Hard drive failures--time or useage?

My online backup would count as that second location. I agree with having two backups, not just one.

My computer is a Mac so I keep one drive permanently attached for hourly Time Machine backups. I keep a second in one of those locking pistol cases in the trunk of my car and backup to it every couple of weeks. I like having a drive that’s not in the same location as the computer in case there’s a catastrophe like a fire or burglary.

I recently replaced my external HDD with an SSD and the difference is remarkable.
 
Time if the drive is run continuously 24/7 or usage if it's not. For a back up drive the question should be all but irrelevant because a single back up copy should never be relied on. There should always be more than one copy and they should not all be stored in the same building. Most people can achieve that easily by having hard local copies and cloud copies.
 
Interesting point there. If I have all my files backed up to the cloud, then I can just keep using this same old hard drive until it fails.

I have some of the files only backed up to the external hard drive, and some of them backed up to the cloud and to the external drive. If I were to back up everything to the cloud, then I could just use this external drive until it fails. At some point, 2 tb would be too small, but that won't be for a while.

Time if the drive is run continuously 24/7 or usage if it's not. For a back up drive the question should be all but irrelevant because a single back up copy should never be relied on. There should always be more than one copy and they should not all be stored in the same building. Most people can achieve that easily by having hard local copies and cloud copies.
 
I've worked with both HDD and SSD industry before.

If you are using it just for archiving or backup, HDD has a data retention by design for 5 years. It doesn't mean it will die after 5 years, just that whatever you wrote to it 5 years ago will start losing its magnetic field after that so you should rewrite it again. In comparison the power off data retention of SSD in room temp is 1 year, and if you turn it off and leave it in a drawer the data stored in its floating gates or whatever gates it uses would stay for 1 year then gradually lose accuracy. If you keep its power on constantly then it will over time refresh itself.

I have a Seagate from 2004 that was still going, that's before their reliability decline days, I keep it at my parents' home just in case my house burn down and lose all my photos.

Most HDD failure come from either design issue, bearing wear and tear, lubricant between media and head deteriorate or spread unevenly, and over time data retention loss on data written long ago but not rewritten again (5 years). If you don't drop or bump your drive they tend to last a very long time.
 
Hard drive failures are completely random. Drives with even occasional use can die randomly just as a drive that has run a long time.

I run my drives until they fail or are too small to be practical. I keep 3 backups and store one drive offsite at all times so whether a drive dies or not is irrelevant. This is how I store all precious family photos / videos, childhood home videos, etc.

The simple answer is to buy a new drive and use the old as a redundant backup.

For long term storage I only use spinning hard drives. SSDs aren’t suitable for long term storage.
 
Years ago, there was a study by Google of the non-SSD hard drives that they used for storage re. lifespan.

Dunno if the science still holds....but back then drive failures were front-loaded. If the drive survived the first 6 months, it is likely to last past the published "mean time to failure" and the failure randomly at that time

if you want your drive to last as long as possible, the only real factors under your control are to keep it away from heat (make sure your computer is appropriately cooled) and hard shocks (we're talking long drops onto concrete, etc.)

https://research.google.com/archive/disk_failures.pdf
https://search.brave.com/search?q=google+hard+drive+study
 
HDD lifespan is not random at all. They are dictated by how many read write arm strokes, head flying hours, and mechanical shock taken. Obviously design and manufacturing defects happen but within the same design they are not random at all, and data centers have statistics to know when to retire drives before they reach unsafe lifespan, and stop shipment from manufacturers if the defective rates are above what they will put up with.

Most are design to target 5 years life expectancy when running typical workload they are designed for and given 1 year warranty.
 
HDD lifespan is not random at all. They are dictated by how many read write arm strokes, head flying hours, and mechanical shock taken.

Seeing inside a hard drive while it's working would make us surprised they last as long as they do.

We did just that years ago when I worked in IT at a school. One of the technicians made a transparent vacuum moulded cover for an old hard drive. We compromised it's life of course by removing and replacing the cover outside of a clean room but it worked for a good while running Windows as a demonstration unit for the students. What you see is the heads flying back and forth at a tremendous speed. Hard discs don't work sequentially like a gramophone record. In order to minimise seek time, when a drive writes data it starts in the middle track and then oscillates back and forth either side of the middle hence the frantic motion of the heads. Except when parked the heads don't touch the disc, they float over it, so any likely mechanical wear is going to be in the head assembly.

Hands up who's old enough to remember when you had to manually park the heads on a hard drive before shutting down. Parking was positioning the heads on an unused part of the platter so it didn't physically damage the platter when started up again
 
Hands up who's old enough to remember when you had to manually park the heads on a hard drive before shutting down. Parking was positioning the heads on an unused part of the platter so it didn't physically damage the platter when started up again
(y)
 
We use these on system critical backups, have several and they are very fast

1746878186038.webp


Another way is to install Google Drive Desktop and have it mirror automatically folders with important data
A small business license for a user is currently only $14/mo and provides several terabytes of Google Drive space
 
Seeing inside a hard drive while it's working would make us surprised they last as long as they do.

We did just that years ago when I worked in IT at a school. One of the technicians made a transparent vacuum moulded cover for an old hard drive. We compromised it's life of course by removing and replacing the cover outside of a clean room but it worked for a good while running Windows as a demonstration unit for the students. What you see is the heads flying back and forth at a tremendous speed. Hard discs don't work sequentially like a gramophone record. In order to minimise seek time, when a drive writes data it starts in the middle track and then oscillates back and forth either side of the middle hence the frantic motion of the heads. Except when parked the heads don't touch the disc, they float over it, so any likely mechanical wear is going to be in the head assembly.

Hands up who's old enough to remember when you had to manually park the heads on a hard drive before shutting down. Parking was positioning the heads on an unused part of the platter so it didn't physically damage the platter when started up again
Even in the mid 2000s they don't park the head on an unused area of the media anymore. They use the inertia of the platter to regen power to move the actuator arms onto a ramp. The only time the head actually touches the medias are probably when they start and unload from the ramp down to the medias (like unloading a parasail off the boat above water).

Seek do cause some minor amount of wear (head fly height are not only controlled by rotational speed and air pressure, but also a heater to expand a variable amount of distance between 2 metal parts on the actuator, small amount of flexing during seek likely can cause some amount of media contacts), and have enough statistics in data center drives to show between frequently used drives vs archive only drives. For home users they likely won't use enough and most damages are either bad designs (flying too low?) or mishandling (bumping the drive while writing (writing to the nearby tracks) or dropping the drive in transit.
 
Last edited:
Back
Top Bottom