The fragmenting of drives

  • Thread starter Thread starter Buck
  • Start date Start date
B

Buck

What causes a system to slow down more from fragmenting, fragmenting
from many small files, or fragmenting from very big files (like 500mb
or 1gb)?
 
Buck said:
What causes a system to slow down more from fragmenting,

Modern systems dont slow down from any fragmenting.
fragmenting from many small files,

Nope, those dont fragment enough to matter.
or fragmenting from very big files (like 500mb or 1gb)?

While those do fragment, they're normally used in situations
where there is no slowdown. Most obviously with avi files etc
where the fragmentation is completely irrelevant, they get
played at a rate thats determined by the frame rate regardless.
 
Modern systems dont slow down from any fragmenting.

Are you saying today's hard drives don't slow down from fragmenting,
or WindowsXP doesn't slow down from fragmenting?

While those do fragment, they're normally used in situations
where there is no slowdown. Most obviously with avi files etc
where the fragmentation is completely irrelevant, they get
played at a rate thats determined by the frame rate regardless.

Yes that's a good example. If fragmentation is caused by large avi
files, they might still play fine, but can they make your OS sluggish?
 
Are you saying today's hard drives don't slow down from fragmenting,
or WindowsXP doesn't slow down from fragmenting?

Both. Modern hard drives seek so quickly that the head move
from one fragment to another is so fast that its not noticeable,
and any version of Win is doing quite a lot with the hard drive
most of the time, even when say just web browsing and caching
all sorts of stuff from web pages that the heads are moving
around a hell of a lot most of the time that the occasional
extra head move from one fragment to another gets swamped.

And the tiny files in the cache dont get fragmented much either.
Yes that's a good example. If fragmentation is caused by large avi
files, they might still play fine, but can they make your OS sluggish?

Nope, essentially because the head movement from one fragment
to another happens so quickly with modern hard drives that it
doesnt have any effect on what else the OS is doing.
 
Rod Speed said:
Both. Modern hard drives seek so quickly that the head move
from one fragment to another is so fast that its not noticeable,

Bullshit.
Slow down is a function of access time vs sustained transfer rate (STR) .
STR has increased far more than access time has decreased in the past years.
and any version of Win is doing quite a lot with the hard drive
most of the time, even when say just web browsing and caching
all sorts of stuff from web pages that the heads are moving
around a hell of a lot most of the time that the occasional
extra head move from one fragment to another gets swamped.

And the tiny files in the cache dont get fragmented much either.



Nope, essentially because the head movement from one fragment
to another happens so quickly with modern hard drives that it
doesn't have any effect on what else the OS is doing.

As if 15 ms would be noticeably slow but 10ms wouldn't be.
 
Both. Modern hard drives seek so quickly that the head move
from one fragment to another is so fast that its not noticeable,
and any version of Win is doing quite a lot with the hard drive
most of the time, even when say just web browsing and caching
all sorts of stuff from web pages that the heads are moving
around a hell of a lot most of the time that the occasional
extra head move from one fragment to another gets swamped.

And the tiny files in the cache dont get fragmented much either.



Nope, essentially because the head movement from one fragment
to another happens so quickly with modern hard drives that it
doesnt have any effect on what else the OS is doing.


So what could be the problem with my OS? It's a little sluggish than
it used to be, and the last thing I was about to try was defragmenting
it. Since the drive is 120GB I haven't done it yet, I have to leave it
over night. I've run Spybot, Ad-Aware, Registry Mechanic, and RegClean
to try to get it running smoother. I've also run SpinRite to fix any
bad sectors on my drive. And also a Fitness Test Utility from Hitachi
with S.M.A.R.T and everything, it checks out fine.
I don't want to do a clean install of WindowsXP even I know that will
most likely fix any sluggishness. Is WindowsXP still prone to getting
sluggish like Windows98 used to, or is it just me?
When my system is being sluggish I can sometimes hear the hard drive
working hard to do it's job. I can hear this sound which I don't know
how to describe. I don't know if this means the hard drive is slowly
dying or if that's normal. Like I said before, I checked it out with a
couple of programs and the drive is fine, but I know those programs
could be wrong.
 
Rod said:
Both. Modern hard drives seek so quickly that the head move
from one fragment to another is so fast that its not noticeable,
and any version of Win is doing quite a lot with the hard drive
most of the time, even when say just web browsing and caching
all sorts of stuff from web pages that the heads are moving
around a hell of a lot most of the time that the occasional
extra head move from one fragment to another gets swamped.

And the tiny files in the cache dont get fragmented much either.



Nope, essentially because the head movement from one fragment
to another happens so quickly with modern hard drives that it
doesnt have any effect on what else the OS is doing.
That's true if the system isn't heavily loaded. Eventually, systems
will reach load levels at which such things matter (unless you buy more
hardware than is really needed to do the job -- your money, your
choice). It's an OS's (as well as hardware's) responsibility to
schedule I/O to use head movement efficiently.
 
Rod said:
Modern systems dont slow down from any fragmenting.

I beg to differ on that. My laptop becomes very slow do to fragmenting
and it runs XP....well, you did say modern...
 
So what could be the problem with my OS? It's a little sluggish than it used
to be,

That can be something as basic as a virus infection.

Have a look at the task manager, see what the spare cpu capacity
is, sort by cpu use and see what is using most of the cpu resources.
and the last thing I was about to try was defragmenting it.
Since the drive is 120GB I haven't done it yet, I have to
leave it over night. I've run Spybot, Ad-Aware, Registry
Mechanic, and RegClean to try to get it running smoother.

Worth trying the Microsoft free spyware scanner.
I've also run SpinRite to fix any bad sectors on my drive.

Yeah, flakey sectors can do it.
And also a Fitness Test Utility from Hitachi with
S.M.A.R.T and everything, it checks out fine.

OK, its clearly not the drive failing then.
I don't want to do a clean install of WindowsXP even
I know that will most likely fix any sluggishness.

Thats not as bad as it used to be with the
new Files and Settings Transfer Wizard.
Is WindowsXP still prone to getting sluggish like Windows98 used to,

Mine dont.
or is it just me?

Thats possible.
When my system is being sluggish I can sometimes hear the
hard drive working hard to do it's job. I can hear this sound
which I don't know how to describe. I don't know if this
means the hard drive is slowly dying or if that's normal.

Thats a real worry. Check the SMART data with Everest, some
of the others can claim the smart data is fine even when it isnt.
Post the Everest SMART data here.
Like I said before, I checked it out with a couple of programs
and the drive is fine, but I know those programs could be wrong.

Yeah, looks like the drive is dying.
 
CJT said:
Rod Speed wrote
That's true if the system isn't heavily loaded.

Few modern desktop system ever are for long.
Eventually, systems will reach load levels at which such things matter
Bullshit.

(unless you buy more hardware than is really needed to do the job

Virtually every modern desktop system is that today.
-- your money, your choice). It's an OS's (as well as hardware's)
responsibility to schedule I/O to use head movement efficiently.

The modern reality is that few modern desktop systems use
the hard drive agressively enough for that to be relevant.

Servers are a different matter entirely.
 
Rod said:
Few modern desktop system ever are for long.


Virtually every modern desktop system is that today.


The modern reality is that few modern desktop systems use
the hard drive agressively enough for that to be relevant.

For that matter, hard disk performance is irrelevant to most modern
desktop systems, because the disk is barely used. In fact, desktop
systems probably shouldn't even have hard disks.
Servers are a different matter entirely.
I didn't notice anything in the original post that excluded servers.
 
CJT said:
Rod Speed wrote
For that matter, hard disk performance is irrelevant to most modern desktop
systems, because the disk is barely used.

Indeed, and thats why fragmentation is irrelevant.
In fact, desktop systems probably shouldn't even have hard disks.

Thats crap for most personal systems.
I didn't notice anything in the original post that excluded servers.

He clearly wasnt talking about a server.
 
Are you saying today's hard drives don't slow down from fragmenting,
or WindowsXP doesn't slow down from fragmenting?

That is untrue. Modern Filesystems fragment far less than older ones,
but e.g. FAT32 is an old filesystem. And they will slow down on reading
and writing, just a bit later since there is write buffering and read
caching. Reading a fragmented file for the first time is still massively
slower as reading an unfragmented file. Writing a file to fragmented
space witll slow down massively when the write buffer is filled or a
flush condition is triggered.

Of course HDDs have gotten faster, so the impact of fragmentation is
not so obvious as it used to be.

Arno
 
That is untrue.
Nope.

Modern Filesystems fragment far less than
older ones, but e.g. FAT32 is an old filesystem.

Nope, and the use of free space has changed over time too.
And they will slow down on reading and writing, just a
bit later since there is write buffering and read caching.

Wrong again.
Reading a fragmented file for the first time is still
massively slower as reading an unfragmented file.

Wrong again. Few could even pick it in a proper randomised
double blind trial without being allowed to use a ute that
displays the fragmentation status, so there is no 'massively'
Writing a file to fragmented space witll slow down massively
when the write buffer is filled or a flush condition is triggered.

Wrong again. Few could even pick it in a proper randomised
double blind trial without being allowed to use a ute that
displays the fragmentation status, so there is no 'massively'
Of course HDDs have gotten faster, so the impact
of fragmentation is not so obvious as it used to be.

And there is no 'massively', stupid.
 
Wrong again. Few could even pick it in a proper randomised
double blind trial without being allowed to use a ute that
displays the fragmentation status, so there is no 'massively'

Take two drive, with the same data. One massively fragmented (all
files spread all over the drive), and one organised in a nice way,
with all the files in contiguous sectors.
Copy one of the two drive on an empty one. Copy the other one on an
empty one.
Even you could pick the fragmented one in a double blind trial.

Nick
 
Arno Wagner said:
That is untrue.

Can a question be untrue.
Modern Filesystems fragment far less than older ones,
but e.g. FAT32 is an old filesystem. And they will slow down on reading
and writing, just a bit later since there is write buffering and read
caching. Reading a fragmented file for the first time is still massively
slower as reading an unfragmented file. Writing a file to fragmented
space witll slow down massively when the write buffer is filled or a
flush condition is triggered.
Of course HDDs have gotten faster, so the impact of fragmentation is
not so obvious as it used to be.

That's nonsense. Just because they have become so much faster is why
seeking still impacts on that speed dramatically. You said that yourself.
 
Back
Top