In a message dated 6/29/98 5:41:10 PM Eastern Daylight Time,
[log in to unmask] writes:

<< > Regarding the space advantages of using 4K... c,mon it's obvious!! Just
 > check out how many small files you have!!! I just did a quick find of all
 > files below 4K and I got 1090 just from the Windows directory alone. If
 this

 I'm trying to clarify what you did here.  I just did this search -- at most
 4K of the Windows directory -- and only got 175 files and a lot of them
 were folders.  Did you search only the Windows directory or did this
 "include subfolders"?  If you included subfolders, you may have also
 included your Temporary Internet Files folder and that folder does contain
 a LOT of small files.  When I included subfolders on my work PC, I got 1800
 files.  Searching all of my 2GB hard drive found only another 3500 files
 under 4K.  I'm wondering if the Windows folder is a representative sample
 .. >>
Hi,
  This thread seems to be dealing with the "slack" space on a hard drive when
using large cluster sizes. Small files are only one cause of wasting space
this way. ANY file that is slightly larger than a multiple of the cluster size
will do the same thing.
  For instance, if the cluster size is 32K, and a file is 33K (or 65K, or 97K,
etc...), then it will waste 31K of the last cluster it occupies. So only
searching for small files does not give a *true* picture of how much space is
really being lost due to the "slack".
  If we assume a true random relationship between file size and cluster size
(which is pretty close to real life), then the AVERAGE space lost to slack
(per file) will be approximately 1/2 of the cluster size (some files will
waste more, others less). A better way to measure the approximate total wasted
disk space due to slack would be the total number of files on the hard drive,
times 1/2 the cluster size.
  If there are 10,000 files on a drive, and the cluster size is 32KB (FAT16
partition larger than 1 gig), then the approximate wasted space due to slack
is 10,000 x 16K (16,000), or about 160 Megabytes. Doing the same math with a
FAT32 drive using 4KB clusters and 10,000 files, we get about 20 Megabytes of
wasted space. So a smaller cluster size does indeed waste less space (140 MB
of disk space would be reclaimed in this example by converting the drive to
FAT32).
  Another thing to notice here is that the number of small files does not
affect this waste (per file) due to slack, AT ALL, since a large file is just
as likely to be a space waster as a small one (although the percentage of
waste will be less for the larger file). The average waste per file will be
approximately 1/2 the cluster size, regardless of whether it is a small, or
large file.

Hope this helps,
Peter Hogan
[log in to unmask]