Based on another thread on here, I wondered, out of interest, wether anybody 
else has noticed that a drive thats been running for years fine, say, in a 
horizontal plane, makes a different noise when mounted vertically in a 
different system, and dies quickly?.
Recently, I have replaced bits in some of my servers, that were fine, but 
needed extra storage etc ( I am talking about a small boot drive thats been 
on for say 5 years, 24/7 no problems) and it only lasts a few weeks after 
removal and used for fun somewhere else

I am firmly coming to believe that a drive left running may last for years, 
but changing its orientation upsets the bearings etc?
Is this something anybody else has known?

Probably not explained very well, but basically i dont like moving used 
drives, that have been in the rack, as it seems they will work forever if 
never moved (and some have been hammered VERY hard) but always die soon 
after removal. I would hate to post one etc, as that must be worse. Maybe, 
once they are settled, and "warmed up" they may never die if never moved?
Not sure what the MTBF is measured in?    Life if left running?. Life if 
warmed up, cooled down every day? Life if shipped around by parcel-farce, 
booted round the floor few times, and another computer dropped on it etc?

Just curious.

1/ I agree generally that electronics likes to be turned on and left
on.

2/ moving hard disks always also involves new airflow, sure, shock
kills hard disks, but heat kills them 10x faster than everything else
put together IMHO

3/ HD MTBF does not work like that at all I'm afraid...

Remember this is done at the manufacturers, under ideal power supply
and cooling conditions.

New model HD model number HD001

Take the first 500 off the production line, HD001-00001 to HD001-00500
and put them in the test rig.

power up the test rig.

500 disks x 24 hours in a day = 12,000 disk hours a day

83 days 8 hours 0 minutes 0 second later the first hard disk fails.

12,000 x 83.3r = 1,000,000

Hence MTBF is 1 million hours, even though no single disk had run
longer than 83 days....

1 million hours is 114 years, how else do you think they arrive at
these MTBF numbers?
WD didn't start shipping raptors in 1884 AD.

=================

Can I suggest you have a read of
http://www.logicsmith.com/birth.html

"Guy Fawkes"  wrote in message 
news:4da3cc9c-52a4-4b2d-84bb-5a026ae7ab95@r33g2000yqn.googlegroups.com...
> 1/ I agree generally that electronics likes to be turned on and left
> on.
>
> 2/ moving hard disks always also involves new airflow, sure, shock
> kills hard disks, but heat kills them 10x faster than everything else
> put together IMHO
>
> 3/ HD MTBF does not work like that at all I'm afraid...
>
> Remember this is done at the manufacturers, under ideal power supply
> and cooling conditions.
>
> New model HD model number HD001
>
> Take the first 500 off the production line, HD001-00001 to HD001-00500
> and put them in the test rig.
>
> power up the test rig.
>
> 500 disks x 24 hours in a day = 12,000 disk hours a day
>
> 83 days 8 hours 0 minutes 0 second later the first hard disk fails.
>
> 12,000 x 83.3r = 1,000,000
>
> Hence MTBF is 1 million hours, even though no single disk had run
> longer than 83 days....
>
> 1 million hours is 114 years, how else do you think they arrive at
> these MTBF numbers?
> WD didn't start shipping raptors in 1884 AD.
>
> =================
>
> Can I suggest you have a read of
> http://www.logicsmith.com/birth.html



Good to see somebody else has the same view.
I dont like analogies, but you could say........"3 month old car"

It could have done 10,000 miles, and had the nuts thrashed out of it, or sat 
in a  driveway?
Who knows, but one of the 2 may last a bit longer.
Either way, it may still break down, and as second user, may be hard to sort 
under warranty?

On 30/06/2009 in message <rYu2m.49787$OO7.19789@text.news.virginmedia.com> 
Neil wrote:

>I am firmly coming to believe that a drive left running may last for 
>years, but changing its orientation upsets the bearings etc?
>Is this something anybody else has known?

I remember an article in a computer mag from many years ago that said you 
should partition and format a drive in the orientation it would be used in 
to avoid problems because the head may move slightly when orientation is 
changed.

-- 
Jeff Gaines Damerham Hampshire UK
That's an amazing invention but who would ever want to use one of them?
(President Hayes speaking to Alexander Graham Bell on the invention of the 
telephone)

Jeff Gaines wrote:
> On 30/06/2009 in message 
> <rYu2m.49787$OO7.19789@text.news.virginmedia.com> Neil wrote:
> 
>> I am firmly coming to believe that a drive left running may last for 
>> years, but changing its orientation upsets the bearings etc?
>> Is this something anybody else has known?
> 
> I remember an article in a computer mag from many years ago that said 
> you should partition and format a drive in the orientation it would be 
> used in to avoid problems because the head may move slightly when 
> orientation is changed.

That might have probably applied to a drive using a stepper motor to 
position the head - don't think a drive using a voice coil suffers the 
effect.

Mind you, some of the journalists writting back then would have written 
about the earths magnetic field and the phase of the moon having some 
effect on the Amstrad PC 1512 of the period :-)

-- 
Adrian C

"Adrian C" <email@here.invalid> wrote in message 
news:7b0s6eF21uvlaU1@mid.individual.net...
> Jeff Gaines wrote:
>> On 30/06/2009 in message 
>> <rYu2m.49787$OO7.19789@text.news.virginmedia.com> Neil wrote:
>>
>>> I am firmly coming to believe that a drive left running may last for 
>>> years, but changing its orientation upsets the bearings etc?
>>> Is this something anybody else has known?
>>
>> I remember an article in a computer mag from many years ago that said you 
>> should partition and format a drive in the orientation it would be used 
>> in to avoid problems because the head may move slightly when orientation 
>> is changed.
>
> That might have probably applied to a drive using a stepper motor to 
> position the head - don't think a drive using a voice coil suffers the 
> effect.

Drives used to be approved for use in one position: normally horizontal. If 
you wanted to mount them vertically, then you had to buy a special model.
When were stepper motors last used in hard disks?

Adrian C wrote:

> Jeff Gaines wrote:
>> On 30/06/2009 in message
>> <rYu2m.49787$OO7.19789@text.news.virginmedia.com> Neil wrote:
>> 
>>> I am firmly coming to believe that a drive left running may last for
>>> years, but changing its orientation upsets the bearings etc?
>>> Is this something anybody else has known?
>> 
>> I remember an article in a computer mag from many years ago that said
>> you should partition and format a drive in the orientation it would be
>> used in to avoid problems because the head may move slightly when
>> orientation is changed.
> 
> That might have probably applied to a drive using a stepper motor to
> position the head - don't think a drive using a voice coil suffers the
> effect.

The problem was mainly caused by temperature changes making the platters
expand and contract - as the steppers move the head to fixed positions the
track alignment changes with temperature and if a drive is formatted when
cold it could have data reading errors when used hot, gravity causing the
head position to sag when drive is mounted on it's side and not sag when
mounted flat has the same effect.

However modern drives use voice coils to move the head and there is 1 extra
platter surface that carries no data and is purely used to hold a head
position locating servo track - as all platters generally expand and
contract with temperature at the same rate and all heads sag by same amount
when drive is positioned on it's side this makes the drive always position
the heads spot on the data (if heads position is right by servo track it's
spot on the data as well) so the stepper to data misalignment problem can
no longer happen.

As others have pointed out it's only really high temperatures that kill
modern drives - I'm guessing either the platters can expand enough to cause
a head crash (the head to platter gap reduces with heat until they touch)
or like most magnets the platters lose their magnetic field when heated
until the servo track is damaged and data tracks cannot be found (less of a
problem on data platters as a lowlevel format should restore things on
these) causing the familiar clicking drive as it hunts for position.

On 1 July, 23:09, Nigel Feltham  wrote:

> As others have pointed out it's only really high temperatures that kill
> modern drives - I'm guessing either the platters can expand enough to cause
> a head crash (the head to platter gap reduces with heat until they touch)
> or like most magnets the platters lose their magnetic field when heated
> until the servo track is damaged and data tracks cannot be found (less of a
> problem on data platters as a lowlevel format should restore things on
> these) causing the familiar clicking drive as it hunts for position.

servo tracks are pretty much built into the data surface nowadays,
they have to be to get the areal density and seek times etc.

death by heat is in 99.99% of cases integrated electronics death by
heat

I have NEVER EVER installed a hard drive on any of my own computers
that did not have forced air cooling over both the top and bottom of
each drive, and I have never ever had a heat death.

In fact excepting stupid mistakes (yanking power lead out while spun
up, dropping it, getting oil in it, etc) I have only ever had one hard
disk failure, which was a deathstar, generally I retire them first,
but having said that have some old RAQs here with 20 and 30 gig hard
drives that are still going strong powered up 24/365.

> As others have pointed out it's only really high temperatures that 
> kill
> modern drives
>

I don't think anyone has yet pointed to the Google study 
(http://labs.google.com/papers/disk_failures.pdf) which underlined this 
(and possibly required even more emphasis on the *really* high):

"Failures do not increase when the average temperature increases. In 
fact, there is a clear trend showing that lower temperatures are 
associated with higher failure rates. Only at very high temperatures is 
there a slight reversal of this trend".
-- 
R

On 2 July, 12:32, "neverwas"  wrote:
> > As others have pointed out it's only really high temperatures that
> > kill
> > modern drives
>
> I don't think anyone has yet pointed to the Google study
> (http://labs.google.com/papers/disk_failures.pdf) which underlined this
> (and possibly required even more emphasis on the *really* high):
>
> "Failures do not increase when the average temperature increases. In
> fact, there is a clear trend showing that lower temperatures are
> associated with higher failure rates. Only at very high temperatures is
> there a slight reversal of this trend".
> --
> R

The google test was WAAAAAY too simplistic on the temp front, but then
given a sample of 100k units that is understandable.

e.g. google took SMART data which said "disk 87564 is currently at
44'C"

Which is all very well, but ...

1/ the the temp sensor is no more accurate than any other LM type mobo
temp sensor.

2/ the temp sensor reading assumes the entire physical drive is at the
same temperature.

3/ the temp sensor is not placed in the worst hotspots.

a thermal image of a hard disk shows just as much difference as of a
person outside in winter in warm clothing, in practice an "average"
SMART hard disk temperature reading of say 55 degrees can easily mean
that certain components are running at 90 degrees or more.

when you remember that the point at which electronics "burns up" is
usually around 150 to 175 degrees C, you can see that it is easy to
get that 90 degree hotspot up to burnout point simply by wrapping it
is something insulating, like fluff.

Laptop type HD will get rid of heat better and thus tolerate allegedly
higher temperatures, because the MAXIMUM distance from any point in
the disk to the outside is about 4mm, as opposed to 10mm or more for a
full size disk, thin stuff dumps heat better than blocks.

I'll say it again, I have NEVER had a heat death on one of my own hard
disks on stuff I build for myself, because I ALWAYS have forced air
cooling on both sides of the HD on everything I build.

In article , Nigel
Feltham  writes

>However modern drives use voice coils to move the head and there is 1 extra
>platter surface that carries no data and is purely used to hold a head
>position locating servo track

No. All modern drives use an embedded servo (i.e. the servo's part of
the track.)

>As others have pointed out it's only really high temperatures that kill
>modern drives - I'm guessing either the platters can expand enough to cause
>a head crash (the head to platter gap reduces with heat until they touch)

No.  The head is aerodynamic.  it floats on a cushion of air above the
platter, so it doesn't matter if the platter expands (if it does at all)
Any expansion of the platter is going to be radial.

>or like most magnets the platters lose their magnetic field when heated
>until the servo track is damaged

that would require the platter to reach its Curie point - hundreds of
degrees.

-- 
(\__/)   
(='.'=)  Bunny says Windows 7 is Vi$ta reloaded.
(")_(")  http://imgs.xkcd.com/comics/windows_7.png

Guy Fawkes wrote:
>
> Hence MTBF is 1 million hours, even though no single disk had run
> longer than 83 days....
>
> 1 million hours is 114 years, how else do you think they arrive at
> these MTBF numbers?
> WD didn't start shipping raptors in 1884 AD.

It may be worse than that. Say the manufacturer unearths an academic study 
that shows that disks run at 20 degrees over the optimal level fail 10 times 
quicker. So, take your 500 disks in the test rig, crank up the temperature 
by 20 degrees. After 8.3 days, one dies. But you multiply that by 10 because 
of the study, so you then still end up with an **estimated** mtbf of 1m 
hours. All done with only 100,000 disk hours.