Differences between hard disks rated 24 hours/day vs 8 hours/day?

[larrymoencurly]

How do drives with estimated 5-year lifespans differ when they're rated for24 hours of operation per day vs. 8, other than in warranty and price and maybe the ability to synchronize in RAID? Are any components better in thedrives with the 24 hours/day ratings?

 

[Percival P. Cassidy]

How do drives with estimated 5-year lifespans differ when they're rated for 24 hours of operation per day vs. 8, other than in warranty and price and maybe the ability to synchronize in RAID? Are any components better in the drives with the 24 hours/day ratings?

Many years ago I read a Seagate White Paper that claimed that their
higher-spec drives were engineered for multi-drive use to avoid
vibrational feedback -- perhaps other features as well that I've now
forgotten, perhaps cooler running.

They said that their consumer-level drives were intended for
single-drive use only -- i.e., not in multi-drive arrays.

Perce

 

[Rod Speed]

How do drives with estimated 5-year lifespans differ
when they're rated for 24 hours of operation per day
vs. 8, other than in warranty and price

They don't.

and maybe the ability to synchronize in RAID?

That doesn't either.

Are any components better in the drives with the 24 hours/day ratings?

Nope.

 

[Arno]

How do drives with estimated 5-year lifespans differ when they're rated
for 24 hours of operation per day vs. 8, other than in warranty and price
and maybe the ability to synchronize in RAID? Are any components better
in the drives with the 24 hours/day ratings?

Generally, 5 years is not the lifetime, but the "component
lifetime", i.e. the time the MTBF is valid for. If treated
well, basically all drives can be run 24/7 and should die
from old age somewhere between 5 and 20 years. They can
die from otehr causes before, of course.

I have some notebook drives in a fileserver, that run 24/7.
The oldest one is 3.5 years, and still fine, despite a 800'000
head-load count (cought it very late). I do have about
one firmware crash every 1.5 years of drive operation
though.

BTW, "low vibration" is not 24/7, but "RAID drive" or
"array drive", as that can indeed be a problem when drives
are coupled mechanically.

Arno

 

[larrymoencurly]

(e-mail address removed) wrote:

Generally, 5 years is not the lifetime, but the "component
lifetime", i.e. the time the MTBF is valid for.

I noticed that drives made 20-30 years ago had MTBF ratings in the 5 figures, like 50,000 hours, and their estimated lifespans were similar. Then in the 1990s 6-figure MTBFs became common, and I think I even saw some hard drives rated for 2 million hours MTBF, but the estimated lifespans didn't increase. I realize MTBF isn't the same as lifespan, but why did one statistic go up by over a magnitude while the other stayed about the same?

 

[Arno]

I noticed that drives made 20-30 years ago had MTBF ratings in the 5
figures, like 50,000 hours, and their estimated lifespans were similar.
Then in the 1990s 6-figure MTBFs became common, and I think I even saw
some hard drives rated for 2 million hours MTBF, but the estimated
lifespans didn't increase. I realize MTBF isn't the same as lifespan, but
why did one statistic go up by over a magnitude while the other stayed
about the same?

Simple: "Component life" is not a statistic. It is a definition.
It says, "we have investigated the MTBF for this lifetime,
afterwards we do not know".

It also has some statistical meaning, as the MTBF is not
measured for 5 years, but scaled up in various ways. And 5 years
is the time basically all electronics components do not go into
"old-age" and simple accelerated ageing models work.

The thing is though, if the MTBF drops down to 50% for 5...10 years,
it is still pretty good for modern drives.

Arno