Thread 106472300

Alright /g/ents, I'm looking for the most long-lasting hard drives OVER 1TB.
I want to store a bunch of stuff and I want to know which hard drives will last for over a decade. They can be both internal or external. I'm in the EU so American pricing is kind of irrelevant to me.

>>106472300 (OP)
Just use tape.

my cock

>>106472300 (OP)
>WD Gold
definitely not Western Kiketal

>>106472300 (OP)
If you are buying the major brands there is no real difference, failures tend to be pretty random and you can get a bum drive from any manufacturer. Also reminder that if you don't power them on regularly bitrot will claim your shit in a few years or even months on whatever drive you use.

>>106472300 (OP)
These will

>>106472300 (OP)
check external hdds on amazon, sometimes they are on sale way cheaper than internal one... buy 5 of them, take them out of the case and put them in RAID5 (or RAID6 if you are extra schizo) NAS

>>106472492
>Also reminder that if you don't power them on regularly bitrot will claim your shit
How would power in them on help fight bitrot? It's not like spinning the platters does anything special.

any HDD over 8TB has helium in it, helium leaks out through tiny pores in the casing. you at most have 10 years with those drives, so youll have to buy a new one and xfer your data every 10 years. You should be doing that anyways but i do have a drive from 2012 that is still functioning, now if it had helium in it i doubt itd work because all those platters spinning without helium would overheat and itd die.
>>106472829
think he meant on SSDs, but if you just install something on an SSD and just leave it, not an OS drive, its essentially just a flash drive and will be fine without needing to error correct. you only need to power it on if you have been wearing the ssd out by writing and deleting a bunch on it so the cells are worn out

>>106472300 (OP)
If you're worried about the reliability of individual drives you're doing it wrong. Have at least 2 copies on different drives and validate them periodically. Either by filesystem scrub or sfv files. IME A-grade drives are less likely to have media degradation issues, but it's no substitute for a spare copy because you're guaranteed to have some instance of data loss eventually.

>>106472300 (OP)
wd gold are pretty good, got 2 myself. Also hgst.
But not 10y.. they're like good for 5y and you usually get more out of them

>>106472300 (OP)
>What are the most durable hard drives?
hard drives are rated for power on hours
dont boot it often and dont leave it powered on for long
that is the most durable hard drive

>>106472300 (OP)
Just buy anything and use RAID. Either it fails in the first year or two or lasts for a decade.

Just check the backblaze blog

https://www.backblaze.com/blog/backblaze-drive-stats-for-q1-2025/

>>106472300 (OP)
>>106473898
this

RAID 10

then just hotswap the failed HDD for a fresh one. I've had two fail in 7 years.

>>106472968
>you at most have 10 years with those drives
Lies
At my work we track literally millions of drives. Even the earliest HGST Helium drives don't show any signs of leaking.
We would know as it was tracked that some technician decided to open HDD ESD bags with a knife and caused a few thousand drives to leak from literal gouges in the case. They were preemptively replaced before they even failed.
Don't abuse your drives and you won't have an issue.

I get "new thing bad" but just because your miserable doesn't mean your BS is acceptable.

>>106474257
>heliumposting
ayylmao

There is no durable hard drive. They will all fail. It's why you buy two.

>>106474998
but I'm poor :(

>>106475047
Then buy smaller ones and just store less data.
If you're too poor for that then you should just start deleting stuff.

>>106472300 (OP)
I've never had a WD fail on me.

>>106475057
>just store less data
lol no

>>106473832
Oversimplification. They're rated by head park cycles, motor start / stops, and media is also graded. I've only killed a HDD with power-on hours once and it took 13 years.

There's no reason to pay for enterprise drives with crazy overbuilt mechanical if you're never going to load it 24/7, but 7200 RPM / uncut platters will most def see a reduction in data loss.

>>106472300 (OP)
Use BD disks

>>106472300 (OP)
you should be more concerned with storage.haha.

They Need to be in a dry room with steady temperatures inn a bin with silica packets. stored at least 2 ft above the floor. this applies for game discs and ammo as well.

>>106474257
>I get "new thing bad" but just because your miserable doesn't mean your BS is acceptable.
umm actually this is basic chemistry, this is why trying to hide coke from sniff dogs doesnt work because all structures have space between their atoms and eventually everything leaks out. Helium drives at that scale have not even been on the market for 10 years so how can you tell me that your work tracks and shows no leaks when the drives arent even old enough to prove if they leak or not lol

>>106478016
> leaks out
Not really.
First, it’s not under any pressure. In fact, what they’re trying to do is approach a vacuum.
Second, it’s not in a balloon, it’s in a hermetically sealed steel encasement.

>>106472968
So... what about keeping your HDDs in a sealed plexiglass box with helium inside? Or is that too autistic even for /g/?

>>106478119
Little by little leaks out over time through the space between atoms regardless of what the encasement is. How about you go to school before trying to argue over something you know nothing about while i have 2 chemistry degrees. Eventually the helium will leak out to a critical point where there just isnt enough to support the "smoothing" effects of it against all the discs in big drives. At that point the drive will be usable but will overheat so if you arent proactive in making backups before this point occurs you will most likely destroy your drive in the process of data transfer.

>>106475137
Partitioning when logically possible also helps with HDD drive longevity too, so the head won't have to search the entire disk to find one thing and it will spin less.

>>106478249
its literally easier to just make copies so if a time ever comes where the helium inside is no longer enough to support the cooling effects on the discs you dont have to worry about losing your drive before it shatters from the heat.

>>106478437
They do have a helium sensor. It's one of those failures you get plenty of warning on.

>>106478483
I doubt a normie will know how to even see what the sensor has to say. All im saying is its new tech and we just dont know the caveats of the tech yet because the time hasnt passed to let us know what they are.

Who knows maybe other components in the drive will fail sooner than any significant He leaks out.

But if like many people, you keep a drive for 20-25 years then it might be a significant caveat to deal with.
I know several collectors of old PCs that have a 1999 mac and it still boots up. Now the HDD in that doesnt have Helium, but what if it did? Would it instantly fail from the overheating? Would the helium leaking be so miniscule even in the 20-25 year time frame that it wouldnt matter?
We just dont know. As an archivist and an entropy fighter I always try to prevent the deterioration of data whenever I can.
Thats why my point was no matter if the He leaking is significant long term relative to other components failing is. YOU SHOULD BE BACKING UP YOUR DATA ALWAYS.

Nothing sadder than watching lost media youtube channels.

>>106478437
Making a sealed box and filling it with helium is trivial.
If you want to pay the price tag on rebuying all your harddrives every 10 years thats fine, some of us would prefer to rebuy them all every 20, if possible, for a minor amount of effort. Most of us have enough drives that the cost is in the thousands.
Are you one of those consumerist shills that are paid to just get people to never fix anything, never preserve anything just buy buy buy?

>>106478552
>Are you one of those consumerist shills that are paid to just get people to never fix anything, never preserve anything just buy buy buy?
quite the opposite. I too wish to be able to use these drives in 20 years. Im saying these drives havent been around 20 years for us to know the rate of He gas loss in normal use. Lets say all other components dont fail and you do have like my buddy who has a 26 year old HDD that still works in his collection, then if the gas leaks and you havent been making copies/buying extra back up drives that also are obv helium to fit the same large about of data, you might/potenitally be hit in the face with a "oh oh your drive is failed because the space between discs is too tight and the helium all leaked out, now you cant use this drive to even READ let alone transfer the data"

its literally because i care about keeping things and not just buy buy buy that im bringing this up so people in the archiving hobby can think about/look out for this as a potential fail point in these modern high capacity drives

>>106472968
>any HDD over 8TB has helium in it
wrong, picrel is a 8tb drive using air. they stopped making 8-10tb helium drives and they only make air ones, and they are the worst possible hdds to get. you might still find used 8-10tb helium drives though, but they'd be 5+ years by now.

helium drives starts from 12tb and above.

>helium leaks out through tiny pores in the casing.
no, they are sealed airtight, it's why they have a different physical design. you need to open them up like a sardine can for the helium to leak out.

I have 2 helium drives both clocking at 2.5 years and they both show 100% helium levels in the SMART table.

>>106478422
>Little by little leaks out over time through the space between atoms

That would take thousands if not millions of years. The read heads would've failed by then.

>>106478552
>Most of us have enough drives that the cost is in the thousands.
Your 20 year old drive is e-waste, full stop.
At some point they aren't worth the space in keeping when literal stacks of HDDs can be replaced with a pair of mirrored drives

>>106478580
>so people in the archiving hobby
SInce when does archiving mean keeping old HDDs?

>>106478584
>no, they are sealed airtight, it's why they have a different physical design. you need to open them up like a sardine can for the helium to leak out.
there is no such thing as an unleakable casing in our universe bro. im talking mostly about 12tb + drives. if people start hoarding 12tb+ drives over the next few years who knows if this will eventually be a problem

>>106478599
>SInce when does archiving mean keeping old HDDs?
youre archiving old hardware like my buddy

>>106478597
Yes — helium eventually leaks out.

Helium atoms are so tiny and non-reactive that they can diffuse through seals, welds, and even very small imperfections in metals or glass over time. That’s why helium-filled HDDs aren’t expected to hold it forever.

Reality in drives: Manufacturers use hermetic sealing and welded enclosures, but even then helium slowly diffuses. Modern helium HDDs are designed with a service life of about 5 years before helium loss could start affecting performance.

Diffusion mechanism: It’s not that helium slips “between atoms” in the literal lattice, but that it passes through defects, grain boundaries, or permeable materials in seals.

End result: The drive will lose helium gradually until it’s replaced by normal air, raising drag on the platters and increasing heat/power consumption.

So yes, diffusion means helium leaks out eventually, but engineering makes the leak slow enough that it outlasts the intended lifespan of the drive.


>Modern helium HDDs are designed with a service life of about 5 years before helium loss could start affecting performance.

>>106478525
>All im saying is its new tech and we just dont know the caveats of the tech yet because the time hasnt passed to let us know what they are.

It's not new tech. It has been around a decade.

>>106478552
>some of us would prefer to rebuy them all every 20, if possible,
That won't be possible, HDDs don't generally last that long when in use.

And even just after 10 years, a new drive would end up quite a bit faster. My 2TB drives from 2012 peak at ~140MB/s (4 platters), my 4TB 2016 ones peak at ~180MB/s (3 platters), the helium drives I bought around 2020 or 2021 reach up to 240MB/s (8 platters) and the 20TB ones I got last year go as high as 290MB/s (10 platters).
And we are just on the cusp of HAMR/MAMR increasing platter density, same as how PMR did in the late 2000s to mid 2010s, when we went from 500MB platters to 2TB ones.

A 20 year old hard drive would be so fucking slow that even assuming it works properly, you'd want to get it replaced with a new one.

>>106478627
>m gradually until it’s replaced by normal air,
This is where you are wrong
Air cannot enter the hermetic seal, this is why ultimately why helium drives are going to be fine.
As diffusion happens a pressure differential is created and it means the concentration differential decreases which slows the rate of diffusion. After a few years the diffusion rate slows to a crawl

>>106472307
fpbp
LTO tape is like 2$/TB

>>106478525
>I doubt a normie will know how to even see what the sensor has to say
They don't have to. It's a pre-fail SMART indicator. They get a message from their BIOS and OS.

>>106478674
>A 20 year old hard drive would be so fucking slow that even assuming it works properly, you'd want to get it replaced with a new one.
yep thats what im saying. we dont know how long these he drives will last. now also i think HAMR/MAMR will also use helium rright away. so itll face having that extra caveat on top of whatever other caveats HAMR tech brings to the table

>>106478696
>we dont know how long these he drives will last
You don't really know how long anything will last
To say one thing will last more than another because you have a surviving example is post-hoc rationalization, not one based on careful analyst on what made the drive last.

Data preservation is ultimately an active endeavor, being faith based is asking for trouble
https://arstechnica.com/gadgets/2024/09/music-industrys-1990s-hard-drives-like-all-hdds-are-dying/

>>106478745
>Data preservation is ultimately an active endeavor, being faith based is asking for trouble
>You don't really know how long anything will last
why i even said what i initially said. Because we dont know everyone should be making backups. like you said active endevour

>>106478613
>there is no such thing as an unleakable casing in our universe bro.
If you start taking universal entropy into account then it's best not to use any computer at all because they will all eventually break.

>>106478627
OK, show me a some research on how 5-10 year old helium drives retain their helium levels. They've been around since 2016 so there must be tons and tons of those around with their helium levels failing below the threshold.

Except you won't find any that specifically died because they lost their helium filling, because barring some construction error (that would kill air filled drives too), this would only happen on a timeline so long that the read/write heads would fail before the helium leak affects performance.

>>106478696
>we dont know how long these he drives will last
They've been around since 2016 and still lasting.

>now also i think HAMR/MAMR will also use helium rright away.
I think you are stuck in 2016 in your head because HAMR/MAMR drives are already on the CONSUMER market, let alone enterprise. I have four.

Obviously they all use helium, because the big money is not in making 6TB external drives cost less, it's in shipping 30+TB enterprise drives to datacenters. They'll only start making air versions once they shifted all production lines to HAMR/MAMR, since it's not worth making the smaller drives for now.

>>106478760
>show me a some research on how 5-10 year old helium drives retain their helium levels.
wish i had such research we could end this lol

>>106478760
>I have four.
doubt. 40TB HAMR isnt even due till 2026

shitgate sucks, i had 3 fail on me

>>106478784
the data shows that its better than WD but ive never had WD fail meanwhile ive had shitgate fail several times in the past

so duno how to gage that

>>106478790
>data
sauce

>>106478796
https://www.youtube.com/watch?v=IgJ6YolLxYE&t=775s

>>106478766
I can at least tell you that Google seems to have no issue with the earliest 8TB He drives.
They are fucking ANAL about using hardware till the wheels fall off and they would know about mass failures.

>>106478804
>toshiba and WD keep on winning

>>106478804
>Ultrastar is #1

>>106478790
I believe WD started to fuck their (consumer) shit right around the time Seagate started getting their shit together
probably most of the higher quality WD stuff is because they bought out Hitachi (HGST)

>>106472819
>>106473898
>>106474136
RAID is obsolete. use truenas (core), or linux (if you're brave since zfs was made for *bsds) with ZFS. (raidz2 means 2 drives can fail before the entire thing fails for example)

ZFS is literally enterprise grade and handles shit for you, RAID is a toy

If WD hired competent engineers to improve their HDDs instead of pajeet shills, their business wouldn't begin failing so hard like it does now.

>>106478766
So you admit you got fuck all except your cocked up theory on how the drives will all kill themselves because helium atom absorption is a thing that exists.

>>106478924
ZFS RAID is literally just software raid.

>>106478982
I think he means hardware RAID. Hardware RAID is indeed completely obsolete simply because it's an abstraction at the wrong level. Presenting a bunch of drives as "one big drive" to the filesystem on the hardware level will never work and is a retarded idea to begin with.

>>106478760
> show me a some research on how
Translation: “I think absence of evidence means ‘evidence for absence’”

>>106478924
> zfs made for bsds
Zfs is made for Solaris
I heard that bsds are using the openzfs for linux with a shim layer.

>>106472492
They do not need to be powered on regularly and bit rot has literally never been proven to be a thing.

>>106472722
Yes, both are completely unrelated. He's just a dumb fuck

>>106478433
It spins the same amount regardless with the fuck are you talking about, also, journaled file system.

>>106478924
It is still a raid, stupid ass