they are still under nda for specifics.. They are actually 6-7 degrees cooler idle and 3-4 load than the 4TB 7200rpm US drives. I can't verify the actual power decrease % but it makes sense based on the temps. The drives are noticeably but not not significantly lighter than a standard 4TB US (No Helium jokes please). The external speeds are at least on par with their 4TB brethren for random and better for sequential. Another interesting point is they can be completely submerged in certain liquids for liquid cooling which you will hear more about coming soon.

Any idea why HGST is still not using 1TB platters?
 
Any idea why HGST is still not using 1TB platters?

We asked and got 2 different answers at different times. From the technical people we got first an answer about the platter substrate material and areal density being an issue, and later we just heard it was an "internal decision" not related to technology. I personally wouldn't expect them to have used more heads and platters than they really needed for cost reasons, but anything is possible. I do know that on their current timeline SMR drives are poised to drop 2Q14 and HAMR 2Q15 and that WILL bring platter counts down &/or arial densities up compared to current availabilities.
 
These 6TB drives will only be viable at <$300. Personally, I wouldn't spend more than $250 for one.
 
These 6TB drives will only be viable at <$300. Personally, I wouldn't spend more than $250 for one.

For end users maybe. Prosumers & [H] users would pay a bit more to keep the noise/case size/port count down. For enterprise users these are viable at triple the price. If I can decrease (rack needs, square footage, controller/connection counts, license costs, cooling costs and more) for my bulk storage by 50% (Over 4TB, 100% if you are at 3TB) it is a deal!
 
Last edited:
I can see the use for these things even if they are expensive.

Imagine I can have a 3-drive raid1 and still have 6 TB net. That is a huge simplification of the setup and a huge improvement in overall safety. The resync risk is under control by sharing the resync read load evenly between the two non-failed drives and you can partition the drives into different filesystems, syncing important ones first.

Compared to the fragile raid6 that takes more disks and more power and has individual disks that are useless outside the array that's a big improvement.
 
Helium tends to leak through seals and HGST has had to develop hermetic seal technology to stop this from happening. This means the He6 could be used in a liquid cooling scheme as the liquid cannot get into the drive and damage it. Current air-cooled drives are unsuitable as the cooling liquid could penetrate the drive.


Anyone else catch that the article author has no concept of basic physics or phase change cooling?
 
The capacity per port is wonderful, but I don't know how I feel about seven platters or it being a first-gen product. I've been pretty happy with the power usage and thermals of my WD reds.
 
My biggest worry would be the hermetic seal. I mean if that is somehow compromised over time. Also would it have at least the same ability to tolerate vibration since the heads will be flying closer to the platters (or would they not)?
 
My biggest worry would be the hermetic seal. I mean if that is somehow compromised over time. Also would it have at least the same ability to tolerate vibration since the heads will be flying closer to the platters (or would they not)?

I doubt the flying height will be significantly different. It is not randomly determined -- the designers control it by adjusting the spring tension in the heads to obtain the desired flying height.

As for the hermetic seals -- there are decades of data on hermetic seals and helium leak rates. The traditional method for measuring leak rates of hermetically sealed devices is to introduce helium into the device (either before sealing, or after sealing by placing the device in a pressurized helium chamber) and then put the device in a vacuum chamber with a mass spectrometer to detect helium.

I expect the HGST designers used existing hermetic seal technology and helium leak data, perhaps with a few tweaks, to develop the technology for the hermetic seals on the HDDs. I expect that they did plenty of vibration, G-shock, and thermal cycling tests to qualify the technology.
 
I can see a very strong value in these drives for data storage facilities. To them...any thing reduces port count, foot print, and power has a very tangible and calculable benefit. Can't wait to see an update from the Backblaze blog.
 
What about:

-Price
-RPM
-Noise
-Transfer rates
-Heat

Any info on that?

-Enterprise only for now so it'll be steep
-Same
-Less
-Same
-Less

According to what I've read:
Helium causes less turbulence for lower noise. They are also able to make everything thinner and lighter thanks to the lower turbulence thus decreasing power.

The cost of the helium vs less materials needed should be a net positive for them. So it's a win.
 
Should wait and see...

Care to back up your assertion or are you just a natural doubter? Sorry, common sense indicates that they wouldn't make this drive if it didn't have a tangible benefit to their largest consumption audience.
 
Back
Top