I have quantum fireball from 2000, so 26 years old, still going strong.
5 years doesn't seem that long for a drive that cost hundreds of dollars! Persistence is the point.
Just wondering why Seagate seems like the bottom of the barrel in the longevity department. Western Digital drives seem to fail a lot less frequently on average in this dataset and in my life experience.
To Seagate's credit, I do have 8x24TB drives that have been working fine for the past 4 years. Hopefully can last a few more until the compute hardware shortages pass.
That's a brand I haven't heard of in a long time. I had a 8gb HDD from the brand in 2000 until my sister kicked the computer case out of frustration which ended up shorting some chips on the drive. I mourned the loss of my music collection for quite a long time.
I have two particularly notorious Seagate periods:
Seagate bought Conner when Conner had released several models w/
leaky seals. Bad sectors started at the outer edge of the
platters and grew inward. We had a lot of these drives
out there and Seagate refused to honor Conner's drive
warranties.
The 7200.10 series had super high failure rates. I wound up
replacing every one in my care, within 2 years. The 7200.11
drives weren't much better.
I think the last Seagate lines I truly trusted were the ST series of MFM and RLL drives.
I do not know, but the last time when I have bought a Seagate HDD, I had a very nasty and unpleasant surprise.
Last year I have bought a 22 TB Seagate Expansion Desktop external HDD, because it was cheaper than the other 22 or 24 TB HDDs available at that time.
I had read carefully its datasheet before buying and there was nothing suspicious there, so I assumed that it must be cheaper just because it is a slow HDD. I did not care about the speed, it was for storing data archives infrequently accessed.
Only after receiving it I discovered what was not said in the datasheet, that this Seagate HDD does not support S.M.A.R.T., so there is no way to test it to see if it works OK and there is no way to discover when errors have happened, e.g. to see when the HDD becomes too old, so you need to migrate your data.
I have never imagined that in 2025 it is possible to buy a HDD that does not support S.M.A.R.T., especially in HDDs with a capacity over 20 TB, and moreover without giving a prominent notice about such a misfeature in the datasheet.
Before this, in 2024 I had bought a 24 TB Seagate SkyHawk, which had S.M.A.R.T., as expected. Since then, after the Seagate Expansion fiasco, I have bought a 22 TB external WD HDD, at the same price with the Seagate, and which has S.M.A.R.T., as it is normal.
I cannot see how removing S.M.A.R.T. support can reduce costs, as it is just a firmware feature. I any case a manufacturer that removes testing and error reporting features from its products clearly does not give a s*t about data corruption and HDD failure rates.
With an external drive the SMART info might be hidden behind the USB-to-SATA bridge, smartctl has support for some of those but sometimes needs to be told with an extra argument.
The Seagate Expansion is made as an external drive by Seagate itself, so it was not put in some random enclosure.
If Seagate has chosen in 2025 to use some archaic bridge that does not pass the SMART commands, it is on them. That would be even more stupid than not implementing SMART in the HDD firmware.
As I have said, the previous external Seagate that I had bought in 2024 had SMART that worked fine over USB. I have a large number of external HDDs, most from WD. Some have been packaged by the HDD vendor as USB drives, others I have assembled myself into enclosures with SATA-to-USB bridges.
On all of them SMART works perfectly, except in this Seagate Expansion Desktop, where the drive replies that SMART is not supported.
Whenever I buy a HDD, I first run the long SMART self-test, to determine whether it can be used safely or I should return it immediately, even if the long self-test takes a couple of days on modern over 20 TB HDDs.
I started to use this procedure after I had some problems with a batch of WD drives, 2 decades ago, where all the drives had very frequent errors since the first few days of use. After running the SMART self-tests, which all failed, the seller could not deny an immediate replacement.
You know, there are external drive USB controllers that linux blocks/blacklists SMART passthrough when using UAS due to paranoia and historical problems, but this can overridden.
I guess it is possible this is not your problem, but the last Seagate external I bought in October worked just fine with this workaround. This is probably safe from a data integrity standpoint, at least with a modern filesystem, but in my case it was no issue as I was only using the SMART to do tests before shucking the drive. Also, I don't know of any modern drive that truly doesn't support SMART.
Doesnt surprise me, Seagate is marching to its own drum. My experience defiantly mirrors others' higher than average failure rate as well.
My latest 'fun' experience with them, also, came in the form of an Ironwolf drive which is 'detected' on usb-to-sata interface when plugged in, around %15 of the time. While it starts up consistently on a plain SATA interface. This makes it unusable for what I need. Again, no other drive or MFG ever fails on this usbSata, just the new ironwolf, which it appears is actually for the chineese market, but was sold on newegg, but this is not necessarily seagate's fault, nevertheless.
Even without using ZFS (I prefer XFS as significantly faster) all the files that I store have content hash values in extended attributes, for data integrity verification (and also for data deduplication).
Whenever I write that drive, after a power cycle (to be sure that the files are read from disks and not from some cache) I run a script that checks the integrity of the files, to be sure that I can remove them from elsewhere without risking data loss.
With SMART-enabled drives, I usually do that only in the rare cases when a drive reports corrected errors, because I have seen cases when a drive miscorrected some errors, resulting in corrupted files. With a HDD without SMART, when the drives finds errors, but it believes to have corrected them successfully, there is no external sign that something could have gone wrong.
It's regrettable that you would spread misinformation, and easily explained/debunked information at that. This is a reasonably well-known issue and even with the extremely bad and limited information you provided (no model numbers or other info), I already know exactly what your problem is.
nope. i don't even think ibm were being widely sold at all. ultrastar rings a bell, somewhat, but definitely not ibm hard drives. and i'm talking early 2000s here.
Pretty sure that's cos every AI company wants to have a copy of all the worlds information stored in each of their data centers, and hard drives are the best storage medium for that.
Design your training strategy carefully and you can do streaming rather than random reads from the drives and get enough performance.
To be honest, they mostly have 8 TB and up in stock at the moment, but the price was around 35 EUR for the 1 TB drive which is kind of insane when similar drives at local stores cost 110 EUR right now - I remember when those Seagate Barracuda drives were about 40-50 EUR brand new!
I still use SSDs for boot drives, video games and also fragmented data (like Garage for something S3 compatible, which chunks the files), but for backups and longer term storage HDDs are great.
Same with DRAM, old-node chips during the Post-COVID chip shortage, and toilet paper during COVID. Lots of things are commodities until the demand spikes.
I'm still running ~40 WUH721414ALE6L4 purchased ~2020 for $110 ea. They're $320 ea now, used.
I never thought I would own commodity hardware that would increase in value over time. When this AI bubble pops like dotcom 1.0, the definancialization is going to be painful.
Can y'all offer a backup plan targeted at home users with a NAS? The current two tiers either don't support that or are incredibly cost prohibitive to a random schmo who just wants his NAS backed up.
Genuinely curious: What would such an alternative provide to the market? Would it be cheaper, last longer, be some medium that offered different performance/longevity characteristics? There is flash storage which provides some tradeoffs with price (especially now!) and performance. Spinning disks seem to be in a sweet spot of relatively cheap adequately performant with an acceptable lifetime/failure rate for a lot of needs... What market need is missing? (Again - not trying to debate, I am genuinely curious as I am not in the storage industry at all.)
Some cheap and durable archival storage would be nice. Perhaps by storing data in grooves on ceramic discs. (Which is essentially what cds did). Though, preferably without high speed rotation, and reading more than one bit at a time by e.g. taking a picture of the surface through a microscope. How about 1 PB per disk?
I was wondering why we dont have something like that.
Unfortunately, it turns out that, with Blue-ray discs, we're already approaching the optics limit of what visible light (easily) gets us, and they're already cheating with multi-layer storage for the 100GB option. Thus youd need a complicated EUV or electron beam setup for smaller feature sizes.
Say i want my 1PB disk in cd format. A cd has r=6cm. Thus A~=100cm^2. A/1PB= 1.25e-18m^2= 1.25 nm^2, or about 5x5 atoms per bit :/
I guess we're better off by just scaling up flash production and stacking those elements vertically.
Edit: Turns out we need volumetric storage.
By using a material that is transparent to the laser wavelength, by intersecting two precisely focussed beams, the local intensity suffices to absorb two photons at once (stops being transparent) allowing you to select a volumetric point to interact with.
This allows a couple hundred of layers until refraction breaks our resolution.
Seagate continues the tradition of having the highest failure rates of any manufacturer, on average.
Why is that?
As explained at https://www.backblaze.com/blog/backblaze-drive-stats-for-q3-..., a large proportion of Backblaze's Seagate inventory are rather old drives for a datacenter (now 5-9 years in service), so a high failure rate is expected.
I have quantum fireball from 2000, so 26 years old, still going strong.
5 years doesn't seem that long for a drive that cost hundreds of dollars! Persistence is the point.
Just wondering why Seagate seems like the bottom of the barrel in the longevity department. Western Digital drives seem to fail a lot less frequently on average in this dataset and in my life experience.
To Seagate's credit, I do have 8x24TB drives that have been working fine for the past 4 years. Hopefully can last a few more until the compute hardware shortages pass.
That's a brand I haven't heard of in a long time. I had a 8gb HDD from the brand in 2000 until my sister kicked the computer case out of frustration which ended up shorting some chips on the drive. I mourned the loss of my music collection for quite a long time.
Well, I have a 200mb maxtor IDE works just fine to this day.
The classic Deathstars™ ["Deskstar"]
I used a 1.2gb Fireball for my main drive around that era — it is so comically loud!.
Only expected if its Seagate. Backblaze Hitachi drives had miniscule failure rates thru their whole life cycle.
I have two particularly notorious Seagate periods:
I think the last Seagate lines I truly trusted were the ST series of MFM and RLL drives.I do not know, but the last time when I have bought a Seagate HDD, I had a very nasty and unpleasant surprise.
Last year I have bought a 22 TB Seagate Expansion Desktop external HDD, because it was cheaper than the other 22 or 24 TB HDDs available at that time.
I had read carefully its datasheet before buying and there was nothing suspicious there, so I assumed that it must be cheaper just because it is a slow HDD. I did not care about the speed, it was for storing data archives infrequently accessed.
Only after receiving it I discovered what was not said in the datasheet, that this Seagate HDD does not support S.M.A.R.T., so there is no way to test it to see if it works OK and there is no way to discover when errors have happened, e.g. to see when the HDD becomes too old, so you need to migrate your data.
I have never imagined that in 2025 it is possible to buy a HDD that does not support S.M.A.R.T., especially in HDDs with a capacity over 20 TB, and moreover without giving a prominent notice about such a misfeature in the datasheet.
Before this, in 2024 I had bought a 24 TB Seagate SkyHawk, which had S.M.A.R.T., as expected. Since then, after the Seagate Expansion fiasco, I have bought a 22 TB external WD HDD, at the same price with the Seagate, and which has S.M.A.R.T., as it is normal.
I cannot see how removing S.M.A.R.T. support can reduce costs, as it is just a firmware feature. I any case a manufacturer that removes testing and error reporting features from its products clearly does not give a s*t about data corruption and HDD failure rates.
With an external drive the SMART info might be hidden behind the USB-to-SATA bridge, smartctl has support for some of those but sometimes needs to be told with an extra argument.
The Seagate Expansion is made as an external drive by Seagate itself, so it was not put in some random enclosure.
If Seagate has chosen in 2025 to use some archaic bridge that does not pass the SMART commands, it is on them. That would be even more stupid than not implementing SMART in the HDD firmware.
As I have said, the previous external Seagate that I had bought in 2024 had SMART that worked fine over USB. I have a large number of external HDDs, most from WD. Some have been packaged by the HDD vendor as USB drives, others I have assembled myself into enclosures with SATA-to-USB bridges.
On all of them SMART works perfectly, except in this Seagate Expansion Desktop, where the drive replies that SMART is not supported.
Whenever I buy a HDD, I first run the long SMART self-test, to determine whether it can be used safely or I should return it immediately, even if the long self-test takes a couple of days on modern over 20 TB HDDs.
I started to use this procedure after I had some problems with a batch of WD drives, 2 decades ago, where all the drives had very frequent errors since the first few days of use. After running the SMART self-tests, which all failed, the seller could not deny an immediate replacement.
You know, there are external drive USB controllers that linux blocks/blacklists SMART passthrough when using UAS due to paranoia and historical problems, but this can overridden.
https://www.mcgarrah.org/usb-drive-smart/
I guess it is possible this is not your problem, but the last Seagate external I bought in October worked just fine with this workaround. This is probably safe from a data integrity standpoint, at least with a modern filesystem, but in my case it was no issue as I was only using the SMART to do tests before shucking the drive. Also, I don't know of any modern drive that truly doesn't support SMART.
Doesnt surprise me, Seagate is marching to its own drum. My experience defiantly mirrors others' higher than average failure rate as well.
My latest 'fun' experience with them, also, came in the form of an Ironwolf drive which is 'detected' on usb-to-sata interface when plugged in, around %15 of the time. While it starts up consistently on a plain SATA interface. This makes it unusable for what I need. Again, no other drive or MFG ever fails on this usbSata, just the new ironwolf, which it appears is actually for the chineese market, but was sold on newegg, but this is not necessarily seagate's fault, nevertheless.
Do you have MacOS by any chance?
MacOS does not support S.M.A.R.T over USB.
Nope. I use Linux and I also have a great number of external HDDs, both from Seagate and from WD.
On all of them, except on the Seagate Expansion model, SMART works fine over USB.
Seagate Expansion is made as an external drive by Seagate, so it is not some custom enclosure that could have been incompatible with the drive.
>there is no way to discover when errors have happened
There is: use ZFS and scrub.
But yeah, crazy that it doesn't support SMART!
Even without using ZFS (I prefer XFS as significantly faster) all the files that I store have content hash values in extended attributes, for data integrity verification (and also for data deduplication).
Whenever I write that drive, after a power cycle (to be sure that the files are read from disks and not from some cache) I run a script that checks the integrity of the files, to be sure that I can remove them from elsewhere without risking data loss.
With SMART-enabled drives, I usually do that only in the rare cases when a drive reports corrected errors, because I have seen cases when a drive miscorrected some errors, resulting in corrupted files. With a HDD without SMART, when the drives finds errors, but it believes to have corrected them successfully, there is no external sign that something could have gone wrong.
It's regrettable that you would spread misinformation, and easily explained/debunked information at that. This is a reasonably well-known issue and even with the extremely bad and limited information you provided (no model numbers or other info), I already know exactly what your problem is.
It actually looks like they're getting better, if the changes from last year to this year are any indication.
"back in my day", seagate was "the shit". only much later, hitachi drives came to be popular and wd, sort of.
By Hitachi, you mean IBM Ultrastar. IBM drives tended to be the best.
>IBM drives tended to be the best.
The IBM Deskstar 75GXP entirely earned its nickname of Deathstar.
https://en.wikipedia.org/wiki/Deskstar#IBM_Deskstar_75GXP_fa...
nope. i don't even think ibm were being widely sold at all. ultrastar rings a bell, somewhat, but definitely not ibm hard drives. and i'm talking early 2000s here.
Flash storage costs have gone way up.
I wonder if backblaze's business has seen any changes given that their assets are platter drives
Retail HDD prices have also gone up ~50% over the past couple of months.
Pretty sure that's cos every AI company wants to have a copy of all the worlds information stored in each of their data centers, and hard drives are the best storage medium for that.
Design your training strategy carefully and you can do streaming rather than random reads from the drives and get enough performance.
In modern storage tech, network is the i/o bottleneck, but HDD
Thankfully for personal stuff there's at least the option of buying white label or recertified drives.
Recently got some 1 TB ones from Datablocks as backups, did a short writeup on my blog: https://blog.kronis.dev/blog/buying-some-drives-from-datablo... (though a better post made its way on HN by someone else a while ago, too)
To be honest, they mostly have 8 TB and up in stock at the moment, but the price was around 35 EUR for the 1 TB drive which is kind of insane when similar drives at local stores cost 110 EUR right now - I remember when those Seagate Barracuda drives were about 40-50 EUR brand new!
For the US, there's also goHardDrive, though I haven't used them myself: https://www.goharddrive.com/category-s/69.htm
I still use SSDs for boot drives, video games and also fragmented data (like Garage for something S3 compatible, which chunks the files), but for backups and longer term storage HDDs are great.
Interesting, SSD have kind doubled.. storage used to be such a commodity item and now it's a gold rush
Same with DRAM, old-node chips during the Post-COVID chip shortage, and toilet paper during COVID. Lots of things are commodities until the demand spikes.
My drives have an AFR of 0.41
It looks like I picked a good vintage which is good because the same drives are approaching 2x the price today.
I'm still running ~40 WUH721414ALE6L4 purchased ~2020 for $110 ea. They're $320 ea now, used.
I never thought I would own commodity hardware that would increase in value over time. When this AI bubble pops like dotcom 1.0, the definancialization is going to be painful.
Super interesting thanks for sharing.
Can y'all offer a backup plan targeted at home users with a NAS? The current two tiers either don't support that or are incredibly cost prohibitive to a random schmo who just wants his NAS backed up.
"As of the end of 2025, Backblaze was monitoring 341,664 drives used to store data."
Three hundred thousand drives. That figure seems insane. Google and AWS must have truly staggering amounts.
The sad thing is that they need 330k hard drives to store a shitton of data. And there are no hard drive alternatives on the market.
Genuinely curious: What would such an alternative provide to the market? Would it be cheaper, last longer, be some medium that offered different performance/longevity characteristics? There is flash storage which provides some tradeoffs with price (especially now!) and performance. Spinning disks seem to be in a sweet spot of relatively cheap adequately performant with an acceptable lifetime/failure rate for a lot of needs... What market need is missing? (Again - not trying to debate, I am genuinely curious as I am not in the storage industry at all.)
Some cheap and durable archival storage would be nice. Perhaps by storing data in grooves on ceramic discs. (Which is essentially what cds did). Though, preferably without high speed rotation, and reading more than one bit at a time by e.g. taking a picture of the surface through a microscope. How about 1 PB per disk?
I was wondering why we dont have something like that.
Unfortunately, it turns out that, with Blue-ray discs, we're already approaching the optics limit of what visible light (easily) gets us, and they're already cheating with multi-layer storage for the 100GB option. Thus youd need a complicated EUV or electron beam setup for smaller feature sizes.
Say i want my 1PB disk in cd format. A cd has r=6cm. Thus A~=100cm^2. A/1PB= 1.25e-18m^2= 1.25 nm^2, or about 5x5 atoms per bit :/
I guess we're better off by just scaling up flash production and stacking those elements vertically.
Edit: Turns out we need volumetric storage. By using a material that is transparent to the laser wavelength, by intersecting two precisely focussed beams, the local intensity suffices to absorb two photons at once (stops being transparent) allowing you to select a volumetric point to interact with. This allows a couple hundred of layers until refraction breaks our resolution.