The end of the Hard Disks

[Zoli]

...
Be really careful with optical media for long-term storage. In a place I used to work, we would have old archives, first on CD, then DVD and eventually Bluray. I experienced problems which ranged from discs being rendered totally unreadable, to certain sectors being unreadable (which is no good for archive files unless there is redundancy built-into the file). And we're talking disks that ranged from 10 to only a few years old.

It was a standard office-type set up, but after hours, the air conditioning would shut off. I'm not sure whether that contributed significantly to the humidity and therefore accelerated the damage to the discs or not?
...

That's typical for the dye-based disks; however, the M-disks are based on different technology. I would place them in similar category like the mass-produced/printed disks for longevity.

...
The moral of the story is test and rotate your backups often.

That's absolutely true for all the backup media types.

[David Hess]

Just for kicks I broke out some old DVDs that that I burned 18 years ago and read them with both drives without problems.

[coppercone2]

https://hardware.slashdot.org/story/22/08/02/1837229/the-dirty-carbon-secret-behind-solid-state-memory-drives

[Mechatrommer]

a little bit too duplicated...

Looks like HDDs are greener than SSDs:
The Dirty Carbon Secret Behind Solid State Memory Drives (https://www.discovermagazine.com/technology/the-dirty-carbon-secret-behind-solid-state-memory-drives)

doesnt that mean more trees can breath on co2 and in return produce more o2 for us to breath on, and then we can multiply more until the earth is overcrowded, and then we'll either need to move to mars or kill each other to rebalance the population count.

[BruceEckel]

Clicking sound from a HDD is so called “Click of death” because that means the Hard drive is now broken and must be repaired or replaced.

HDD and SSD can and do fail suddenly. I have had a 1TB external drive fail on me like this, it showed no prior signs of damage before it suddenly developed click of death.

[tooki]

Lemme fix that for you:

Clicking sound from a HDD is so called “Click of death” because that means the Hard drive is now broken and must be repaired or replaced.

For all intents and purposes, hard disks cannot be repaired.

HDD and SSD can and do fail suddenly. I have had a 1TB external drive fail on me like this, it showed no prior signs of damage before it suddenly developed click of death.

For the sake of data integrity and avoiding failures, I replace them long before they start clicking, by using SMART to identify issues. Not all failures can be predicted, but some SMART parameters (or rapid changes in them) are strong predictors of imminent failure.

[Mechatrommer]

Lemme fix that for you:

Clicking sound from a HDD is so called “Click of death” because that means the Hard drive is now broken and must be repaired or replaced.

For all intents and purposes, hard disks cannot be repaired.

it depends, sometime it can...

For the sake of data integrity and avoiding failures, I replace them long before they start clicking, by using SMART to identify issues. Not all failures can be predicted, but some SMART parameters (or rapid changes in them) are strong predictors of imminent failure.

power glitch can cause some SMART parm to be marked. and disk clone will clone the SMART parm as well, so a marked SMART value as bad, can be all perfect green again. i set 10 years as limit, or if i start to hear weird sound, i dont usually remember to check SMART values.

[Kjelt]

For all intents and purposes, hard disks cannot be repaired.

Repair as in you continue using it as a normal hdd no, but to get the data back you can some times.

If the error is in the drive controller board you can buy a donor hdd and replace one rom from the original to the donor board and swap them.
Example


Sometimes it is as simple as blown resistors/fuse/diode ofcourse you have to ask what caused this but hey if it works for a couple of hours that is enough to get the data off it.
Example


If it is a mechanical issue inside the drive then you better hire a pro company esp with current helium filled disks.
So best advice: back everything up !

[tooki]

Lemme fix that for you:

Clicking sound from a HDD is so called “Click of death” because that means the Hard drive is now broken and must be repaired or replaced.

For all intents and purposes, hard disks cannot be repaired.

it depends, sometime it can...

A band-aid fix to recover data is NOT A REPAIR, and that video even says so.

power glitch can cause some SMART parm to be marked.

Not the ones that matter in regard to predicting failure.

and disk clone will clone the SMART parm as well,

Complete nonsense.

so a marked SMART value as bad, can be all perfect green again.

Certain test logs have a limited number of entries, such that the oldest tests get cycles out.

Not the important stuff though.

[tooki]

For all intents and purposes, hard disks cannot be repaired.

Repair as in you continue using it as a normal hdd no, but to get the data back you can some times.

If the error is in the drive controller board you can buy a donor hdd and replace one rom from the original to the donor board and swap them.
Example


Sometimes it is as simple as blown resistors/fuse/diode ofcourse you have to ask what caused this but hey if it works for a couple of hours that is enough to get the data off it.
Example


If it is a mechanical issue inside the drive then you better hire a pro company esp with current helium filled disks.
So best advice: back everything up !

I said “for all intents and purposes”, not “under all circumstances without exception”. Data recovery (as a result of a failure to back up properly) is the ONLY situation where any work on a failed or failing hard drive makes sense, and then the purpose is not to restore it regular use. In any situation other than data recovery, the labor cost of a repair attempt will exceed the cost of replacing the drive with a new one.

[james_s]

When I was a broke teenager somebody gave me a pair of dead 1GB hard drives, quite large by the standards of the day. I swapped the PCB from one onto the other and got one of the drives working and used it for several years. I also repaired a 8GB SCSI drive once that had a drink spilled on the computer while it was running, my friend had his whole music collection on it. Replacing one of the mosfets driving the spindle motor fixed it and he was still using it when I lost touch with him a few years later. That latter could be called data recovery, the former was a result of poverty and having more time than money.

[Kjelt]

I said “for all intents and purposes”, not “under all circumstances without exception”. Data recovery (as a result of a failure to back up properly) is the ONLY situation where any work on a failed or failing hard drive makes sense, and then the purpose is not to restore it regular use. In any situation other than data recovery, the labor cost of a repair attempt will exceed the cost of replacing the drive with a new one.

Well as a non native english speaker I do count "backup of data" as a valid purpose to repair a hdd, other cases indeed it makes no economical sense also regarding reliability etc.
But as far as reliability goes, then the same goes for instance any electrical device having suffered severe ESD due to lightning or whatever and debatable also any T&M device that has suffered damaged on its measurements inputs, at least if you need the data to be exact and reliable, unless done properly , recalibrated and certified by the manufacturer.

[tooki]

I said “for all intents and purposes”, not “under all circumstances without exception”. Data recovery (as a result of a failure to back up properly) is the ONLY situation where any work on a failed or failing hard drive makes sense, and then the purpose is not to restore it regular use. In any situation other than data recovery, the labor cost of a repair attempt will exceed the cost of replacing the drive with a new one.

Well as a non native english speaker I do count "backup of data" as a valid purpose to repair a hdd, other cases indeed it makes no economical sense also regarding reliability etc.

To me, as a native English speaker, "repair" has the connotation of a durable repair intended to return the item to regular service. Otherwise, we'd call it a "temporary repair", temporary fix", etc.

(FYI, "backup" specifically means making an extra preventative copy in advance to protect against data loss. Copying data off after drive failure is data recovery, the exact opposite of backing up.)

Anything involving anyone but the manufacturer or manufacturer-authorized data recovery provider opening the drive (as in the reply above yours) is, given the fickle nature of HDD mechanisms, inherently a quick-and-dirty fix, not a lasting repair, since letting non-cleanroom air into the drive is asking for trouble.

But as far as reliability goes, then the same goes for instance any electrical device having suffered severe ESD due to lightning or whatever and debatable also any T&M device that has suffered damaged on its measurements inputs, at least if you need the data to be exact and reliable, unless done properly , recalibrated and certified by the manufacturer.

That's comparing apples to oranges. Or apples to pencil erasers. Completely irrelevant. We're not talking about drives that have failed due to abuse. The issue is that non-abused hard disks will just... fail. And almost all of the time, it's mechanical failure.

[Kjelt]

True. They can fail as anything mechanical, electric , pneumatic etc can fail.
I don't think harddrives are particular worse if taken care of (used) properly.
In datacenters using enterprise grade harddisks that are 24/7 on in temperature and humidity controlled environment 99% if not higher of the harddisks are preventively changed after three or four years without any failure.
Personally from the 74 harddisks I ever owned I have one fail and one has a SMART entry which is not good. A lot of these harddisks were never 24/7 on, were moved, shaken, temperature cycles, lots of power on/off cycles etc.

You probably know the backblaze reports, they reveal that the AFR of hdds is equal to ssds now who had put their money on that ?

https://www.dpreview.com/news/7921185885/backblaze-reveals-failure-rates-for-hard-drives-in-2021-releases-an-ssd-report-for-the-first-time

[CatalinaWOW]

True. They can fail as anything mechanical, electric , pneumatic etc can fail.
I don't think harddrives are particular worse if taken care of (used) properly.
In datacenters using enterprise grade harddisks that are 24/7 on in temperature and humidity controlled environment 99% if not higher of the harddisks are preventively changed after three or four years without any failure.
Personally from the 74 harddisks I ever owned I have one fail and one has a SMART entry which is not good. A lot of these harddisks were never 24/7 on, were moved, shaken, temperature cycles, lots of power on/off cycles etc.

You probably know the backblaze reports, they reveal that the AFR of hdds is equal to ssds now who had put their money on that ?

https://www.dpreview.com/news/7921185885/backblaze-reveals-failure-rates-for-hard-drives-in-2021-releases-an-ssd-report-for-the-first-time

I agree that hard drive reliability is not a particular cause for concern, but my personal experience has been significantly worse than what you report.  My total quantity is smaller, though I don't have an exact count.  It is in the 30-40 range.  And I have had two hard failures, one drive that developed start-up reliability failures and one that got a SMART entry and was losing/corrupting data.  Perhaps not surprisingly the failures have occurred in drives 7-12 years old.  New enough to be relatively high capacity with tight tolerances and less margins and old enough to have seen significant use.  The failure rate has been high enough to keep me reminded of the importance of backups and fortunately I haven't lost any meaningful data for a couple of decades. 

I do also have to modify that last statement.   As the total amount of stored material has increased I have lost data - not because it wasn't on the drives, but because I can't locate it.  I know this happens because I have found some of it, sometimes years after the need.

[Kjelt]

Perhaps not surprisingly the failures have occurred in drives 7-12 years old.  New enough to be relatively high capacity with tight tolerances and less margins and old enough to have seen significant use. 

Check the smart values, probably high start stop count?
I only buy enterprise grade for my NASses and minimize the power cycles for the 24/7 on devices. The others I just turn on when I need them.
This costs energy for standby but does prolong the life , but perhaps that is just a believe of mine. Anyway I was planning to also preventively replace my 24/7 hdds and nas every 4 years but due to the bad economy and relatively low feature new synology nasses (still no 10Gb NIC) I kept waiting. 46000+ hours not bad and still going strong.

[Connecteur]

Rummaging through some old stuff today, I found my old music player, a Panasonic SV-SD50.  I don't exactly know how long it was sitting idle, but I think it was at least 20 years.  It had in it one of the biggest capacity SD cards I could afford at the time, 256 MB.

I've always been curious about just how long an SD card will hold data; I'd heard 10 years mentioned several times, but I'm sure that was a guess.  Anyhow, I put a AAA battery into the player, and all the music was still there, intact.  I couldn't say if such longevity applies to more modern SD and micro-SD cards, but they did use larger dies back then, so i couldn't say for sure.  But I was surprised at how long this one retained data.

[Kjelt]

Couple of years later, hdd's still going strong bought 5x 18TB for backup of all my stuff.
Last year bought a new NAS with SSD's but due to the price kept it to 4x4TB.
Since then they became more expensive (+30%) 

[Gyro]

Rummaging through some old stuff today, I found my old music player, a Panasonic SV-SD50.  I don't exactly know how long it was sitting idle, but I think it was at least 20 years.  It had in it one of the biggest capacity SD cards I could afford at the time, 256 MB.

I've always been curious about just how long an SD card will hold data; I'd heard 10 years mentioned several times, but I'm sure that was a guess.  Anyhow, I put a AAA battery into the player, and all the music was still there, intact.  I couldn't say if such longevity applies to more modern SD and micro-SD cards, but they did use larger dies back then, so i couldn't say for sure.  But I was surprised at how long this one retained data.

Double post warning. See main thread...  https://www.eevblog.com/forum/chat/longevity-of-sd-data/

[rkabz]

The hard disk will never die...die-hard computer nerds will keep them alive forever! Pic of still working Seagate ST251, no additional defects, no untoward bearing noise.

[DiTBho]

Yesterday, after two months! of waiting, due to various problems with customs in Japan or I don't know what other problems importing things, I finally received my package!
Contains the fastest Toshiba SCSI DVD-RAM/120mm ever made! and never marketed outside Japan

• Single-sided, one layer, 4.7 GB, 120mm cartridges
• Double-sided, single layer, 9.4GB, 120mm cartridge

and a Fujitsu SCSI DVD-RAM/80mm

• Single-sided, one layer, 1.46 GB, 80mm cartridge

And I'm really happy. It's a big, heavy, surprisingly quiet unit that only takes cartridges, just the way I like it!
On the one hand the 4GB cartridges are more sufficient for my backups and on the other hand the DVD-RAMs really help to manage the HDDs regarding data redundancy.

The unit that reads smaller 80mm cartridges looks a bit like that weird data-media thing Neo uses in the first Matrix movie when the white rabbit knows at his door 

[Kjelt]

Yesterday, after two months! of waiting, due to various problems with customs in Japan or I don't know what other problems importing things, I finally received my package!
Contains the fastest Toshiba SCSI DVD-RAM/120mm ever made! and never marketed outside Japan

Unless you have some retro hardware that needs these why would you now start with using obsolete hardware?
If something dies you need to scavenge the planet for attick finds from Japan.
Cool yes for in the display cabinet, practical 

[DiTBho]

Unless you have some retro hardware that needs these why would you now start with using obsolete hardware?
If something dies you need to scavenge the planet for attick finds from Japan.
Cool yes for in the display cabinet, practical 

What would be the modern alternative? Use those crappy USB sticks that lose data if you don't power them? Using SSD as backup is even worse.
Yes, I also support both of them, because customers want them, but I don't use them for my personal needs.

Where I started working in 2001, they took 16 DVD-RAM v2.0 units around 2003, and they're all still working.
Now I no longer work as an internal collaborator, now I'm a freelancer, every now and then they send me DVD-RAMs with documentation and (C, asm, Octave, ... ) sources of things they want me to fix.

I have two other DVD-RAM drives installed.
The two I imported from Japan in particular have

• the 120mm drive
  • it's about 2X times faster than the fastest DVD drive marketed in Europe and America
  • better still - it has a 32Mbyte buffer, never seen before, very useful as a "disk cache", especially if you want to use DVD-RAM as "laser memory" (there are jbox software that make abundant use of it)
  • integrates an RFID reader/writer. If the cartridge has an eeprom-contactless tube, it is very convenient to use the RFID as a "label" or "index" for the contents of the DVDRAM without even having to read it!
  • It has all metal gears, interchangeable belts, high-end motor and bearings, metal rails. Mechanics easily removable in every single part, and it's 2U tall, so it has excellent airflow too!
• the 80mm ... well there are very few 80mm units. It has nothing special other than supporting a small and rare format. I don't think I'll use it seriously, but I took a dozen 80mm cartridges

For me DVD-RAM is very practical to handle, has no trouble traveling, and is very long-lasting. It proved this with facts. While many CD-RWs burned (by a Plextor at 1X speed) in 2008 are already unreadable, no DVD-RAM burned in 2004 has ever lost a single byte!

Other than that, I'm not using hard to find cartridges, and I got about 400 from Japan, 120mm/4GB cartridges are common, but these have a 2Kbyte RFID, which the Toschiba DVD-RAM is able to read and write .

Then on the Linux kernel side... the driver is a pleasure 

[Kjelt]

What would be the modern alternative? 

Portable exchangeable durable: don't know.

My usecases are different, 4GB is what I create in two days on holiday pictures alone.
Documentation code etc. my data directory now is about 1.5TB
So I use NASses.
Backed up two dayly and again weekly on diferent NASses available on evry mobile or fixed device in the house.
For data transfer to 3rd parties I use portable SSDs.

[DiTBho]

My usecases are different, 4GB is what I create in two days on holiday pictures alone.
Documentation code etc. my data directory now is about 1.5TB
So I use NASses.
Backed up two dayly and again weekly on diferent NASses available on evry mobile or fixed device in the house.
For data transfer to 3rd parties I use portable SSDs.

Yes, I was talking about source code (text files), which is what I'm most interested in, since that's what I do for a living.
Typically it's less than 4,498,528 bytes total for 1 year of projects, which must have several redundant back-ups.
DVD-RAM is just one of them, but it seems the most reliable.

It saved me from bad situations in which both the USB stick and the SSD one of customers left unused for 1 year had corrupted the data.
Obviously I also had other copies at home, but I didn't have them with me at the time, while I had the DVD-RAMs in a box in the car.

For some things that I absolutely cannot lose, I also have a WORM copy.
(If you work with government agencies, or if you do avionics work, sooner or later someone will ask you to burn a WORM)

For bigger things, like photos, audio (Wav, mp3, ogs), video (H264/265, MPG4), etc. HDDs and tapes are the best and only possible solution, especially due to the time it takes to write large amounts of data.
Still SSD, no. Sure they are faster than HDDs, but Old and modern Flash is no good for secure data storage.

Unless you buy several SSDs and cycle them through, copying data from one to the other, and always with another backup medium in parallel.