Is Microsoft DEFRAG the best defragger?

Just to add my 2 cents to this aging thread:
http://groups.google.com/groups?hl=f...gbl%26rnum%3D1
....

Many people out there are saying that today's system are so fast that
there should be no need to prolong hard disk defrag process so that it
would be better optimized for program startup.

I agree on that specific point: program startup is short ennough now
(for systems built on recent hardware) to avoid spending extra time
defragging a drive in a fancy but more time consuming way. (excuse my
lack of vocabulary - english is not my primary language!

BUT...

I recall using Norton Utilities (DOS) defragger back in time where I
had a 486 DX33 and a 220MB hard drive (or so). I found it would be
funny to optimize loading speed of Win 3.1, because I had so many
fonts that it took lengthy additional minutes for this OS to load
(fonts where loaded in full into memory, I think) - anoying. So I took
advantage of Norton's SpeedDisk options which made possible to
organize physical file placement on the drive (You know: place all
'*.tmp' at the end, other type at beginning, and so on...).

I forced positionning of all *.ttf (true type fonts), *.fon, *.fot (or
something like that) as first data on the drive, in a sequential,
alphabetical order - so they where gathered together as a "solid"
suite of font datas. I then assured myself that the loading sequence
of those fonts was logically the same in Windows' interest - meaning I
edited the win.ini (or whichever ".ini" that apply but I don't recall
it!) to gather and then sort font specific lines in the same
alphabetical order. Saved the '.ini', and rebooted...

Day VS night!

I could HEAR it when Windows was getting at this portion of boot
sequence: the hard drive was suddenly transfering datas to memory in
silence, as a the head was not seaking everywhere! Simply reading
track-to-track information as if it was a big, single, defragemented
file! Running smooth as a rolling tape, but fast as hell (as
absolutely NO seek time was involved)! But there was HUNDREDS of font
files... so what would have been the benifit of a simple file based
data defragging? Shorter task time, yes, but this kind of
accelleration would have been impossible, and magnetic heads of the
drive would have been constantly moving, stressed as usual.

SO now... imagine a complete Win98 (or why not XP) defrag process
which would analyse and optimize as many file as possible to get them
aligned on the drive almost exactly as they will be call each time
Windows boot up: wouldn't it be great? =D Benefit in time would
largely overcome this time to time longer defrag procedure (hey! My
'defragmented' XP system is so big it now takes me a full 3m30s to
boot, the drive bathed in seaking noise! A shame!).

I'm pretty sure a good analysis of system bootup requests could be
done automatically before defrag, so miroring this structure to file
position on drive could be done for most of them. Leave sufficient
head space between those 'parented data blocks', and you leave room
for subsequent FAST defraging even if some files have been added or
changed... until those empty spaces are filled and complete block
rebuild should be required (LONGER defrag time then). Not every file
should be moved that way, but MANY of them could (those which are only
'loaded', rarely modified).

This is MY opinion: WINDOWS OS startup IS THE MOST ANOYING PART OF
TODAY'S COMPUTING EXPERIENCE. True, 'programs' loads fast ennough
now... BUT a 3m30s wait BEFORE having a chance to finally launch one
of those applications is way too much!

I will get my two SATA Raptors at the end of this week.

Part of my decision to get them IS this unacceptably poor performance
on bootup (with my Western Digital WD800JB 7200rpm, 80GB). I may have
delayed my decision if my main drive would not have gave me serious
failure signals
recently. But I was to wait a better price drop and get them anyway -
I am fed up with those lenghty transfer times!

An ingenious defrager could have made me postpone this buy after a
better price drop, or even after competition's response (will come!).


Feedback on this?

While your reasoning makes plenty of sense in re Win 3.11 and any Win9x system
(including ME), a major portion of the time spent loading XP is due to
housekeeping tasks. No amount of analysis and defragmentation is going to
seriously affect those processes. Even where it might, the files involved would
change during every session and thus you would not only have to defrag before
every restart to see much advantage, the defragmentation logic created during
the previous startup would be obsolete by that time.

With XP, your time is better spent reviewing background processes, services,
etc., to make sure that you are only running things that you really need to run.
Even then, XP is going to take longer to load than any comparably configured 9x
system--it's the nature of the beast. Of course, XP is much better equipped to
leave on all the time, for days or even weeks on end, using appropriate
power-saving settings.

Consider, also--newer systems are by nature larger and more complicated systems.
Compare the startup time of a well-equipped, full-service 98 system in Sept. of
1998 to a comparably well-equipped, full-service system in Sept. of 2004 and
you'll get similar results.

--
Gary S. Terhune
MS MVP for Win9x

"VROMB" wrote in message
om...
Just to add my 2 cents to this aging thread:

http://groups.google.com/groups?hl=f...gbl%26rnum%3D1
...

Many people out there are saying that today's system are so fast that
there should be no need to prolong hard disk defrag process so that it
would be better optimized for program startup.

I agree on that specific point: program startup is short ennough now
(for systems built on recent hardware) to avoid spending extra time
defragging a drive in a fancy but more time consuming way. (excuse my
lack of vocabulary - english is not my primary language!

BUT...

I recall using Norton Utilities (DOS) defragger back in time where I
had a 486 DX33 and a 220MB hard drive (or so). I found it would be
funny to optimize loading speed of Win 3.1, because I had so many
fonts that it took lengthy additional minutes for this OS to load
(fonts where loaded in full into memory, I think) - anoying. So I took
advantage of Norton's SpeedDisk options which made possible to
organize physical file placement on the drive (You know: place all
'*.tmp' at the end, other type at beginning, and so on...).

I forced positionning of all *.ttf (true type fonts), *.fon, *.fot (or
something like that) as first data on the drive, in a sequential,
alphabetical order - so they where gathered together as a "solid"
suite of font datas. I then assured myself that the loading sequence
of those fonts was logically the same in Windows' interest - meaning I
edited the win.ini (or whichever ".ini" that apply but I don't recall
it!) to gather and then sort font specific lines in the same
alphabetical order. Saved the '.ini', and rebooted...

Day VS night!

I could HEAR it when Windows was getting at this portion of boot
sequence: the hard drive was suddenly transfering datas to memory in
silence, as a the head was not seaking everywhere! Simply reading
track-to-track information as if it was a big, single, defragemented
file! Running smooth as a rolling tape, but fast as hell (as
absolutely NO seek time was involved)! But there was HUNDREDS of font
files... so what would have been the benifit of a simple file based
data defragging? Shorter task time, yes, but this kind of
accelleration would have been impossible, and magnetic heads of the
drive would have been constantly moving, stressed as usual.

SO now... imagine a complete Win98 (or why not XP) defrag process
which would analyse and optimize as many file as possible to get them
aligned on the drive almost exactly as they will be call each time
Windows boot up: wouldn't it be great? =D Benefit in time would
largely overcome this time to time longer defrag procedure (hey! My
'defragmented' XP system is so big it now takes me a full 3m30s to
boot, the drive bathed in seaking noise! A shame!).

I'm pretty sure a good analysis of system bootup requests could be
done automatically before defrag, so miroring this structure to file
position on drive could be done for most of them. Leave sufficient
head space between those 'parented data blocks', and you leave room
for subsequent FAST defraging even if some files have been added or
changed... until those empty spaces are filled and complete block
rebuild should be required (LONGER defrag time then). Not every file
should be moved that way, but MANY of them could (those which are only
'loaded', rarely modified).

This is MY opinion: WINDOWS OS startup IS THE MOST ANOYING PART OF
TODAY'S COMPUTING EXPERIENCE. True, 'programs' loads fast ennough
now... BUT a 3m30s wait BEFORE having a chance to finally launch one
of those applications is way too much!

I will get my two SATA Raptors at the end of this week.

Part of my decision to get them IS this unacceptably poor performance
on bootup (with my Western Digital WD800JB 7200rpm, 80GB). I may have
delayed my decision if my main drive would not have gave me serious
failure signals
recently. But I was to wait a better price drop and get them anyway -
I am fed up with those lenghty transfer times!

An ingenious defrager could have made me postpone this buy after a
better price drop, or even after competition's response (will come!).


Feedback on this?

"Gary S. Terhune" wrote in message ...
While your reasoning makes plenty of sense in re Win 3.11 and any Win9x system
(including ME), a major portion of the time spent loading XP is due to
housekeeping tasks. No amount of analysis and defragmentation is going to
seriously affect those processes. Even where it might, the files involved would
change during every session and thus you would not only have to defrag before
every restart to see much advantage, the defragmentation logic created during
the previous startup would be obsolete by that time.
[-Snip-]

I was not taking my dreams for reality! I know Win XP do not behave
"passively" ennough to expect a performance boost from a strategical
defrag. Sad, this is.

I wonder if an OS as big as Win XP could have been tought in a way it
could optimize itself constantly, as to obtain similar time cut at
bootup? (each time processor load is low, for instance)...

As OSes and silicons are growing bigger and bigger, how to exploit
this extra hardware power seems to evolve inconsistantly. Hard drive
have always been the slowest component for data transition, and it
will stay that way as long we'll depend on highly mechanical interface
(when are "Hard-Silicon Memory Packs" of 80GB or more going to be
available? . I persist saying that defragmentation should not be
taken as a simple FILE optimization process, but should first serve
user's experience by optimizing some TIME consuming tasks. Powering a
system and wait after it to finally get into iExplorer and then open
latest email where a phone number relies before dialing it is a 4min
task on my side... damn TOO LONG. Better take my net capable cell
phone to get this email. Not a winning comparision for a $1200.00
hardware/software bundle!

I recall a time where booting a computer consisted of switching it on.
I wish OSes could one day be sold in a solid state cartdrige (with a
standardized interconnection to motherboard), like TRS-80, Atari or
C-64 where built more than 20 years ago! Programs where then available
right after a single switch on toggle. Wouldn't it be great if we
could switch from Microsoft to Linux or Apple OSes in a matter of
seconds? 1 GB (or more) of battery maintained RAM would allow this,
beside permitting constant data modification in it (opposed to ROM
pack). Fragmented RAM is of no object. Hard drive would serve only for
their HUGE capacity, for data and software collection.

Ahhhh.... dreaming of a better world!

And you just keep on dreaming, VROMB, s. That's the first step in making
things happen, and some of those things you're dreaming sound pretty darned good
to me!

Good talkin' to ya,

--
Gary S. Terhune
MS MVP for Win9x

"VROMB" wrote in message
om...
"Gary S. Terhune" wrote in message
...

Ahhhh.... dreaming of a better world!

Why don't you hardwire the Flash Memory chips from a 2GB
flash disk onto your north (or was it south) bridge
connector and save windows on there? ^_^

On a more realistic note your system will be faster if you
have 15000rpm harddrives, why not add 8 in scsi raid 0.
that way for each byte the pc has to save, each hdd will
save one of the bits...

Okay, way more realistically, why don't you just leave
your pc on? And if you can't, try suspend to harddrive aka
hibernate. This way when you put ur pc back on it skips
POST and Windows boot up and goes straight into Windows
where you left off. I tried playing a video and going into
hibernate, when I woke the computer up it resumed the
video where it left off :]

In article ,
says...

I recall a time where booting a computer consisted of switching it on.
I wish OSes could one day be sold in a solid state cartdrige (with a
standardized interconnection to motherboard), like TRS-80, Atari or
C-64 where built more than 20 years ago! Programs where then available
right after a single switch on toggle. Wouldn't it be great if we
could switch from Microsoft to Linux or Apple OSes in a matter of
seconds? 1 GB (or more) of battery maintained RAM would allow this,
beside permitting constant data modification in it (opposed to ROM
pack). Fragmented RAM is of no object. Hard drive would serve only for
their HUGE capacity, for data and software collection.

Ahhhh.... dreaming of a better world!


Err...You *CAN* do this with Linux and USB Flashcard Readers/Drives.

There are Linux distros that can be configured to boot off 8 meg and up
Flashcards if you have a BIOS that will let you boot off a USB Mass Storage
Device.