How Often Disk Defrag

[VanguardLH]

You've missed the whole point. I AM talking about regular, periodic
maintenance, i.e. defragging, just ONLY when it's needed,

Yet, as shown here, NO ONE knows just when a defrag's cost will result
in a savings (wear, electricity, or by whatever measurement) over
continuing to access fragmented files. Saying that you know when to
defrag is no more steeped in science or proof than someone who admits
they don't know and just goes ahead to schedule the defrag anyway. You
don't know "when it's needed". You just wait until the percentage of
fragmentation gets above some arbitrary threshold upon which you've
chosen (by the way, some defraggers can "schedule" themself to defrag
when fragmentation gets above some threshold, again, arbitrarily chosen
by the user). The other user admits they don't know "when it's needed"
so figure they'll just run it or schedule it at short intervals since
they don't perceive the added wear (which may not exceed the added wear
to access fragmented files) as a risk to the survival of their device.

Your measure is just as arbitrary and not based on valid premises as is
someone else making an arbitrary choice of when to schedule a defrag.

and not some pulled-out-of-the-air schedule (daily, weekly, monthly,
etc.) based on nothing more than a regular, recurring time period.

You're just "pulling out of the air" some fragmentation threshold upon
which you've arbitrarily chosen is when you will defrag. Say it tells
you there is 20% fragmentation. Okay, so how many fragments for each
file to which they belong reside outside the same cylinder? How big or
small are those fragments? How big is the gap between the fragments
(i.e., the size of the unallocated space or allocated space for other
files between the fragments getting moved)?

I tell you, as my employee, that there 10% of a pile of 1000 blocks are
the wrong color and you have to separate them. It's an easy job if
they're the size and weight of 3-gram Lego blocks but a hell of a lot
more work if they're the size and weight of 40-pound construction
blocks. The 10% figure is worthless as to how much work it takes to
segregate the different colored blocks.

Why do you think defraggers have an analyzer tool separate from the
defragging part? It's there so the user can test the drive for the degree
of fragmentation, and decide if they want to defrag.

As I stated, showing the level of current fragmentation does not
indicate how long you get an ROI on wear for defragging those files.
You will add more disk wear with the defrag (several small ones or a few
big ones) but at some point you are hoping that the saved wear from
accessing defragmented files exceeds the wear by the defrag and
accessing fragmented files. Showing the current level of fragmentation
gives you no usable information to know when that ROI point will be met.

I gave up a long time ago on using the percentage of fragmentation or
the number of fragments to give a measure of how "bad" was the state of
my hard disk. 10 fragments that are huge in size (many mega- or
gigabytes) will take a lot longer than 1000 fragments that are 15KB in
size. One time you'll see 10% fragmentation or a count of 50 fragments
and the defrag takes under 10 minutes, or less. Another time you have
5% fragmentation or a count of 10 fragments but the defrag takes an
hour, or more. Depends on how big are those fragments that get moved
around to get rid of the gaps. Obviously moving 10 fragments where the
smallest fragment is 100MB (1GB, or more, total) will involve more
movement than 1000 fragments where the largest was 10KB (10MB, or less,
total). The level of fragmentation doesn't tell you how much you will
be moving.

The level of fragmentation may give a clue as to why a file takes longer
to load (assuming you load the entire file) but it doesn't tell you when
to defragment or how much will get moved (which equates to time to do
the defrag).

I have a data disk where, for example, there are virtual machine files
that are 8GB in size (but could go up to 16GB). Just defragging to get
rid of a single gap (i.e., there are 2 fragments for the huge file)
would involve a LOT of head movement to rearrange the huge chunks just
to get rid of one gap. The defragger might say there is only 1%
fragmentation or 12 fragments over there but that gives absolutely no
clue as to how many bytes are going to get moved which equates into how
long the defrag will take.

The level of fragmentation (percentage or fragment count) really doesn't
provide you a decent gauge to figure out when you should defrag.

You won't be saving on disk wear by doing the major defrag at infrequent
intervals versus doing minor defrags at frequent intervals. Do a lot
all at once but infrequently or a little each time and frequently.
Please explain how there is any difference in disk wear. If left, the
[snip]

The more the drive head moves, the more wear and tear. The more you
defrag, the more the head moves. Even if the defragging takes less time,
because there's less fragmentation, the head still has to move numerous
times to accomplish the defragging whether there are a few fragmented
files or many. It's not like the software "know" innately which files
are fragmented, and only works on those files. It has to read all the
files each time you defrag.

Geez, what defragger do you use? The ones that I use do a scan before
defrag so they can build up a list of eligible files for the type of
defrag operation that you choose to run. If you use the same defrag
algorithm each time, it shouldn't be moving all files but just the ones
that are fragmented (and perhaps have more fragments than a threshold
you configure in the defragger, if an available option). The only
non-fragmented files that should get moved are to make room, say, for
the MFT reserve area or pagefile.

If you keep switching defrag algorithms, like using one defragger for
daily defrags and a different defragger for monthly or yearly defrags,
then you will get a lot of movement. You are using tools whose
author(s) make decisions on what they believe is best defragging. One
defragger will undo "best" placement done by another defragger that
decides a different placement is better. You need to use just ONE
defragger (and why if you use a 3rd party defragger then you need to
disable the one included in Windows).

 

[J. P. Gilliver (John)]

In message <[email protected]>, VanguardLH <[email protected]>
writes:
[]

hour, or more. Depends on how big are those fragments that get moved
around to get rid of the gaps. Obviously moving 10 fragments where the
smallest fragment is 100MB (1GB, or more, total) will involve more
movement than 1000 fragments where the largest was 10KB (10MB, or less,
total). The level of fragmentation doesn't tell you how much you will
be moving.

The level of fragmentation may give a clue as to why a file takes longer
to load (assuming you load the entire file) but it doesn't tell you when
to defragment or how much will get moved (which equates to time to do
the defrag).

[]
At least one of the defraggers I have has the ability to set a threshold
on fragment _size_.

 

[Stan Brown]

[about whether disk defrag is scheduled in Windows by default]

I said in another post that on my computer, Defrag as a service is set
to Manual start and isn't running (i.e., wasn't running when I looked).
I thought it meant it wasn't happening here.

But the above remarks made me look at Task Scheduler, and lo!, Defrag
is in there, scheduled to run at 1:00 AM every Wednesday of every Week.
It last ran at 3/07/2012 19:04:43, according to the entry in the
Scheduler.

I've got 43 tasks, but none of them looks like defragger, unless it
has another name that I'm failing to recognize. I can't be certain,
but I'm pretty sure I never disabled it. If I had, I would have been
unlikely to delete it, but would most likely simply have scheduled it
for far in the future (as I did with the annoying Java Update).

Maybe this is another difference in Home Premium? Do those of you
with defrag enabled by default have Professional or Ultimate?

 

[Zaphod Beeblebrox]

[about whether disk defrag is scheduled in Windows by default]

I said in another post that on my computer, Defrag as a service is set
to Manual start and isn't running (i.e., wasn't running when I looked).
I thought it meant it wasn't happening here.

But the above remarks made me look at Task Scheduler, and lo!, Defrag
is in there, scheduled to run at 1:00 AM every Wednesday of every Week.
It last ran at 3/07/2012 19:04:43, according to the entry in the
Scheduler.

I've got 43 tasks, but none of them looks like defragger, unless it
has another name that I'm failing to recognize. I can't be certain,
but I'm pretty sure I never disabled it. If I had, I would have been
unlikely to delete it, but would most likely simply have scheduled it
for far in the future (as I did with the annoying Java Update).

Maybe this is another difference in Home Premium? Do those of you
with defrag enabled by default have Professional or Ultimate?

Both Professional and Enterprise here, I'll check a Home Premium
tonight (if I can pry it out of my son's hands long enough).

--
Zaphod

"So [Trillian], two heads is what does it for a girl?"
"...Anything else [Zaphod]'s got two of?"
- Arthur Dent

 

[Dave-UK]

[about whether disk defrag is scheduled in Windows by default]

I said in another post that on my computer, Defrag as a service is set
to Manual start and isn't running (i.e., wasn't running when I looked).
I thought it meant it wasn't happening here.

But the above remarks made me look at Task Scheduler, and lo!, Defrag
is in there, scheduled to run at 1:00 AM every Wednesday of every Week.
It last ran at 3/07/2012 19:04:43, according to the entry in the
Scheduler.

I've got 43 tasks, but none of them looks like defragger, unless it
has another name that I'm failing to recognize. I can't be certain,
but I'm pretty sure I never disabled it. If I had, I would have been
unlikely to delete it, but would most likely simply have scheduled it
for far in the future (as I did with the annoying Java Update).

Maybe this is another difference in Home Premium? Do those of you
with defrag enabled by default have Professional or Ultimate?

Mine is here (Win 7 Pro):

Control Panel(icon view) > Administrative tools
Task Scheduler > Task Scheduler Library
Microsoft > Windows > Defrag.

 

[Paul]

You're just "pulling out of the air" some fragmentation threshold upon
which you've arbitrarily chosen is when you will defrag.

My metric is pretty simple.

"Man, this computer is a pig today."

That's when I know it's time. If certain operations have become
unbearably slow, it's time for a cleanup. I did this yesterday
on the machine I'm typing this on, and my log shows the time
before was Oct29,2011. So it's been like a bit over four months
between "spring cleaning" attempts.

*******

And I don't just defrag, I lay the file system down again with
Robocopy. That works on WinXP. I doubt that would be safe with
Windows 7, because of all the tricks Windows 7 has.

And doing a backup and restore, using Windows 7 System Image,
won't "clean" anything, because as far as I know, a VSS based
copy would carefully preserve the fragmentation state of the
computer. Many backup programs will be VSS based, so you'd
suspect they would behave in a similar manner. A program
like the old NTBackup might be different, but again,
my suspicion is, if you brought forward a copy of NTBackup
and used it, it would just trash Windows 7 and the restored
copy would be a mess.

So I don't actually know of a good way (guaranteed) to clean up
Windows 7, like I do with the WinXP machine I'm typing on.

And that makes me "perfectly happy", to use that Windows 7
built-in defragmenter. My scheduled defragmentation on the
Windows 7 laptop, never gets to run on that machine, and
when I check manually, there really isn't enough fragmentation
on there to worry about. The Windows 7 machine was slow from
day #1, so it's not like I'd notice if it got slower. On WinXP,
I can feel that "take out the garbage" effect a lot more.

Paul

 

[Ken Blake]

The more the drive head moves, the more wear and tear. The more you
defrag, the more the head moves. Even if the defragging takes less time,
because there's less fragmentation, the head still has to move numerous
times to accomplish the defragging whether there are a few fragmented
files or many.

If the drive is greatly fragmented, it takes more head movement to
accomplish the defragging, but the *result* of degragging is that the
heads then have to move less. The general view is that it's somewhere
around a wash.

 

[Wolf K]

If the drive is greatly fragmented, it takes more head movement to
accomplish the defragging, but the *result* of degragging is that the
heads then have to move less. The general view is that it's somewhere
around a wash.

You let the machine defrag while you are doing something else. You do do
something else once in a while, no?

;-)

Wolf K.

 

[Ken Blake]

You let the machine defrag while you are doing something else. You do do
something else once in a while, no?

??? I've read your two sentences four or five times now,but I still
have no idea why you think your comment has anything to do with the
point I was making.

 

[Char Jackson]

I've got 43 tasks, but none of them looks like defragger, unless it
has another name that I'm failing to recognize. I can't be certain,
but I'm pretty sure I never disabled it. If I had, I would have been
unlikely to delete it, but would most likely simply have scheduled it
for far in the future (as I did with the annoying Java Update).

Maybe this is another difference in Home Premium? Do those of you
with defrag enabled by default have Professional or Ultimate?

I have a mix of Home Premium and Ultimate. Both types have the defrag
task here.

 

[VanguardLH]

At least one of the defraggers I have has the ability to set a threshold
on fragment _size_.

Which one? Freeware or payware? I recall seeing the option but it's
been too long since I last trialed several defraggers.

That solves the problem of defragging a hard disk that had just 1 file
that had 2 fragments but those fragments were gigabytes in size so you
don't waste a long time and generate lots of head movement to make
contiguous the huge file when such a defrag will definitely have a very
long ROI before any aggregate head movement across that one gap for the
file will exceed the head movement for moving all those bytes. However,
again this is probably an arbitrary value chosen by the user rather than
based on any historical data regarding head movement to defrag those
fragments that exceed in size the configured threshold versus the head
movement needed only for the adjoining clusters of the file to access
them across the gap. So like the argument about using fragmentation
percentage as some measure of when to defrag which is an arbitrary
choice, the fragment size threshold is another arbitrary choice.

I currently use MyDefrag (which is much better than its old jkdefrag
version). However, it still sucks regarding its GUI. It's a bare bones
GUI to Windows' defrag API. To modify how a defrag operates, you have
to go separately editing script files. I see a file size specified in
some scripts but not a fragment size (although that parameter may be
available as yet another directive in the script). However, I'll admit
that I really am not interested in figuring out this author's directive
to figure out how to customize his scripts.

 

[VanguardLH]

If the drive is greatly fragmented, it takes more head movement to
accomplish the defragging, but the *result* of degragging is that the
heads then have to move less. The general view is that it's somewhere
around a wash.

Yeah, as far as the "head/disk wear" argument goes, you could add the
wear a little now with a small defrag, add a bunch all at once with a
big defrag, or don't defrag at all and let the heads bounce around to
find the fragments. The wear argument never seems to prove itself.

However, defragging is more about file load times and shortening them so
YOU don't have to wait as long. Yet most apps just open a handle on a
file and then read/write them in bits rather than load the entire file
in memory. I've hit some programs that decided to dump their entire
signature database into memory so that program's memory footprint was
huge. Defragging was more important when you had much slower hard
disks. The percentage of time to load a file might be the same but the
actual time would be smaller with faster hard disks that users may never
realize there's a difference either in loading a file or in accessing
different clusters for it.

Like you said, when responsiveness to load or access files gets
noticeably slow then do a defrag. However, by the time you really
notice that slow down, it's been happening for a long time. The slow
down will be incremental, not catastrophic. So, in the meantime, you've
incurred a slowdown that lasted until it increased to a threshold large
enough for you to start wondering what to check on why your computer is
slow. Since the head/disk wear issue is a wash, why not do little
defrags at frequent intervals so file load/access time remains as fast
as possible so you never get to the threshold of noticing an extreme
slow down. It takes a hell of a lot of fragmentation to generate a slow
down that you'll notice.

 

[Wolf K]

??? I've read your two sentences four or five times now,but I still
have no idea why you think your comment has anything to do with the
point I was making.

I read your point as implying that waiting for defragging to finish
could take time. Looks like I was read _deeply_ into your text. Well
below the surface. Down where there's no light at all.....



Wolf K.

 

[Ken Blake]

I read your point as implying that waiting for defragging to finish
could take time. Looks like I was read _deeply_ into your text. Well
below the surface. Down where there's no light at all.....

LOL! OK, no big problem.

 

[Ken Blake]

Yeah, as far as the "head/disk wear" argument goes, you could add the
wear a little now with a small defrag, add a bunch all at once with a
big defrag, or don't defrag at all and let the heads bounce around to
find the fragments. The wear argument never seems to prove itself.

Yes, exactly my point.

However, defragging is more about file load times and shortening them so
YOU don't have to wait as long.

Yes, exactly.

 

[Gene Wirchenko]

How often should you use Disk Defrag? I now have mine set for once a
month. Thanks.

Who cares? That is my answer. I have never observed a
difference with before and after although I grant it might happen.
Running defrag, one can feel one is doing something to admin one's
system. Whether anything of use is actually occurring, who knows?

Try measuring access speed before defragging and after. If the
difference is critical enough to you, then you needed to defrag.
Without measuring, for all you know, the defragging might end up
slowing access.

Sincerely,

Gene Wirchenko

 

[Rodney Pont]

Yeah, as far as the "head/disk wear" argument goes, you could add the
wear a little now with a small defrag, add a bunch all at once with a
big defrag, or don't defrag at all and let the heads bounce around to
find the fragments. The wear argument never seems to prove itself.

The head is flying above the disk whether it's seeking or not so I
don't see how a defrag can change the head/disk wear at all. Defraging
causes head movement so would cause wear on the head motor bearings but
so would reading a fragmented file and I can't see it making much
difference either way.

 

[BillW50]

In

The head is flying above the disk whether it's seeking or not so I
don't see how a defrag can change the head/disk wear at all. Defraging
causes head movement so would cause wear on the head motor bearings
but so would reading a fragmented file and I can't see it making much
difference either way.

Really? You read every single file on your drives everyday? Not me, 90+%
of them are rarely ever read. And I can use my hard drive all day and
the drive temperatures stay low. But if I ever defrag the hard drive, it
heats up into the red zone. And you can't see any difference?

 

[J. P. Gilliver (John)]

VanguardLH writes: []

The level of fragmentation may give a clue as to why a file takes
longer to load (assuming you load the entire file) but it doesn't
tell you when to defragment or how much will get moved (which
equates to time to do the defrag).

At least one of the defraggers I have has the ability to set a threshold
on fragment _size_.

Which one? Freeware or payware? I recall seeing the option but it's
been too long since I last trialed several defraggers.[/QUOTE]

Auslogics Disk Defrag. (Version 3.3.1.2.) Free for personal use - I
don't know what business users get charged. The "Skip fragments larger
than" (if you tick it) has a dropdown offering 1 MB, 10 MB, 50 MB, 100
MB, 1GB, 10 GB, with the default I think at 10 MB.
[]

file will exceed the head movement for moving all those bytes. However,
again this is probably an arbitrary value chosen by the user rather than
based on any historical data regarding head movement to defrag those
fragments that exceed in size the configured threshold versus the head
movement needed only for the adjoining clusters of the file to access
them across the gap. So like the argument about using fragmentation

[Are you a lawyer (-:?!? - Your sentences with no punctuation though
strictly grammatical are of extreme length!]
[]

 

[Stefan Patric]

In spite of your remark elsewhere about wear and tear, I suspect that
the weekly defragging is harmless, so I'd let those who feel better
about it just go on doing it.

And if there's no (or very little) fragmentation, pointless, too. Like
running a washing machine with no laundry in it. ;-)

I originally was just answering the OP's question: How often should you
defrag? Answer: When it's needed. Many, who do otherwise, disagreed.
But, as you say, there are those who derive comfort from doing
unnecessary tasks.

Stef