The FBI built a new lab in Portland... I attended a presentation about it a couple years back. This is done on a daily basis there, AFAIK. One would also suspect that crime lords and rouge intelligence agencies have this capability also. So glad my tin foil hat is comfy, and that my account must be considered fictional.

   Ben

PS - If desktop fabrication evolves, it is likely that masses of old data could be recoverable *far* beyond the limits of the original reader/writer/media... unless they become physically destroyed or broken down, like CD rot I guess.


T. Joseph CARTER wrote the following on 3/27/2005 10:52 PM:

On Sun, Mar 27, 2005 at 10:53:41PM +0000, [EMAIL PROTECTED] wrote:


Q3)Shred sounds excellent for removing all traces of previous files, but 25X?, 
isn't that overkill?

Q4) I've heard it suggested that 4x through using the dd command to write zeros through the mbr to the end of the drive is enough hide sensitive old files, how true is that?



If your filesystem is reduntant or journals writes other than to file metadata (Ext3 does not do a full journal, only the metadata), shred won't work very well.

Actually, writing zeros over the device once is sufficient unless you are
going to actually take the drive apart.  If you're going to do that,
writing zeros four times is still going to leave some residual trace of
the former data.

More paranoid erasures involve writing several alternating bit patterns
over the media just to really foul things up.  For example, if you write
over each byte with:

0xff, 0x00, 0xff, 0x55, 0xaa, 0xff, 0x00

you can be sure that the average attempt to read data straight off the
platter will simply fail.

Let's get more paranoid than that even:

0xff, 0x00, 0x55, 0xaa, 0x55, 0xaa, 0x33, 0xcc, 0x33, 0xcc, 0xf0, 0x0f,
0xf0, 0x0f, 0xe7, 0x18, 0xe7, 0x18, 0xc3, 0x3c, 0xc3, 0x3c, 0x7e, 0xe7,
0x7e, 0xe7, 0xff, 0x00, 0xff, 0x00

Then write something from /dev/urandom to that byte, and clear it back to
0. If someone can pull the original byte off the platter after that, I'd
be fairly impressed.




_______________________________________________
EUGLUG mailing list
[email protected]
http://www.euglug.org/mailman/listinfo/euglug

Reply via email to