Assuming that like me you use ypur retro PCs solely for gaming and nothing mission critical - why replace a working part in an already assembled system?

Regular backups of important data are necessary regardless of drive age.
Catastrophical failure that destroys other hardware is probably not more likely on old HDDs than on newer flash storage.

It is like with LAN cards, modems, 2D or 3D acceleration and everything. It all depends on the cpu. And how good is the acceleration.

Think of IDE as a verry fast framebuffer and SCSI as a 2D accelerated. Thats simple.

maxtherabbit wrote:

Smack2k wrote:

Thanks for the information....

This is an old 386 DX-33. The SCSI Card is an Adaptec AHA-1522A currently. I think I am going to switch over to CF Card just for peace of mind that the drive wont suddenly crash as the drive is years old....

my argument would be: "if the drive has survived this long without failing, it will likely keep going"

for just about everything after the ST-502 era, my experience has been that the vast majority of drive failures that aren't directly attributable to physical abuse, occur within the first 2-3 years of operation - if the drive has survived 10+ years I'd have a tremendous degree of confidence that it lacks manufacturing defects

Errr... That is not how mechanical devices work. "Well, if your car has made it past the first 20 thousand miles, it will likely last forever."
This sort of thinking makes no sense. HDDs are under wear and tear every minute they are in use. The longer they have been in use, the higher the chance of failure becomes. The vast majority of drive don't die because of manufacturing faults, but rather as a result of natural end of life. For a HDD, failing motor, failing magnetic field, failed head(s) etc. are all natural causes.

I understand mechanical wear, but my contention is that this:

LunarG wrote:

The vast majority of drive don't die because of manufacturing faults, but rather as a result of natural end of life.

is simply untrue. I know all mechanical devices will someday wear out, but in my experience that's not what kills the vast majority of hard drives.

most of my dead drives have either been DOA or died in their first few months of use or due to improper handling, so I agree that natural end of life is probably not the vast majority ^^

maxtherabbit wrote:

I understand mechanical wear, but my contention is that this:

LunarG wrote:

The vast majority of drive don't die because of manufacturing faults, but rather as a result of natural end of life.

is simply untrue. I know all mechanical devices will someday wear out, but in my experience that's not what kills the vast majority of hard drives.

You chose to refer to only part of my statement... If the motor in a HDD fails because of long time friction, that is natural end of life. If the disk surface's magnetic field deteriorated enough that it starts developing bad sectors, that is natural end of life. If the head after 5 years suddenly violently slams into the disk surface in a catastrophic way, this is natural end of life. These are mechanical devices, not living beings. Failure after end of warranty basically constitutes natural end of life. The older a drive gets the more likely it is to suddenly fail. That is inevitable. Some lasts a looong time past the MTBF though, which is great, and generally SCSI drives were, and still are, more expensive than ATA and SATA drives and are usually made to higher standards. But once they pass their 10th anniversary, they are certainly living on borrowed time. Just because they've "lasted this long" doesn't mean they will last forever. In fact, quite the opposite.

imi wrote:

most of my dead drives have either been DOA or died in their first few months of use or due to improper handling, so I agree that natural end of life is probably not the vast majority ^^

Somebody breaking a drive by improper handling is a competely different thing. This is something the manufacturer has no control over. Personally I have never received a new DoA drive.
I have had drives fail prematurely in the past. Quantum Fireball drives to be exact, and at that time Quantum even made a public announcenent that a large number of drives had a manufacturing fault. I got mine replaced, and had no further issues.

LunarG wrote:

maxtherabbit wrote:

I understand mechanical wear, but my contention is that this:

LunarG wrote:

The vast majority of drive don't die because of manufacturing faults, but rather as a result of natural end of life.

is simply untrue. I know all mechanical devices will someday wear out, but in my experience that's not what kills the vast majority of hard drives.

You chose to refer to only part of my statement... If the motor in a HDD fails because of long time friction, that is natural end of life. If the disk surface's magnetic field deteriorated enough that it starts developing bad sectors, that is natural end of life. If the head after 5 years suddenly violently slams into the disk surface in a catastrophic way, this is natural end of life. These are mechanical devices, not living beings. Failure after end of warranty basically constitutes natural end of life. The older a drive gets the more likely it is to suddenly fail. That is inevitable. Some lasts a looong time past the MTBF though, which is great, and generally SCSI drives were, and still are, more expensive than ATA and SATA drives and are usually made to higher standards. But once they pass their 10th anniversary, they are certainly living on borrowed time. Just because they've "lasted this long" doesn't mean they will last forever. In fact, quite the opposite.

a 10yr old working drive is quite likely to become a 20, 30 or even 40yr old working drive - I never even said "forever" that was your word

the level of mechanical wear incurred by HDD head and spindle bearings in normal use is probably several orders of magnitude lower than that of a crankshaft or connecting rod bearing, the car analogy is really not appropriate

maxtherabbit wrote:

obviously I didn't mean "forever" literally, but I do think that a 10yr old working drive is quite likely to become a 20, 30 or even 40yr old working drive

Obviously only working 10 years old drives can become 20, 30 or 40yr working drives, But I have a problem with that "quite likely".

The statistics captured at very high samples shows that the like-hood of failure is significantly higher as the drive gets older. And once you reach 3 or 4 years (depending on source) the failure rate starts to grow exponentially. A drive a this age is actually more likely to fail than the very high likehood of early failure due to manufacturing defects.

Your argument that it survived so long so it will work fine could work, but only during first 3 years of drive's life or so.

If you have a working 20+ drive, you should be thanking God every-time you successfully start it.

https://en.wikipedia.org/wiki/Bathtub_curve

mpe wrote:

If you have a working 20+ drive, you should be thanking God every-time you successfully start it.

I have armloads of working 20+ drives, and I don't believe in god

most drives get retired before they're even 10 years old, hence why drive failures are more likely to be premature ones, because the late drive failures of old age simply won't happen.
of course if you would use every drive till the end of it's life span quite literally... then death of old age would be more common, but that's just not the case :p

I am also running a 30 years old ST251 in my Macintosh II. Fingers crossed 😀