Why I Trust Helium HDDs Even Less Than SSDs

[Shaina11]

Let me start by saying I already don't trust SSDs because of their limited write cycles, thus as someone who is constantly dealing with large files, e.g. ISOs, BIN/CUEs, reinstalling VMs, it would die rather quickly, hence I always use air-based HDDs that have no such limitation. And I could care less about the speed increase from the SSD, HDDs have always been fine for me.

And now we have helium-filled HDDs. Sealed to prevent helium from getting out eh? 5-year warranty eh? Bull. If it can't last 10 years, pfft.

Now before you say anything, I know you're going to say having a life-expectancy of 10 years for an HDD is ridiculous, I know. However, being someone who builds and repairs vintage PCs with HDDs over 10+ years of age, and still working perfectly, you end up expecting modern hardware to last longer than it currently does. (All these cheap pieces of crap, technology used to be built like tanks, now we're cutting corners, using cheap materials, etc.) Eventually these "modern" machines will be considered vintage. And if the helium evaporates from the HDD before then... well. You're screwed. Meanwhile 30+ year old PCs would still be chugging away without helium.

Until they find a guaranteed way to never ever let the helium escape,(doubtful.) I trust SSDs more. (We need 10+ TB capacities though, I mean come on.) I still trust conventional HDDs more than both of the above.

Also, SMR? Massive risk of data loss/corruption when a power outage strikes. No thank you! And HAMR HDDs? How long will the laser live before burning out? Or can it even burn out? I might trust HAMR more depending on the answer to the question asked.

 

[BadActor]

When Intel/Micron perfect XPoint, it's going to change everything. They are talking about 10,000,000 write cycles compared to NAND's 10,000.

 

[jsmithepa]

One thing I started to notice, todays' HD less reliable than old. Heck I still have a few PATA drives sitting around, 10+ year old and still going, their small size of course now make them obsolete. My 2 years old terabyte Seagate just died on me, 2 years! practically new.

 

[Shaun o]

Are you finished with your very fine observation of the current status of hard drives and SSD drives.

It all depends on how the user of the system treats the drives, and the frequency they are used in day to day usage.
Plus the environment the drives are physically installed in.

And yes I have a few drives here about 8 years of age that still work fine using IDE connections probably like you have also, floating around in boxes or on the shelves.

Helium is used in drives because it can deal with higher operating temperatures of a mechanical drive.
It also by filling it with helium as I am sure you know. Creates a better pressure balance preventing dust also that gets into the drive by chance, forcing it to the outer edges of the drive, where it has less chance on settling on the disk platter damaging the disk platter or the read and write heads of the mechanical drive, so they say.

How reliable it turns out to be is a matter of time, if proof is in the pudding.

It`s surprising what people do to SSD drives and HDD`s that break quickly.
But the main cause is not taking care of them from a user stand point, or setting up conditions to keep them running for longer up time.

And I see your point about HAMR hard drives also.
All laser fitted devices degrade far more quickly in respect to the beam and power out put strength.
And the loss of calibration to the beam width.

The more they are in constant use.

And I do suspect that cheaper components are the cause of drives failing far before there stated life expectancy also.
And the stated or quoted running hours, stated or promoted, on some drives.

Toshiba drives I avoid like the plague, the failure rate of them is shocking.

 

[mrmez]

What a weird rant. Where did that come from? Hahaha.

Minor points...
-What makes you think the helium WILL escape?
Considering what goes into making an HDD, I'm sure an airtight case is the easiest thing to do.
-And IF the helium escapes, what makes you think this will massively shorten the drive's lifespan? Especially if your vintage drives are already lasting 10+ years.

Also... people tend to upgrade their drives more because they run out of space than anything else. And vintage users probably care little about losing their 200Gb drive.

 

[nitrium]

Let me start by saying I already don't trust SSDs because of their limited write cycles, thus as someone who is constantly dealing with large files, e.g. ISOs, BIN/CUEs, reinstalling VMs, it would die rather quickly, hence I always use air-based HDDs that have no such limitation. And I could care less about the speed increase from the SSD, HDDs have always been fine for me.

You should be able to write about 600 TB of data to a 240GB SSD even with the crappy TLC NAND drives before they start to fail. The Samsung Pro 240GB MLC NAND drives will take a full petabyte before they run into trouble. Is that seriously the sort of numbers you encounter over a few years of use (or whatever you mean by "rather quickly"? I mean you could write ~150,000 4GB ISO's to a 240GB TLC drive before you wear out the NAND! To put this into perspective, say you turnover 20 4GB ISOs PER DAY every day of the year - it will still take ~20 YEARS to wear out a 240GB TLC NAND drive and double that for an MLC one!!! Does anyone keep a drive, SSD or HDD, around for that long???
http://techreport.com/review/26058/the-ssd-endurance-experiment-data-retention-after-600tb

 

[Shaina11]

BadActor

When Intel/Micron perfect XPoint, it's going to change everything. They are talking about 10,000,000 write cycles compared to NAND's 10,000.

Now that could be something depend upon the capacity and performance.

jsmithepa

One thing I started to notice, todays' HD less reliable than old. Heck I still have a few PATA drives sitting around, 10+ year old and still going, their small size of course now make them obsolete. My 2 years old terabyte Seagate just died on me, 2 years! practically new.

I really wish they would have better quality control these days, if they'd just use better materials, do much more testing, and stop cutting corners we might actually have a decent lifecycle for modern hardware. What happened to the days of DOS? The Commodore? Atari? Amiga? When things were actually well made.

Shaun o

Are you finished with you very fine observation of the current status of hard drives and SSD drives.

It all depends on how the user of the system treats the drives, and the frequency they are used in day to day usage.
Plus the environment the drives are physically installed in.

And yes I have a few drives here about 8 years of age that still work fine using IDE connections probably like you have also, floating around in boxes or on the shelves.

Helium is used in drives because it can deal with higher operating temperatures of a mechanical drive.
It also by filling it with helium as I am sure you know. Creates a better pressure balance preventing dust also that gets into the drive by chance, forcing it to the outer edges of the drive, where it has less chance on settling on the disk platter damaging the disk platter ot the read and write heads of the mechanical drive, so they say.
How reliable it turns out to be is a matter of time, if proof is in the pudding.

It`s surprising what people do to SSD drives and HDD`s that break quickly.
But the main cause is not taking care of them from a user stand point, or setting up conditions to keep them running for longer up time.

And I see your point about HAMR hard drives also.
All laser fitted devices degrade far more quickly in respect to the beam and power out put strength.
And the loss of calibration to the beam width.

The more they are in constant use.

Aye, I've heard of many failed drives from Seagate especially. Maxtor doesn't seem too hot either. Western Digital and HGST seem to be the most reliable brands when it comes to HDDs, though I question their helium drives. Maybe time will prove me wrong, and maybe helium can be trapped. (Still doubtful.) But time will tell.

mrmez

What a weird rant. Where did that come from? Hahaha.

Minor points...
-What makes you think the helium WILL escape?
Considering what goes into making an HDD, I'm sure an airtight case is the easiest thing to do.
-And IF the helium escapes, what makes you think this will massively shorten the drive's lifespan? Especially if your vintage drives are already lasting 10+ years.

Also... people tend to upgrade their drives more because they run out of space than anything else. And vintage users probably care little about losing their 200Gb drive.

Let's see. Helium is one of the lightest gases in existence and can escape almost anything. It can leak straight through the metal of the drive, sealed or not. It doesn't care about the seal, it goes right through the metal.

If the drives rely on helium to float the head, and use less power to drive the platters, if it disappears, down falls the head, bye bye 10 TB+ of data. What fun that'll be to recover.

Also, vintage drives rely on air, not helium. Therefore no risk of it evaporating.

nitrium

You should be able to write about 600 TB of data to an SSD even with the crappy TLC NAND drives before they start to fail. The Samsung Pro MLC NAND drives will take a full petabyte before they run into trouble. Is that seriously the sort of numbers you encounter over a few years of use (or whatever you mean by "rather quickly"? I mean you could write ~150,000 4GB ISO's to a TLC drive before you wear out the NAND!!!
http://techreport.com/review/26058/the-ssd-endurance-ex...

I will admit, I myself have never used an SSD and am going off of what I've researched. It just so seems that with my particular needs, an SSD would not last near as long as an HDD.

When I'm working with ISOs - Bin/Cues, I'm also editing them, thus backup the old file, write the new file, make alterations if it didn't come out properly, repeat, delete the old file, and store the new one, make any future alternations if needed, and backup to a second drive. VMs along with many, many snapshots could cause many write cycles. (Enough many(s) in there?)

I just trust HDDs more at the moment. When SSDs get better I may eventually try them. Until then, I'm sticking with obsolete tech.

 

[nitrium]

I will admit, I myself have never used an SSD and am going off of what I've researched. It just so seems that with my particular needs, an SSD would not last near as long as an HDD.
I just trust HDDs more at the moment. When SSDs get better I may eventually try them. Until then, I'm sticking with obsolete tech.

Fair enough. But limited write endurance is not a major issue for SSDs. I edited my reply above to illustrate just how badly you can abuse an SSD before you risk wearing out the NAND (i.e. writing 20 4GB ISOs PER DAY every day of the year it will still take you ~20 YEARS to wear out a 240GB TLC NAND drive and > 40 years for an MLC one).

 

[InvalidError]

The seals do not need to keep helium in, they only need to keep everything else out.

If some helium leaks out, the drive will operate under slight negative relative pressure and that will prevent any further helium from leaking out: helium cannot leak out with a partial vacuum holding it in.

 

[Shaina11]

Fair enough. But limited write endurance is not a major issue for SSDs. I edited my reply above to illustrate just how badly you can abuse an SSD before you risk wearing out the NAND (i.e. writing 20 4GB ISOs PER DAY every day of the year it will still take you ~20 YEARS to wear out a 240GB TLC NAND drive and > 40 years for an MLC one).

You do have a very good point there. I also wonder how 3D NAND will change things up in the future.

InvalidError

The seals do not need to keep helium in, they only need to keep everything else out.

If some helium leaks out, the drive will operate under slight negative relative pressure and that will prevent any further helium from leaking out: helium cannot leak out with a partial vacuum holding it in.

Though wouldn't a vacuum be bad for the drive? Causing it to be slightly unbalanced?

 

[InvalidError]

But limited write endurance is not a major issue for SSDs.

When you see a 250GB SSD with only 150TBW of endurance (ex.: 850 EVO), it is difficult to not have doubts.

If I count all the background and scheduled junk Windows does that I cannot disable or schedule on a less frequent basis plus Firefox and Chrome's obsession with updating their bookmarks, cookies and other crap every 10-15 seconds even when minimized, I would probably rake in 5-10TBW/year before I even start doing anything useful.

 

[InvalidError]

Though wouldn't a vacuum be bad for the drive? Causing it to be slightly unbalanced?

Not a vacuum, a partial vacuum.

If the drives operate internally at 40C, its internal pressure, assuming the drive was originally filled to 101.3kPa at 20C, would be around 108kPa. If all of that extra 7kPa leaked to atmosphere, leaving the drive at 101.3kPa while warm, the drive would still contain 95kPa of helium pressure after it cooled down, a 6kPa vacuum.

Even more clever would be to precharge the drives at 95kPa and design them accordingly. Saves 5% on helium.

Drives are not designed to operate at exactly 101.3kPa since atmospheric pressure can vary by more than 10kPa simply due to meteorological phenomenon. A sealed drive is no exception since it needs to cope with thermal expansion of the drive enclosure itself and its atmosphere.

 

[Shaina11]

Not a vacuum, a partial vacuum.

If the drives operate internally at 40C, its internal pressure, assuming the drive was originally filled to 101.3kPa at 20C, would be around 108kPa. If all of that extra 7kPa leaked to atmosphere, leaving the drive at 101.3kPa while warm, the drive would still contain 95kPa of helium pressure after it cooled down, a 6kPa vacuum.

Even more clever would be to precharge the drives at 95kPa and design them accordingly. Saves 5% on helium.

Drives are not designed to operate at exactly 101.3kPa since atmospheric pressure can vary by more than 10kPa simply due to meteorological phenomenon. A sealed drive is no exception since it needs to cope with thermal expansion of the drive enclosure itself and its atmosphere.

But would there be a chance for the helium to leak out when the drive is just starting up or shutting down? As the pressure would change, and if it runs at different temperatures, this could also cause a change in pressure. And if the pressure changes, wouldn't the partial vacuum? It still seems like under certain circumstances, helium could still leak out.

 

[nitrium]

But limited write endurance is not a major issue for SSDs.

When you see a 250GB SSD with only 150TBW of endurance (ex.: 850 EVO), it is difficult to not have doubts.

If I count all the background and scheduled junk Windows does that I cannot disable or schedule on a less frequent basis plus Firefox and Chrome's obsession with updating their bookmarks, cookies and other crap every 10-15 seconds even when minimized, I would probably rake in 5-10TBW/year before I even start doing anything useful.

I think that's surely a bit of an exaggeration. I have one of those drives, and it is extensively used for everything (including a lot of games, e.g. 60GB for GTA V) and I've done just over 5 TB in ~2 years (Samsung Magician keeps perfect track). If 150TB is what I need to be worried about, I'm, uhm, not very worried.

 

[Shaina11]

But limited write endurance is not a major issue for SSDs.

When you see a 250GB SSD with only 150TBW of endurance (ex.: 850 EVO), it is difficult to not have doubts.

If I count all the background and scheduled junk Windows does that I cannot disable or schedule on a less frequent basis plus Firefox and Chrome's obsession with updating their bookmarks, cookies and other crap every 10-15 seconds even when minimized, I would probably rake in 5-10TBW/year before I even start doing anything useful.

I think that's surely a bit of an exaggeration. I have one of those drives, and it is extensively used for everything (including a lot of games, e.g. 60GB for GTA V) and I've done just over 5 TB in ~2 years (Samsung Magician keeps perfect track). If 150TB is what I need to be worried about, I'm, uhm, not very worried.

Could be even less of a problem if an HDD is used for the OS, and the SSD is used for file storage or games.

 

[InvalidError]

I think that's surely a bit of an exaggeration.

Keep tabs on bytes written by Chrome and Firefox. I hardly use Chrome yet it still writes 3-5GB/day if I leave it open on a page with an edit box. That's 1.7TB/year for periodically saving the content of an unchanged edit box. Firefox is not as bad but updating the history, cookies, session state, etc. every few seconds still makes it chew around 3GB/day, that's another 1.2TB/year for leaving it open whenever your PC is on. That's already 3TB/year for doing nothing more than light browsing.

Getting number on Windows' background processes is difficult since most of them stop when activity is detected and many are split across a bunch of processes that do not run for long enough to get numbers on.

 

[nitrium]

That's already 3TB/year for doing nothing more than light browsing.

But if your NAND write endurance is 150-600 TB, that's still going to take anywhere from 50-200 YEARS to wear out a shitty 240GB TLC SSD. Like I said, I use Chrome (as my sole browser) and use my SSD for EVERYTHING except backup, and have managed 5 TB in 2 years.

 

[mrmez]

I reckon we file this whole thing under "worrying about s*** that will never effect you"

 

[Shaina11]

I reckon we file this whole thing under "worrying about s*** that will never effect you"

Until you lose your data right after the warranty period, cause, you know, companies time hardware to fail like that.
(Slight sarcasm... or is it? )

 

[mrmez]

If your data is important, it's already considered lost if it isn't backed up.
Our work computers backup hourly locally, and daily off site.

Anyway, all of this is coming from your perspective.
You don't need the speed.
You don't need the space.

What about people who do?
For plenty of people with 1Tb monthly internet, an 8Tb HDD just isn't enough. Cramming that capacity into the same form factor as a 20Gb drive is a pretty amazing feat, and you don't get there by doing the exact same thing you've always done.

There are also people who laugh at 'slow' 550Mb/s SSD's, and instead opt for 2,500Mb/s.

You don't need any of those, which is absolutely fine, but a lot of people do.
Getting 1x 8Tb drive is always going to be far more reliable than getting 8x 1Tb drives (and far more practical).
As is getting 1 SSD rather than a bunch of HDD's in RAID, which still doesn't speed up write speeds.

Again, file this under worrying about stuff you don't need to.
Seems you don't have any need to buy any of these 'risky' drives, so don't worry about them

 

[Shaina11]

I could use the storage actually, I just don't trust it. And I do keep regular backups, hence another reason why I don't trust it. It's kinda the whole point of my rant.

 

[mrmez]

You'll be one of the few who don't.
It's actually extremely impressive that size has increased by orders of magnitude, but errors have actually reduced.
I've never had an HDD catastrophically fail. It's usually an accumulation of bad sectors etc. They almost always give you a warning.
With a good backup procedure, the chance of losing both before you've had a chance to rebuild should never ever happen.

 

[testforecho78]

My first helium drive an 8TB WD helium white was used only as a backup drive and rarely accessed except during a weekly sync with the drive it backs up. I will never buy another Helium drive. It's not even 8 months old and it self-destructed during a cloning operation. Suddenly the platters sped up and went out of control and sounded like the drive was going to explode. I shut the machine down immediately. On reboot the drive powered up but the platters are dead as a door nail and will never spin again. Sure I am rma'ing it back but I will never trust the replacement.

And this was more than likely due to helium leakage. Goodbye helium. I have 8 8TB drives and this is the only one that has ever given me trouble. Bash Seagate all you want but they have all been spinning for a couple of years now. Only WD and Toshiba drives have crashed on me.

 

[testforecho78]

And I agree with the statement that 8TB drives are too small.

 

[Ty_Bower]

If some helium leaks out, the drive will operate under slight negative relative pressure and that will prevent any further helium from leaking out: helium cannot leak out with a partial vacuum holding it in.

First, my deepest apologies for digging up such an old thread. However, it's important to be perfectly clear on certain things. Perhaps it is even more important, considering the approaching age of these early generation helium filled drives.

When dealing with mixtures of gases (such as air, which is a mixture of mostly nitrogen and oxygen, with trace amounts of argon, CO2, H2, Ne, CH4, Kr, N2O, and yes even helium) the often overlooked physical property is the PARTIAL PRESSURE of the gas under consideration. We must remember the high school chemistry lessons.

I'm not sure we know the exact pressure of the helium filling the drive. We can assume it is one atmosphere (~101.325 kPascal), though I see no evidence to support that assumption. We do know that the trace gases in our atmosphere make up no more than 0.1% of its composition, and the helium portion is only a fraction of that. This means the partial pressure of helium in our atmosphere is less than 0.1 kPascal.

The helium inside the drive does not interact with the nitrogen, oxygen, and whatever else is on the outside of the drive. It only cares about other helium molecules on the outside. And it absolutely will leak out until the helium molecules on the inside are in equilibrium with those on the outside. If we started with 101.325 kPa inside the drive, it WILL leak out until it is less than 0.1 kPa. Assuming our initial assumption was valid, that's one thousand times less than that with which we started. Certainly the heads would crash at this point.

My helium drives (HGST) are 2.6 years old now. So far, they still run well - but I am anxious.