Re: XP Defrager

Inline but starts with top posting:

XP Pro, SP3, P4, 1.5 Gig, 4 physical hard drives, two external USB
drives. All NTFS formatted.

Actually, moving ALL the files to the end, beginning or middle of a disk
would have little to no effect on time. The time "hit" of fragmentation
is based on seek times of the head; how long it takes it to move from
one track to another track to another track to ... as opposed to reading
it as sequentially is possible. So, there might be one long seek time
to get from the beginning say, to the end where all the data is, but
after that timing differences would be negligible.
In fact, it's interesting to try to comprehend the amount of data on
an outside track or a drive vs an inner track (one much longer than the
other in physical units). Compared to an inside track, and outside
track can hold more data in one track. Or can it<g>? Then when you add
the separate platters, however many there may be, it gets even more
interesting.

All that said however, anyone who has done much video editing and
rendering knows for a fact that fragmentation can cause enough delays to
cause buffer problems and can seriously extend the amount of time it
takes to complete a render. Most serious video workers defragment their
drives between each editing/rendering session in order to be sure the
render completes before morning. Here, when you're talking about a
rendering time of say about 4 hours, that time can easily become 8 or 10
hours if it's a third or later session without having defragged in
between. Video work is extremely disk intensive. I've often wished I
could afford a 10,000 rpm disk for video work but they are EX-PEN-SIVE!!
It's a lot cheaper to just let it defrag in between. Besides, a defrag
doesn't mean you can't be doing something else with the machine; it
never makes sense to sit and "watch" any long term operation working
IMO.

I would be very appreciative to anyone who can show me a citation from
Microsoft where they indicated that defragging a drive does not improve
timing and performance. I've always known it intuitively and when I
started working with video several years ago I was simply aghast at how
much different fragmentation CAN make. It's pretty simple to set up an
imaginary disk drive on paper and calculate the read or write times
based on the access times of the heads. Accounting for the platters
gets a little confusing I admit, but ... it's not that complex an issue
if you remember any high school math.

JS wrote:
Don't dispute your comments, the only point I
was trying to make was that Microsoft must put
some value as to how much the disk is fragmented
or they would have changed there algorithm in such
a way as to factor out the percentage of fragmentation
from the score.

Actually, they do, somehow. I've seen frag %s in the order of 25% to
30% where sometimes fragmentation is recommended and other times it is
not recommended. I've no idea what the algorithm is, but I noticed this
back inthe days of '98 and it carried forward to XP of course.


It would be interesting to see how much the score
changed if you could (After a clean install) move all
the files to the end of the disk and run there benchmark,
then move the all the file to the beginning of the disk drive
and run the test again.

See top posted information please, at the beginning of this post.


"db" <databaseben at hotmail> wrote in message
news:04FD82C2-CA3A-4EA2-9091-C6F1E522ACED@xxxxxxxxxxxxxxxx
I don't dispute your
score,

however, the score is
a mathematical estimation,
not unlike the percentage
of defragmentation that
windows provides.

the lower the percentage
the higher the score -
mathematically speaking.

however your disk is likely
functioning just as optimal
after the defrag as it was
before the defrag.

I'm sure the efficiency and such of the drive does not change, you're
right. However the time it takes the heads to acquire information does,
partly based on fragmentation.


if anything, I think that having
all the files necessary for
booting together, will improve
boot up time.

I believe that, other than the bootstrap loader, that's why the system
files are always grouped toge ther at the "beginning" of the disk in the
first 1250 sectors or whatever that number is; don't quote me on the
1250. It's also the smallest and shortest of the concentric rings on
the disk, so the drive timing is likely to be neglibibly noticeable in
the overall bootup time.


but that is about all the benefit
to be achieved from defragging.

Interesting comment. So, head read time vs seek times don't come into
play? If I can read 100 sectors in one unit of time, and it takes 5
units of time to get from an early track to one in the middle 50 times,
going from track to track isn't going to add any time to the read
process? I think it is going to. After all, the disks are only
spinning at 7200 rpm, or worse yet, 5400 rpm on the older models.


I think diskkeeper also made
a statement citing that defragging
improves the performance of
anti virus software

IMO it certainly would, depending on the amount of fragmentation and
what was set to be analyzed/read.


what needs to be taken into
consideration is that whatever
the score you achieved after
the defrag will change in a
few hours because the disk
is always in a state of
fragmentation.

Sure. Especially if you want to watch surfing web sites and all the
temp/tmp files they create. But, WHERE fragmentaion occurs is as
important as how many fragments there are. You could have a file in
1,000 fragments, not an unusual number for a drive over half full, but,
if that particular file were never read or written to, and there were
not other similarly fragmented files, the net effect might be near 0.
However, when you do something where the heads do have to go to that
1,000 piece file, it is definitely going to be slower by a measurable
amount and realistically, if other files have similar fragmentation,
it's going to reach a noticeable point fairly quickly.
OTOH, back to your comment, the fragmentation may or may not be of
sufficient magnitude to cause any slowdowns for quite some time. That's
why some people's schedule works out to defrag once every few months to
others at a month, to some like me who at times have to do so on a per
session basis.
If the fragmentation isn't heavy and isn't in the often used files
it's going to have a lot less impact then when it's heavy and is in
often used files. In a way it's a crapshoot and figuring out whether/if
you should fragment is difficult to predict; it depends on what goes on
with the machine.
In addition, there are some defragmenters, one of which I use, will
let you put file at the beginning, middle, end, or last on the platters.
Judicious use of those capabilities can seriously extend the amount of
fragmentation that occurs by leaving lots of room between sets of data.
It's not just that the head has to move that makes a difference, it's
how far the head has to move. Drive specs always give the maximum time
for the heads to move the farthest they could move, and call that that
seek time. So if the fragmentation is building up but the heads only
have to move a few tracks to get to it, that process is much faster than
the specced seek times. As a result it takes much longer for
fragmentation to begin to become obvious in terms of noticeably slower
read and write times. So the % fragmentation for a disk of this
structure can be (and is) a lot higher than say XP's defragger gives.
Then, OTOH, the more fragmentation there is, the longer the defrags take
too, so ... six on one in some ways.


previously, the point I was
trying to make is based on
"time"

whereas one can allocate
a few hours to defrag the disk

But that time, seldom in "hours" on my machine, isn't something that
needs to be watched or managed in any way. One of the best features of
my own defragger is that it will do so during idle times. As a result I
seldom have to actually defrag my disks because they're set to defrag
inthe background during idle times. If defrags are reaching the "hours"
mark, expecially per drive, then it's not being performed often enough.
Just for grins, I just went and defragged drive C, my system drive.
Took it 5 minutes. 5 files were left with single-digit fragments:
Hyberfil.sys and I forget the other ones. Interestingly enough,
pagefile seems to not fragment at all 99% of the time, even though I
have system-managed pagefile settings for my system drive. But then, the
main pagefile is actually on drive E so I guess that might be slightly
skewed, but the system drive is the one that fragments the fastest on
any machine I've ever seen due to the constant registry reads & writes
going on.


or simply accept that defragging
doesn't save any time because
aggregating fragmented files is
normal and by virtue will add a
micro second to aggregate each
fragment into memory.

Please take a look at the seek time for your own hard drive/s. You'll
find it's an awful lot of microseconds PER seek. Hmm, I wonder how
fragmentation will affect solid state drives if/when they ever become
affordable?


my philosophy is that a better
method of managing data is
simply segregating data via disk
partitioning or having multiple
drives.

Well, I don't dispute segregating data types and multiple drives. For
me with video work it's almost a requirement, in fact, to have a drive
dedicated to just rendering, so it doesn't mess up other data. Keeps
the mess all in one place<g>.

Hey db, what branch of the service were you? I was Naval Air Force.
All west coast duty whether shipboard (carriers) or on land. Just
curious, that's all.


Regards,

Twayne




--

db·´¯`·...¸><)))º>
DatabaseBen, Retired Professional
- Systems Analyst
- Database Developer
- Accountancy
- Veteran of the Armed Forces

"share the nirvana" - dbZen

~~~~~~~~~~~~~~~~~~



"JS" <@> wrote in message
news:ONX7wEEpJHA.1340@xxxxxxxxxxxxxxxxxxxxxxx
I installed the Windows 7 beta (Clean Install) and the
performance rating test showed a low number
for my disk drive.

Defragmented the hard drive and the disk performance
score improved by more than 30%. So Microsoft must
test the drive in such a way as to downgrade a drive score
if it fragmented.

--
JS
http://www.pagestart.com


"db" <databaseben at hotmail> wrote in message
news:EBAFD183-9255-4290-AB02-B472C11019A7@xxxxxxxxxxxxxxxx
russinovich disagrees and
feels that defragging does
improve system performance.

and though fat systems may
gain improved performance
from defragging,

I agree with the ms engineers
that on ntfs, it is really a waste
of time.

as we know, the system
partition will never stay
unfragmented.

at best, multiple partitions
on a disk is better at keeping
data segregated and reducing
fragmentation.

--

db·´¯`·...¸><)))º>
DatabaseBen, Retired Professional
- Systems Analyst
- Database Developer
- Accountancy
- Veteran of the Armed Forces

"share the nirvana" - dbZen

~~~~~~~~~~~~~~~~~~



"Tom [Pepper] Willett" <tom@xxxxxxxxxxxxxxxxxxxxxx> wrote in
message news:uNXkxSCpJHA.5228@xxxxxxxxxxxxxxxxxxxxxxx
And if Microsoft engineers have that conclusion, it must be so.
MS is never
wrong.
"db" <databaseben at hotmail> wrote in message
news:D5A6C2E9-3215-4E1A-979A-ABA419BEC224@xxxxxxxxxxxxxxxx
microsoft software engineers
have concluded that defragging
disks that are ntfs is not
necessary.

and after some analysis, I
believe they have a valid
point.

so one can spend much
time defragging the disk.

on the other hand the time
saved by not defragging is
allocated in the time it takes
to aggregate fragmented
files into memory.

but in the end, I think we
people simply like seeing
an organized disk and
defrag it.

also in the end the system
partition never, "never" stays
unfragmented,

"6 of one, half a dozen of
another, as the ol'cliché goes"
-------

in regards to free defrags, I
have also come to learn that
some appear to be defragging
but are not.

ultimately, they are designed
to scan the system, as spywares do.


then provide false statistics/results
eluding that the disk is defrag'd
when it is not.

my suggestion is to simply stick
with the built in windows defrag.

-----------------

db·´¯`·...¸><)))º>
DatabaseBen, Retired Professional
- Systems Analyst
- Database Developer
- Accountancy
- Veteran of the Armed Forces

"share the nirvana" - dbZen

~~~~~~~~~~~~~~~~~~



"Casey" <casey@xxxxxxxxxxxxxxx> wrote in message
news:MPG.24247099859f960898969f@xxxxxxxxxxxxxxxxx

Hi everyone, Running XP-home SP2. Yesterday I read
on the XP Myths site about the design problems with the
build in XP Defrager. So, I am looking for a replacement
defrager that has served others well. Could someone
recommend a (perferably free) defrager.
Thanks you,
Casey



.

Relevant Pages

  • Re: Disk defragmenting
    ... takes to defrag if you're also using the machine at the same time. ... that would hopefully be before the fragmentation becomes bad enough to ... Drives C and D don't fragment much so they get done pretty much when I ... of disk accesses due to fragmentation. ...
    (microsoft.public.windowsxp.general)
  • Re: defrag one TB drives
    ... used internally are in the external drives. ... practices and type of disk usage. ... you will definitely notice the fragmentation; ... You might even have to run defrag multiple times to get a full defrag ...
    (microsoft.public.windowsxp.general)
  • Re: Perfect Disk offline defrag pass is SLOW!!!
    ... How to Gain Access to the System Volume Information Folder ... housekeeping before running Disk Defragmenter. ... Both have 10GB C: drives. ... >>> that might not be enough for the offline defrag to proceed ...
    (microsoft.public.windowsxp.perform_maintain)
  • Re: BSOD - Stop Errors
    ... Volume fragmentation ... Stanley Letter\Flat Stanley Visits Glasgow.doc ... the system to read and write to NTFS file system drives. ... be a physical problem with the disk, ...
    (microsoft.public.windowsxp.general)
  • Re: Defragment
    ... That's the reason Defrag is most effective in Safe ... Disk Defragmenter often says there is no need to Defragment. ... fragmentation that is likely slowing things down, IFF you happen to be using ... shumogy wrote: ...
    (microsoft.public.windowsxp.basics)