Is leaving your computer on all the time (24/7) better for hard drive?

westom · 23 Oct 2012

Damage is due to starting and stopping. Most aggressive and harmful starting and stopping is when heads move. Most drive platter damage is due to contamination even from wear of moving heads and spinning disk. To increase drive life expectancy, eliminate operation when hardware need not be operational. Power off when done to increase hardware life expectancy.

Also destructive is transistor switching. Most damage occurs during an event so violent that its junction even outputs a tiny IR pulse during each switch. Violent transistor switching occurs continuously at thousands and millions times a second. To reduce the number of violent cycles, power a drive down when not in use. So that both transistor and mechanical life expectancy increases.

westom · 23 Oct 2012

intel 4004 said:

What about forcing it to be off by holding the power button? Will it shortens my hard drive life?

All drives first learn about power off when voltage starts dropping. To a disk drive, power off is same whether from shutdown, by yanking the power cord, or by a nation wide blackout. Even when disk drives moved heads with motor oil, disk drives first learned about and did the necessary shutdown procedures when incoming voltages began dropping.

Drive hardware does not care how power is lost. A drive's computer does same power down procedures starting when voltage drops. Drives are never told, in advance, of a power off. No reason to.

bobafetthotmail · 24 Oct 2012

DeaconFrost said:

HDDs are mechanical parts, and the more wear and tear you put on them, the more you increase their chances of failure.

All machinery is rated for a certain amount of actions, power on/off, hours of working, whatever. That's the time it is designed to work flawlessly regardless of how "tiring" the action itself may be.
For most devices though such numbers are so huge that the device itself becomes outdated well before the parts themselves fail. So while common sense is correct (doing it will actually shorten the life of the product), the actual life of the product was so damn long to start with that really, it makes little difference in the end.

The same general common sense applies with SSDs, people get worried about cells wearing out by read-write cycles, and it's true in theory, but when you start looking at numbers you discover that you need to keep writing 5+ GB of material per day, for years if not decades to really wear it out. So lol, no worries.

He started with "For example, the worst drive I ever saw was rated for 40,000 power cycles.", don't know where he took that number, but is in the general neighborhood of most "rated for" specs I've seen thrown around for hardware.

Britton30 · 24 Oct 2012

I've never seen a rating for power cycles for a drive myself. There are specs for MTBF, Mean Time Before Failure, which is generally 1 million hours. Of course there are those that will fail in 10 minutes and then the ones which become outdated as you say well before it dies.

westom · 24 Oct 2012

Britton30 said:

I've never seen a rating for power cycles for a drive myself.

Early transistors were rated by a number of switching cycles. As that number became large, then datasheets also dropped that specification. A number exists. But nobody cares.

100,000 power cycles was a typical drive life expectancy over 20 years ago. Since then, even that number became so large as to be irrelevant. And still some just know otherwise by ignoring numbers.

All devices are damaged by too many power cycles. The informed consumer never makes a hearsay conclusion from that subjective statement. The informed learn from numbers. Perspective (quantitative facts) is the difference between a subjectively educated victim and the informed consumer. Unfortunately too many recite popular hearsay (subjective reasoning) rather than first learn facts (quantitative reasoning).

Two common 'hearsay generated' conclusions are hardware damage from heat and power cycling. Ten degree temperature differences are also too tiny to affect electronic part life expectancy.

linnemeyerhere · 24 Oct 2012

With the adoption of SSD main drives I say if you're going to be away for more than two hours then shut down and save energy, heat, wear and tear........30 to 45 seconds and you're back in business so why not?

Britton30 · 24 Oct 2012

westom said:

Britton30 said:

I've never seen a rating for power cycles for a drive myself.

Early transistors were rated by a number of switching cycles. As that number became large, then datasheets also dropped that specification. A number exists. But nobody cares.

100,000 power cycles was a typical drive life expectancy over 20 years ago. Since then, even that number became so large as to be irrelevant. And still some just know otherwise by ignoring numbers.

All devices are damaged by too many power cycles. The informed consumer never makes a hearsay conclusion from that subjective statement. The informed learn from numbers. Perspective (quantitative facts) is the difference between a subjectively educated victim and the informed consumer. Unfortunately too many recite popular hearsay (subjective reasoning) rather than first learn facts (quantitative reasoning).

Two common 'hearsay generated' conclusions are hardware damage from heat and power cycling. Ten degree temperature differences are also too tiny to affect electronic part life expectancy.

So why even bring up stuff from 10-15-50 years ago that has no bearing today? I certainly don't care either.

westom · 24 Oct 2012

Britton30 said:

So why even bring up stuff from 10-15-50 years ago that has no bearing today?

Power cycling damage to drives is a number that says why power cycling causes virtually no damage. Your question should be directed at others who make damage claims by ignoring perspective (numbers).

You are expected to learn from history - stuff from 10-15-50 years ago. Junk science exists by ignoring perspective (numbers). Only victims and fools ignore stuff even known 10-15-50 years ago.

If you don't care, then why post? Oh. You do care. How much?

DeaconFrost · 24 Oct 2012

No one said damage. That was injected by you to further your eWang Competition. It can shorten the life of any electronic component. Try it with two identical light bulbs if you don't want to believe in the science and fact that you love to tout.

You're new here, so I'll offer this bit of advice. Coming across with the holier-than-thou mindset, inferring that all others who disagree with you are ignorant and fools....not the best tactic for longevity here.

Save the preachy hypothetical rhetoric for the proper audiences. We're here to help and assist.

bobafetthotmail · 24 Oct 2012

No one said damage.

"damage" doesn't mean failure. Every operation the device does it suffers some minimal damage, that accumulates and eventually causes a failure. (in the case of your light bulb, the filament slowly vapourizes and eventually becomes so thin that burns out)

In case you haven't noticed, he isn't claiming that the device life isn't shortened, but that it was so long that it does not matter.

And yes, it's normal to have components that are supposed to last 50 years. I have spotlights whose switch is rated for 10'000 uses (and they bother to tell me this detail on the box).

Britton30 said:

So why even bring up stuff from 10-15-50 years ago that has no bearing today?

Outdated info is good enough to give a lower limit (as components got better, not worse). That lower limit is still high enough to give an answer to the question of the OP.