Date:        Tue, 23 May 2000 19:59:53 +0930 (CST)
    From:        blymn@baesystems.com.au (Brett Lymn)
    Message-ID:  <200005231029.TAA07534@mallee.awadi>


  | You cannot use dd unless the filesystems are _exactly_ the same size

That's pretty true - you can use dd if the target is smaller than the
origin, but the dd will fail when it hits the end of the target, and what
you will end with is a mess, useless for anything ... but you can still do
it.  If the target is larger than the source you can do it as well, but the
filesysem won't ten allow more data to be stored, the excess space on the
target will just be wasted.

  | on _exactly_ the same model disk

That's not important - filesystems are just linear arrays of blocks.
Performance might (just possibly) be a bit worse if the target is
notably different than the source, but everything will still function.

  | in _exactly_ the same place.

And that isn't true at all, the block numbers in a filesystem are all
relative to that filesystem, so where the copy is placed is irrelevant.

  | Personally, I would only use dd if I were copying an entire disk to
  | another one that was the same.  My preference for moving filesystems
  | around is to use dump and restore.

Yes, so would I, but dd can be faster, it certainly doesn't require you
to go in and manually remove the "restoresymtab" file (that way too many
people forget to do after completing a restore sequence - leaving it
will damage the ability of later dump/restore sequences to work), and dd
doesn't fiddle the inode ctimes, that dump/restore does (and you don't have
to newfs the target first, or mount it).   dd also works regardless of
filesystem type (when was the last time you did a dump/restore to copy
a mdsos format floppy...)

kre