Defrag strategy to extend disk life

[AndyHancock]

I looked into defragging utilities in order to minimize the wear on my
hard drive. This differs from most people's main motivation, which
is to improve the speed of data acquisition from the hard drive. For
me, that is a beneficial side effectg.

So far, I've found that freeware doesn't cut it. The short story is
that they don't use the proper security privilege so as to be able to
access and derag all files, regardless of the account to which they
below. I ended up trialing Diskeeper, which seems quite awesome in
that it defrags all the files that the freeware could not. It also
defrags in the background, making use of otherwise lull times in CPU
usage. In addition, the premier pro version supposedly has some
intelligence which figures out which files you access most, and
optimizes the access to them.

Unfortunately, the background defragger seems to go against my reason
for defragging, which is to minimize disk wear rather than speeding up
file access. The background defragging seems to keep the hard drive
perpetually spinning, even in the absence of activity that would cause
defragmentation. So basically, I'm going to forgo that feature and
possibly continue comparing commercial defraggers.

Are there any considerations that I may be overlooking in this
decision?

In the absence of background defragging, would once a week be frequent
enough for defragging? I don't use the machine for video, audio, or
any kind of media. Just plain old surfing and email.

 

[philo]

I looked into defragging utilities in order to minimize the wear on my
hard drive. This differs from most people's main motivation, which
is to improve the speed of data acquisition from the hard drive. For
me, that is a beneficial side effectg.

<snip>

defragging or not defragging
one way or the other is *not* going to affect disk wear in any significant
way.

defragging is useful only in terms of performance.

 

[AndyHancock]

<snip>

defragging or not defragging
one way or the other is *not* going to affect disk wear in any significant
way.

defragging is useful only in terms of performance.

Hello, Philo,

Could you please elaborate on why unfragmented files would not extend
disk life? It seems that noticably more disk churning is needed for
the same amount of data access on a fragmented file. Is there not
more head movement as well?

As well, for non-background defragging, is once a week typically
sufficient?

 

[philo]

Hello, Philo,

Could you please elaborate on why unfragmented files would not extend
disk life? It seems that noticably more disk churning is needed for
the same amount of data access on a fragmented file. Is there not
more head movement as well?

As well, for non-background defragging, is once a week typically
sufficient?

With your machine tunred on...unless the drive is in the sleep mode,
it's always rotating...the armature's movement is not what is going to wear
out a drive.

I don't know if this analogy is any good
but it's like saying that your automobile will fail sooner
if you bring a passenger along with you.

As to non-background defragging..
unless you are really using your machine heavily...I doubt if you need to
defrag
more than once a month. I might defrag my machine 5 or 6 times a year.

 

[Sid Elbow]

With your machine tunred on...unless the drive is in the sleep mode,
it's always rotating...the armature's movement is not what is going to wear
out a drive.

Philo, is there actually evidence for that statement? I don't know first
hand but if I had to guess i would say the constant rotation of the
platter with no starts/stops/speed-changes would involve the least
forces and therefore be the least contributor to wear.

The armature on the other hand *is* accelerating/decelerating/changing
direction with each disk access, so there are constantly changing forces
acting. This is exactly where I would expect the major wear to be. (That
and the platter surface - although I understand that the head rides on
an air gap there so that although any wear would still be a function of
armature movement, it's likely to be minimal).

 

[Bob I]

Philo, is there actually evidence for that statement? I don't know first
hand but if I had to guess i would say the constant rotation of the
platter with no starts/stops/speed-changes would involve the least
forces and therefore be the least contributor to wear.

The armature on the other hand *is* accelerating/decelerating/changing
direction with each disk access, so there are constantly changing forces
acting. This is exactly where I would expect the major wear to be. (That
and the platter surface - although I understand that the head rides on
an air gap there so that although any wear would still be a function of
armature movement, it's likely to be minimal).

FWIW, the repeated defrag activity is more likely to INCREASE the amount
of action the RW head goes through. The additional movement of the heads
to pick up the fragments on a read are nothing compared to activity
caused by a defrag. So one should argue that anal defragging DECREASES
the life expectancy of the drive, NOT increase it.

 

[Sid Elbow]

FWIW, the repeated defrag activity is more likely to INCREASE the amount
of action the RW head goes through. The additional movement of the heads
to pick up the fragments on a read are nothing compared to activity
caused by a defrag. So one should argue that anal defragging DECREASES
the life expectancy of the drive, NOT increase it.

I wouldn't want to be categorical about it in either direction. Whether,
say, a half-hour of intense activity once a week offsets or exceeds the
additional activity due to fragmentation would depend on individual
circumstances I would think. At least, it isn't obvious to me one way or
the other.

 

[Bob I]

I wouldn't want to be categorical about it in either direction. Whether,
say, a half-hour of intense activity once a week offsets or exceeds the
additional activity due to fragmentation would depend on individual
circumstances I would think. At least, it isn't obvious to me one way or
the other.

Actually, the OP appeared to believe that frequent defragging would
somehow extend the drives life expectancy. In the real world, any
difference would not amount to any statistical measurable amount.(in
either direction)

 

[philo]

Philo, is there actually evidence for that statement? I don't know first
hand but if I had to guess i would say the constant rotation of the
platter with no starts/stops/speed-changes would involve the least
forces and therefore be the least contributor to wear.

The armature on the other hand *is* accelerating/decelerating/changing
direction with each disk access, so there are constantly changing forces
acting. This is exactly where I would expect the major wear to be. (That
and the platter surface - although I understand that the head rides on
an air gap there so that although any wear would still be a function of
armature movement, it's likely to be minimal).

I do not have any scientific tests to back me up of course...
but can come up with another analogy perhaps.

If you drive your car from NY to LA and always stayed in the same lane...

vs. taking the same trip but changing lanes often...would you *really* have
driven your car
any significant distance farther or put significanly more wear and tear on
the car's suspension?


I think that if one wrote some software to purposely stress a HD and the
armature
were *continually* oscillated as fast as possible through the greatest range
of movement...
that something like that could cause damage...

but you also need to consider that the very act of defragging a drive will
cause a lot of movement.

I think that such factors as excess heat would play a much larger role in
premature drive failure.


Anyway all this is simply my 2 cents worth
so with inflation and all...may only be a penny's worth!

 

[Dave Onex]

Here's what I do (I weighed out all the same concerns plus a few more)

On most machines I set up an automatic defrag once a week during off hours.
This minimizes impact felt by users and tends to keep the disks in an
efficient layout.

I do find that automated tapes backups are faster when the disk is defragged
as is general machine performance but I'm sensitive to these things.

In any event, I weigh out defrag implementations on a per-machine basis. So
if I had one that had a lot of file creation and erasing going on I would
probably defrag it more.

If you defrag once per week then you're probably cutting down on the amount
of times the head on a drive travels. If you do it on a continual background
basis then you'll have speedy disks but you've dramatically increased the
amount of work that they are doing (as you found out).

Given that all my machines are on RAID 5 arrays and each one has a dedicated
spare (that means I can experience two drive failures before I'm up the
creek) I'm not too worried about how hard the drives work.

My servers are all 'old' in comparison to modern machines and many of my
drives have been in use for 5 years. I buy them (literally) off e-bay for
very little and keep spares. For $50 I can get (5) 15K Seagate's with trays
so.....

But of course it all depends on budget/application etc. At any rate, for
most of my machines a defrag once a week keeps the fragments from getting
out of hand and minimizes the impact to the end users

Best!
Dave

 

[legg]

With your machine tunred on...unless the drive is in the sleep mode,
it's always rotating...the armature's movement is not what is going to wear
out a drive.

I don't know if this analogy is any good
but it's like saying that your automobile will fail sooner
if you bring a passenger along with you.

As to non-background defragging..
unless you are really using your machine heavily...I doubt if you need to
defrag
more than once a month. I might defrag my machine 5 or 6 times a year.

The old Norton Speed Disk had options for files to be 'stored at end
of disk', by preference, and would group files in order of activity or
date of access. The 'most frequently accessed' files were located
adjacent to remaining empty space. It would also place the swap files
for early windows systems in the first sectors.

I think the idea was that seldom accessed files could be located so as
to reduce hardware travel and access time.

Whether this increased disk life or reduced noise, power consumption
or mechanical wear is not obvious.

I think the best strategy for maintaining disk life involves locating
the hardware permanently (no toy trays) where it's not going to get
kicked or dropped or overheat, using good quality connectors, and
replacing your power supply at least every two years.

RL

 

[AndyHancock]

<snip>









With your machine tunred on...unless the drive is in the sleep mode,
it's always rotating...the armature's movement is not what is going to wear
out a drive.

Now that's interesting. That's what that slight continuous whispering
sound is.

But there's more than just the continuous whisper occuring. Whenever
there is disk access, the LED blinks spasmodically and there is
chattering. Probably the head moving. Wouldn't that weigh into the
disk life?

I don't know if this analogy is any good
but it's like saying that your automobile will fail sooner
if you bring a passenger along with you.

As to non-background defragging..
unless you are really using your machine heavily...I doubt if you need to
defrag
more than once a month. I might defrag my machine 5 or 6 times a year.

Cool. I think I'd be too lazy to defrag once a week anyway. Thanks!

 

[AndyHancock]

It seems that way, if we take the chattering of the disk drive to be
some kind of rough proxy indicator of wear on the drive. (That's an
open "if", responses to clarify its inaccuracy or veracity from those
more technically knowledgable are welcome).

Actually, the OP appeared to believe that frequent defragging would
somehow extend the drives life expectancy. In the real world, any
difference would not amount to any statistical measurable amount.(in
either direction)

I was wondering about that possibility for normal defrags, not
constant background defrag that keeps the disk chattering all the
time. The only way to have any confidence beyond speculation is to
conduct trials with a farm of disk drives as the test population (and
with a control group too).

Just based on speculation, defrags streamlines the reads for a little
while, for the cost of the wear due to defragging. Whether the
temporarily streamlined operation more than compensates for the wear
from defrag....who knows. Again, it also depends on whether the
chattering actually takes more out of the disk life than the
continuous spinning (which never stops even in the absence of chatter,
judging from responses in this thread).

As I said in another post, *if* the chattering is taken as the sole
indicator of wear that is occuring (big if), then the constant
background defrag creates much more of it than either non-background
defrag (weekly or biweekly defrags) or just living with unstreamlined
disk reads due to fragmentation.

 

[AndyHancock]

I do not have any scientific tests to back me up of course...
but can come up with another analogy perhaps.

If you drive your car from NY to LA and always stayed in the same lane...

vs. taking the same trip but changing lanes often...would you *really* have
driven your car
any significant distance farther or put significanly more wear and tear on
the car's suspension?

I think that if one wrote some software to purposely stress a HD and the
armature
were *continually* oscillated as fast as possible through the greatest range
of movement...
that something like that could cause damage...

but you also need to consider that the very act of defragging a drive will
cause a lot of movement.

I think that such factors as excess heat would play a much larger role in
premature drive failure.

Anyway all this is simply my 2 cents worth
so with inflation and all...may only be a penny's worth!

LOL! I agree that it's hard to move beyond tallying of considerations
and possible contributing factors without a controlled study. It was
interesting to hear the considerations nevertheless. Thanks!

 

[philo]

Now that's interesting. That's what that slight continuous whispering
sound is.

But there's more than just the continuous whisper occuring. Whenever
there is disk access, the LED blinks spasmodically and there is
chattering. Probably the head moving. Wouldn't that weigh into the
disk life?


Cool. I think I'd be too lazy to defrag once a week anyway. Thanks!

Sure!

That reminds me...
I have not checked my GF's machine
I bet it has not been defragged in 4 months or more...
I better have a look, though it's running just fine.

 

[AndyHancock]

This is a home machine, an 8-year old laptop (approximately). No
RAID. When it dies, it's time for a new machine. It is so old that
it only has USB1, and no native wireless. It takes a card, which
tends to make the fan run almost continuously, thus upping the
probability of another failure mode. Anything that uses up a few
percent of CPU continuously causes the fan to run. It's ironic that
recently boosting the RAM alleviates wear on the disk (since more of
the memory space is in RAM), but it also allows the CPU to run at a
higher percentage and increases wear on the fan.

Anyway, I'm trying to stave off it's demise for as long as possible.
One of its hinges has been replaced, and the hard drive was replaced
around 3-4 years ago. These days, I put it on standby whenever it's
not being used. Hence, scheduled defrags won't work, but it's no big
deal to launch a manual defrag before bed time. Once a week sounds
managable (with some weeks skipped for whatever reason).

About getting drives off e-bay, I'm reticent to using e-bay, as well
as to using second-hand drives. No redundancy in my "system", so I
don't have a chance to replace a drive before losing data.

Thanks for your anecdotes!