On Fri, Mar 29, 2002 at 02:06:44PM -0800, John S. Lyons wrote:
> I've done dd if=/dev/zero of=/dev/hda in the past.  What is the
> advantage of using /dev/ramdom?

There are ways to recover over-written data if you have physical access
to the drive.  Some involve removing the platters and checking them out
with other equipment.  You should have some well-financed enemies to be
worried about this kind of stuff.

However, it's good to be a little bit paranoid.  Someone in the list
recently quoted over-run with 5 passes is still recoverable, and I heard
a while ago a slightly larger number, but I don't remember exactly.

Over-writing with zeroes over and over won't really eliminate those
traces, or if it does, not nearly as well as random data.  The random
data will make it harder to figure out where you're hitting real data
that was present before the wipe.

If you have physical security of a level you are satisfied with, you
shouldn't need to do this, but since we are talking about machines being
thrown out (or at least, I think that's how the thread started), when
you lose physical control of the drive, its good to think of these things.

If you over-write with zeroes, there should be no way to get the
information back with just software.

and as I said in another post, /dev/random will run out quickly, so you
probably want /dev/urandom.

On a related note:

Does anyone know of any disk-wiping utilties for linux that will wipe
free space?  bcwipe does file slack space, but the only way I currently
know to wipe free space involves dd'ing files in until you fully fill the
drive (which has to be done as root to fill over 100%), and therefore
means should be done in maintenance mode, since most services become
unhappy when they can't create any files.  I'd like something I can run
on a server while keeping it up.


and specifically wrt bcwipe, I know on-board cache on a HD is going
to make wiping small files useless, but I'm curious if bcwipe does a
sync call after every pass, since that would make it slower, but more
effective on a system buffer level.


Rob




Attachment: msg05505/pgp00000.pgp
Description: PGP signature

Reply via email to