Are disk bearings really harmed by spin-up?

[Arno Wagner]

I have seen various comments like "we leave our servers running 24/7
because powering up a hard drive causes more wear than leaving it
running." I think it's mostly laziness and apathy about conserving
energy. The TiVo forums discuss it a lot because a standard TiVo HD
runs all the time, buffering 30 minutes of whatever channel it's left
on.

Do IT people who leave servers running 24/7 ever have much choice of
NOT leaving them on 24/7? If not, how can they make scientific
comparisons of drive-bearing life? As long as the head isn't moving,
bearing life seems to be the main concern. On a home PC left on all day
it's far less likely that the drive will be doing anything but spinning
at high RPM for no real reason.

I've heard similar claims that the "shock" of turning on a light bulb
is worse than leaving it on all the time. Usually those comments came
as a way to excuse energy consumption after a debate on the merits of
waste. In reality, bulbs have a finite hours rating and will burn out
faster the longer they are left on, as long as they aren't flipped on
and off as torture. CFL bulbs (w/ballast) don't like to be switched on
and off quickly, but I can't imagine them burning out faster if you
only cycle on/off once in 10 minutes or so.

Would anyone claim that car wheel bearings get as much wear when you
pull out of the driveway vs. a 500 mile nonstop trip? In that case, the
"spin up" would be when you first move the car after sitting. What
exactly causes the "big shock" when a hard drive spins up? The heat
generated from constant spinning would seem to far outweigh it. Why
does Windows have a "Turn off hard disks" feature in Power options if
not to reduce bearing wear?

If anyone has thorough technical articles on hard drive wear, please
post. Specifically, what is so torturous about spinning up the drive,
and how can that brief cycle be quantified, damage-wise against
constant spinning with higher heat levels?

Heat is a primary HDD killer. Still, spin-up induces more stress than
normal operation, and many desktop HDDs are only rated for 50.000
start/stop cycles. That is plenty for once a day. That is far to
little for frequent spin-down. Since frequent spin-down is done in
notebooks, notebook HDDs are usually rated for 500.000 start/stop
cycles or more.

My personal guess is that this 24/7 argument is an urban myth. I
operate numerous HDDs and I have not seen significant differences in
reliability in those that run 24/7 and those that were in computers
started and shut dowen once a day. And clearly you can engineer disks
for more start-stop cycles, but there is no need. As long as you
respect the lifetime-limit on start/stop cycles (usually spread out
over 5 years) you should not induce significanlty higher failure
rates. If you of course operate a desktop HDD in an environment were it
starts/stops 1000 times a day, you might just loose it to wear and
tear in some months. If you operate a disk 24/7 without good cooling
at, say, >70C, you might see the same fast death. In such circumstances
_not_ running the disk permanently might actually extend its life
significantly.

There is one component that is put under very high stress at start-up,
especially in a server with many disks: The Power Supply Unit. PSUs
regularly fail on start-up and far less often during normal
operation. That is one of several reasons why servers are usually
running 24/7, even if they are not needed all the time. But take note
that in Servers HDDs are usually cooled well.

Arno

 

[Jack Tyler]

Completely different to hard drives.

In principle it's the same, just with a lot more load on those wheel
bearings. They take a lot of stress when you drive so I question that
pulling out of the garage is significantly worse than a long trip on a
hot day. Cold engine starts are different because oil has to be pulled
up farther from the crankcase, though there's usually film left on the
cylinder walls and crank bearings. Slick 50's claims of metal on metal
are overhyped. But this is getting off topic.

To reduce power used, just like with monitors and motherboards.

Other posters have claimed that power usage is nil but clearly it's
not. My main angle on this was about saving energy. Multiply 5 watts by
millions of computers and you've saved a lot of power. Do that with
countless other gadgets and you've saved a lot more. People leave work
monitors on all weekend (with goldfish tank displays) when they're
gone. Too much trouble to push the off button or let the screen blank
out? Diesel owners idle their engines way too much, etc..

Anything that doesn't have to run 24/7 could be turned off instead of
blaming it all on Kenneth Lay and the Arabs.

JT

 

[Randy S.]

http://www.pcguide.com/ref/hdd/perf/qual/specCycles-c.html

Well, I'm reading that start/stop cycles are typically in the 30k to
50k range, and that's something I didn't know. I wasn't looking for the
right keywords. One boot per day on a home PC would allow for 109 years
at 40k cycles, which means other components must wear out faster than
bearings! This is the info I needed.

JT

Yes, one boot per day would be negligable, and wouldn't really be a
factor. But power saving modes could increase that a lot. If your HDD
powered down after 10 minutes of idle time (not unrealistic for a laptop
trying to save battery), it may cycle 40-50 times per day easily. So if
you divide 109 by 50, that's only a little over 2 years! Of course this
is an extreme case, and, as was noted in another post, notebook HDD's
are often engineered for more start/stop cycles. The thing to note is
that it *is* a design criteria. The other thing to hope is that as
desktop PC's become more power conserving, the HDD's had better be
designed for higher start/stop cycles.

Randy S.

 

[Randy S.]

Heat is a primary HDD killer. Still, spin-up induces more stress than

normal operation, and many desktop HDDs are only rated for 50.000
start/stop cycles. That is plenty for once a day. That is far to
little for frequent spin-down. Since frequent spin-down is done in
notebooks, notebook HDDs are usually rated for 500.000 start/stop
cycles or more.

I hadn't seen those specs, but it sure as hell makes sense.

If you operate a disk 24/7 without good cooling
at, say, >70C, you might see the same fast death. In such circumstances
_not_ running the disk permanently might actually extend its life
significantly.

Well, yes, if you are running a piece of equipment in out-of-spec
conditions, I would expect minimizing the run time would extend its life
;-).

There is one component that is put under very high stress at start-up,
especially in a server with many disks: The Power Supply Unit. PSUs
regularly fail on start-up and far less often during normal
operation. That is one of several reasons why servers are usually
running 24/7, even if they are not needed all the time. But take note
that in Servers HDDs are usually cooled well.

Also, well designed servers will do a "staggered" start, i.e. they won't
start all of the hard drives at once, but at intervals, like 1 every 2
seconds. HDD startup takes a *lot* of juice, and if you have 5 or 6 or
more HDD's in a system you either have to *way* oversize your power
supply to handle the startup load, or more cheaply, stagger the start so
a smaller power supply can handle it. A staggered start can slow bootup
a *lot*, so it really discourages cold booting a server.

And a good admin will make sure that his drives aren't operating
over-temp, as you say.

Randy S.

 

[zaw]

I understand lubrication phenomena but I'm trying to find a way to
quantify startup wear vs. constant running wear. Maybe it's just hard
to quantify without a lot of guesswork.

Actually Hard drives are designed to run constant for years and years.
I have this PC on for over 3 years Only time It get shut down is when
I have to move it or for system upgrade. Only time I have something go
wrong hardware wise is when the piece shut down, like fan on my CPU died
after I shut it down to add more memory. Then fan on my Nvida card
start to make noise and stop because of dust build up.

HDD are better when they're running and they're mounted Horizontally.
That's where everything is perfect. Bearing on hdd work the best after
they're spinning, everything is designed to work around their normal
temperature.

Also other thing is.. Heads in HDD are gliding with wind that created by
the spinning disk, no physical contact with surface anytime when the
drive is running. When the drive get spin down, no winds so the Heads
need to goto Landing zone and they'll rest on the surface, usually made
with different materials and coating to prevent scratches and drag.
When the drive spin up heads must wait until enough wind to glide and
get to the data area, that mean is they're rubbing on the landing zone
until it reach full speed. The coating get ware off after too many
time, will cause heads dig into media, media will start to flake, get
tossed around by spinning discs and cause data corruptions. Now days
with high capacity drives doesn't take much to destroy data.

I just leave mine on for reliability.

 

[Randy S.]

Rod Speed wrote:




In principle it's the same, just with a lot more load on those wheel
bearings. They take a lot of stress when you drive so I question that
pulling out of the garage is significantly worse than a long trip on a
hot day.

Maybe, though the load profiles are completely different as are the
design conditions. A hard drive operates with very little load, it only
needs to overcome frictional losses in the bearings and air resistance,
and maybe support a tiny bit of weight depending on the orientation of
the drive. A wheel bearing is supporting 1/4 the weight of your car at
all times (well, I suppose it varies, but it should *average* around
1/4), the load support is primary, overcoming frictional losses is a
distant second. Startup wear will still be greater than operating wear,
but I agree that it's doubtful that it's significant in that
application. But that doesn't mean it isn't significant in the HDD
application.

Cold engine starts are different because oil has to be pulled
up farther from the crankcase, though there's usually film left on the
cylinder walls and crank bearings. Slick 50's claims of metal on metal
are overhyped. But this is getting off topic.

So, as you note, different applications show varied significance of
startup wear. There's so little wear at operation on HDD's that startup
wear is bound to be significant. The film at startup on bearings
doesn't do you any good if there isn't enough force to form a pressure
wedge, however I throughly agree that products like Slick 50 overhype
such issues to sell stuff. Modern materials handle startup wear *much*
better than 30 years ago. Shoot, we have cars that need not much more
than an oil change for 100,000 miles now.

Other posters have claimed that power usage is nil but clearly it's
not. My main angle on this was about saving energy. Multiply 5 watts by
millions of computers and you've saved a lot of power. Do that with
countless other gadgets and you've saved a lot more. People leave work
monitors on all weekend (with goldfish tank displays) when they're
gone. Too much trouble to push the off button or let the screen blank
out? Diesel owners idle their engines way too much, etc..

Anything that doesn't have to run 24/7 could be turned off instead of
blaming it all on Kenneth Lay and the Arabs.

I think I mentioned that in a parenthetical comment before. While
anyone drive uses very little power, it can certainly argued that in
*aggregate* we could save quite a bit of power if it was conserved. Is
it cost effective? I can't answer that. I think we're probably headed
towards home systems where data is stored centrally and other units act
as smart (i.e. "thick") clients (think of storing all your
movies/music/pictures on a central device, then viewing them from in
your office, on your tv, in your kitchen, etc). I would say in that
situation, the data storage device would need to be on all the time, but
the other devices could aggessively power down when idle.

Randy S.

 

[Jeff Rife]

Randy S. ([email protected]) wrote in alt.video.ptv.tivo:

The other thing to hope is that as
desktop PC's become more power conserving, the HDD's had better be
designed for higher start/stop cycles.

When the CPU draws 50-100W and a hard drive draws less than 10W, I don't
think powering down the drive really amounts to much of a savings. It was
less than two years ago that the typical hard drive sucked more power than
the CPU, but it has changed so radically now with 2GHz+ CPUs and much more
efficient hard drives that it just isn't a big deal.

I wouldn't be surprised to see sub-5W (at idle, but spinning) 3-1/2" hard
drives in the next two years, while CPU power requirements continue to
skyrocket.

 

[Rod Speed]

Rod Speed wrote

I think I need to find and read those datasheets.
If you have any quick links, please post.

http://www.hgst.com/hdd/support/table3.htm
Generally in the Technical Library documentation link for each drive.
One specific example is
http://www.hitachigst.com/tech/tech...F3819BDB86256CE9005AB0B9/$file/d7k250P_sp.pdf

 

[Rod Speed]

Rod Speed wrote:

Well, I'm reading that start/stop cycles > are typically
in the 30k to 50k range, and that's something I didn't
know. I wasn't looking for the right keywords.

Yeah, thats the main problem.

One boot per day on a home PC would
allow for 109 years at 40k cycles,

But only a couple of years if you set it to power down on 30
mins of inactivity and it gets used at something like that rate,
say to poll for new email and dont turn it off overnight.

which means other components must wear out faster than bearings!

Yes, bearing failure isnt seen much anymore with desktop hard drives.

Tho we havent been using fluid bearings
for all that long, so that may change too.

 

[Rod Speed]

Rod Speed wrote

In principle it's the same,

Nope, nothing like it. The bearings are completely different,
you dont get anything like the spinup torque you get on a
hard drive platter with a car wheel, and the hard drive has
a single bearing on one end of the axle too.

just with a lot more load on those wheel bearings.

Much bigger bearings.

They take a lot of stress when you drive

The main stress on the bearings with car wheels
is when you drop into a pothole at speed etc.

so I question that pulling out of the garage is
significantly worse than a long trip on a hot day.
Nope.

Cold engine starts are different because oil has to be pulled
up farther from the crankcase, though there's usually film left
on the cylinder walls and crank bearings. Slick 50's claims
of metal on metal are overhyped. But this is getting off topic.

And completely different to a hard drive bearing anyway.

Other posters have claimed that power usage is nil but clearly it's not.

Its in all the datasheets. 5W is pretty typical for a modern IDE drive.

My main angle on this was about saving energy. Multiply 5 watts
by millions of computers and you've saved a lot of power.

Still a fart in the bath in total power consumption.

Do that with countless other gadgets and you've saved a lot more.

Still a fart in the bath in total power consumption.

People leave work monitors on all weekend (with
goldfish tank displays) when they're gone. Too much
trouble to push the off button or let the screen blank out?

Hardly the end of civilisation as we know it any time soon.

Diesel owners idle their engines way too much, etc..

Anything that doesn't have to run 24/7 could be turned off

No thanks, I turn hardly anything off.

instead of blaming it all on Kenneth Lay and the Arabs.

Wouldnt fix the problem even if they did
turn everything off that didnt need to be on.

 

[Randy S.]

When the CPU draws 50-100W and a hard drive draws less than 10W, I don't
think powering down the drive really amounts to much of a savings. It was
less than two years ago that the typical hard drive sucked more power than
the CPU, but it has changed so radically now with 2GHz+ CPUs and much more
efficient hard drives that it just isn't a big deal.

I wouldn't be surprised to see sub-5W (at idle, but spinning) 3-1/2" hard
drives in the next two years, while CPU power requirements continue to
skyrocket.

True, but the CPU's are getting more power-aware as well, and can run in
power conservation modes that use less power at the cost of speed. once
mobile dual core chips are out, they could even shut down one processor
completely. So at low power modes, HDD power could still be significant.

Randy S.

 

[Folkert Rienstra]

Lots of hearsay in past discussions. This thread has been a lot more
informative.

Strange how more hearsay suddenly turns into information as soon as the
question is asked by poster as an originator rather than him being a lurker.

 

[Folkert Rienstra]

This is the conventional wisdom. And it's not just disks. Thermal cycling
used to be a serious problem with computers--that's why memory sockets have
latches now. On an original IBM PC that had been running for a couple of
years, sometimes the memory chips would walk completely out of the socket
due to repeated thermal cycling.


Shutting down a large server farm is not something to be done lightly.
Bringing it down and back up in an orderly fashion might take more than
one night.


Yep, and they seem to last and last.


Depends on the circumstances.


The viewpoint is generally based on ex-
perience with other mechanical devices.


Even if the head is moving, bearing life is the main concern as far as
_wear_ goes.

The heads run on an air bearing--the wear is negligible.

However disk seldom die of bearing failure--generally the failure

is a crash

Which basically is a failure of 'air bearing'.
Sounds like "the wear is negligible" may not be so 'negligible' as expected.

or an electronics failure.


Maybe on _your_ system.


You ever notice how light bulbs generally blow when you turn them on, not
when they are just sitting there giving off light? It's called "thermal
shock" and it's a real phenomenon.


The basic problem with any bearing is that at rest the mass supported by
the bearing causes the rotating assembly to sink though the lubricant until
it is touching something solid. When the device of whatever kind is started,
there is a period before the lubricating film reestablishes itself in which
there is metal-to-metal contact. Thus most of the wear occurs at startup.
This is exacerbated by the fact that the lubricant is cold and thus does
not flow well.

The "heat generated from constant spinning", assuming that the drive is not
being operated outside its rated temperature range, has negligible effect
on the durability of the mechanical components--

it would have more effect on the electronics but
the electronic components are outside the capsule.

Except one very crucial pre-amp.
And the rest is bolted on close to the HDA with negligible to no airflow
between them.

 

[Richard Lowen]

....My main angle on this was about saving energy. Multiply 5 watts by
millions of computers and you've saved a lot of power. Do that with
countless other gadgets and you've saved a lot more. People leave work
monitors on all weekend (with goldfish tank displays) when they're
gone. Too much trouble to push the off button or let the screen blank
out? Diesel owners idle their engines way too much, etc..

Anything that doesn't have to run 24/7 could be turned off instead of
blaming it all on Kenneth Lay and the Arabs.

What you and almost everyone else ignores is the energy used to
mine the raw materials and the energy to manufacure the components
and the energy to assemble those components into products. Add to
that the energy to get the workers to the manufacturing site, etc., and
it could very well be that more energy is used to make a product than
to operate it. So what you could have is a big energy WASTE if you
don't prolong the life of your products by minimizing the number of
power on/off cycles. In short, it's the overall cost in energy that counts,
not just the energy of operation.


Rick Lowen

 

[Richard Lowen]

HDD are better when they're running and they're mounted Horizontally.
That's where everything is perfect.

Reps at various HDD manufacturers say that orientation doesn't matter.


Rick Lowen

 

[Peter]

Reps at various HDD manufacturers say that orientation doesn't matter.

Actually that is not true. They restrict hard drive position to
a few orientations.

 

[Eric Gisin]

What you and almost everyone else ignores is the energy used to
mine the raw materials and the energy to manufacure the components
and the energy to assemble those components into products. Add to
that the energy to get the workers to the manufacturing site, etc., and
it could very well be that more energy is used to make a product than
to operate it. So what you could have is a big energy WASTE if you
don't prolong the life of your products by minimizing the number of
power on/off cycles. In short, it's the overall cost in energy that counts,
not just the energy of operation.

Nonsense.

The energy to run a 100W computer for 5 years is 5*365*24*0.1 = 4380 kWh, or
$200-400. That exceeds the cost of entry level computers. Energy to
manufacture it is only a portion of that.

 

[Joe Smith]

Reps at various HDD manufacturers say that orientation doesn't matter.

I've heard that it is OK to run disks horizontal, vertical, or upside
down, but not diagonally. An older Dell PC I worked on yesterday has
one disk vertical at the front of the tower and the other disk upside
down under the floppy drive. Any orthogonal orientation is OK.
-Joe

 

[Richard Lowen]

Nonsense.

The energy to run a 100W computer for 5 years is 5*365*24*0.1 =
4380 kWh, or $200-400. That exceeds the cost of entry level computers.
Energy to manufacture it is only a portion of that.

That's assuming you would run the computer 24/7. But if the question
is whether to turn it off every 1/4 hr for 1/4 of an hour AND turn it off for
12 hours (as most people do), the energy savings is only $25-$50, and
the life of the computer might be cut in half by the frequent power cycling.


Rick Lowen

 

[Arno Wagner]

Also, well designed servers will do a "staggered" start, i.e. they won't
start all of the hard drives at once, but at intervals, like 1 every 2
seconds. HDD startup takes a *lot* of juice, and if you have 5 or 6 or
more HDD's in a system you either have to *way* oversize your power
supply to handle the startup load, or more cheaply, stagger the start so
a smaller power supply can handle it. A staggered start can slow bootup
a *lot*, so it really discourages cold booting a server.

Indeed. That is possibly an other factor.

Personally in my PC servers I go for PSUs that can take the full
spin-up load of all disks at once. That has the advantage that during
normal operation the PSU is not under higher load, which should
increase its lifetime. The limit for this is somewhere around 10-15
disks, since PSUs with more than 550W are dificult to get.

Arno