How Long Can a Hard Disk Last in Constant Use?

  • Thread starter Thread starter Vanadium 50
  • Start date Start date
  • Tags Tags
    longevity
AI Thread Summary
The discussion highlights the durability and longevity of hard drives, with one user sharing their experience of a Western Digital Re disk that has operated for over 12 years without any issues, including no bad or relocated sectors. Another user mentions a 19-year-old drive that still functions, albeit unused for seven years. The conversation emphasizes the unreliability of anecdotal evidence for assessing failure rates, noting that while some drives perform exceptionally well, others can fail unexpectedly. The participants discuss the evolution of hard drive technology, contrasting the higher failure rates of HDDs with their greater recoverability compared to SSDs. There are mentions of personal setups, including a NAS RAID system with older drives and the importance of having spares on hand. Concerns about heat affecting the lifespan of consumer-grade drives are also raised, alongside the practice of conducting regular SMART tests to monitor drive health.
Vanadium 50
Staff Emeritus
Science Advisor
Education Advisor
Gold Member
Messages
35,003
Reaction score
21,702
There are some messages complaining how this, that or the other didn't last. I thought I'd share a counterexample. I was looking at SMART logs and one of my disks has been in constant use for over 12 years - a Western Digital Re.

I believe this is post "Caviar" and pre "Colors". 100,000+ hours in service. No bad sectors. No relocated sectors. I'm impressed.

Slightly less impressive was a 19 year old drive that still works, although its been in a box for 7 years.
 
Computer science news on Phys.org
Indeed. I have a 1 terabyte drive that's been running for probably 8-10 years now with no issues.
 
Anecdotal evidence is a bad thing to rely on for failure rates. Yes, you can be lucky, but you also can have bad luck.
 
More impressive numbers: 43 billion rotations, and almost 7 million miles of travel (at the edge of the platter).

It's also remarkable how the technology has changed: in the first 20 years of IBM PCs you had (in the non-SCSI market) MFM, RLL, ESDI, PATA/IDE and then SATA. Now we have SATA.

FactChecker said:
Anecdotal evidence is a bad thing to rely on for failure rates.
I don't think anyone has said that one good drive is all the testing you need.
 
  • Like
Likes nsaspook and FactChecker
Vanadium 50 said:
I don't think anyone has said that one good drive is all the testing you need.
That's true. The situation is complicated. HDDs can have a higher failure rate but are much more likely to be recoverable. A failed sector on a hard drive contains less data than blocks or chips on an SSD so the data is more likely to be recoverable. See this.
 
I've just set up a NAS RAID around half a year ago with already used (common) disks.

It has an UPS and won't be switched off, ever.

I expect it to serve at least a few years.

... just had to replace a half year old laptop HDD recently. Hurled left and right; switched off daily.
 
My NAS' power supply died thir morning. :frown:

It made that forms in the garbage disposal sound and a minute later than was it. I have a new supply, but connecting it up is harder than it might seem. Lots of cables in a small space.

FWIW, every single disk in it is >5 years old. I have a new spare on the shelf in cased one goes bad, and a spare in the NAS which gets used automatically.
 
Rive said:
I've just set up a NAS RAID around half a year ago
As an aside, I am using mostly NAS and Enterprise drives on mine. I started with "regular" drives and expanded when they failed, and what's left are largely NAS and Enterprise. Two Toshiba X300s (lasted much longer than the N300)'s, 2 IronWolves (or is that Wolfs?) and 3 Seagate Enterprise Capacities (formerly Constellation). Also a HGST Ultrastar for backup.

My suspicion is that heat slowly kills consumer-grade drives.
 
  • #10
Vanadium 50 said:
No bad sectors.
There are probably some small defects in the surface, but error correction has enough margin to correct for this.
 
  • #11
Strictly speaking, no relocated sectors. I do a weekly SMART long test, which should pick up problems.
 

Similar threads

Replies
13
Views
3K
Back
Top