Re: Defragging a directory?

[rubywand@aol.com]

In article <55smt7$1sge@dakota.gate.net>, stever@gate.net Steve Reeves
noted that ProSel's Read Efficiency tests are Block, not File, oriented.
So, there would be no difference in test results for a defragmented 3.5"
diskette and one which was not defragmented.


     This led to some tests which would check file reading and an
interesting discovery about defrag processes.

     The first test involved copying all of the files on a diskette to
/RAM5 using Copy II+ v8.3 on a IIgs. Copy II+ was used in order to avoid
effects of caching when copying under the Finder.

     The test diskettes were a fragmented ProDOS 8 workhorse diskette
(Disk F), a copy of the original diskette defragged via ProSel's
Optimization utility (Disk DP), and a copy of the original diskette
defragged via the process of doing a file-copy to /RAM5 (Disk DC).

     According to ProSel, the fragmented diskette (Disk F) has a
"Fragmentation" index of 280 with 105 fragmented files out of a total of
136. 1291 Blocks are in use and 309 are free.

     The ProSel de-fragmented diskette (Disk DP) has a "Fragmentation"
index of 0 with 0 fragmented files out of a total of 136. 1295 blocks are
in use and 305 are free.

     Disk DC was created by copying all of the files on the original (via
"Select All" file-copy under the Finder) to a just-inited /RAM5. Then,
/RAM5 was disk-copied (i.e. a block for block transfer)  to a diskette.
The idea is to let DOS, do the "defragging" by putting files where it
wants them. 

     A surprise is that Disk DC shows up with some fragmentation!
According to ProSel, it has a "Fragmentation" index of 5 with 4 fragmented
files out of a total of 136. Another surprise is that Disk DC ends up with
more free blocks than the other two: 1260 blocks are in use and 340 are
free-- a gain of 31 blocks over the original!

     In the test, the process of file-copying all of the files on a
diskette to a freshly inited /RAM5 using Copy II+ was timed. Here are the
results:

Disk F (original):                     4 min. 16 sec.
Disk DP (ProSel-defragged):    4 min. 26 sec.
Disk DC (Copy-defragged):      3 min. 55 sec.

     Evidently, while ProSel defragging may be very helpful for hard
disks, it is not very useful for speeding access to files on 3.5" 800k
diskettes. Also, for some reason, there is a loss of a few blocks of
storage. (By the way, a total file-copy under the Finder takes about the
same time as a ProSel defrag.)


     It seemed like a good idea to compare Disk F and Disk DC on a couple
simple ProDOS 8 tasks. Here are the timings:

             Launch a BASIC program    Load a packed super-res pic

Disk F              3.56 sec.                         3.19 sec.
Disk DC           2.75 sec                          2.25 sec.


     There is, roughly, a 30% to 40% speed improvement. In short, the best
way to defrag a 3.5" diskette appears to be via file-copy to RAM disk
(followed by disk copy to a diskette) or to copy files directly to another
diskette. Basically, 'Let DOS do it'.



 

    
 
 



Rubywand