Re: deleting a hard drive
From: Moe Trin (ibuprofin_at_painkiller.example.tld)
Date: Wed, 15 Sep 2004 15:47:34 -0500
In article <wOJ1d.4214$F75.3528@trndny01>, David H. Lipman wrote:
>I trust the US DoD's new methodology.
I'm not sure I would.
>It basically describes writing a pattern such as 01010101 the writing its
The fallacy is that this assumes the ancient FM encoding scheme from the
1970s - and NO disk has used that scheme in _decades_ Modern hard disks
use RLL 2,7 encoding, that takes groups of two, three and four bits and
encodes them into four, six, or eight POSSIBLE flux transitions. The
DATA MAGNETIC CHANGE DATA MAGNETIC CHANGE
10 NTNN 11 TNNN
000 NNNTNN 010 TNNTNN
011 NNTNNN 0010 NNTNNTNN
Thus, the '01010101' bit pattern is encoded on the disk as:
010 10 10 10 10 10 10 10 10
TNNTNN NTNN NTNN NTNN NTNN NTNN NTNN NTNN NTNN
while the '10101010' bit pattern is encoded on the disk as:
10 10 10 10 10 10 10 10 10
NTNN NTNN NTNN NTNN NTNN NTNN NTNN NTNN NTNN
Now, removing the spaces I put in here so that you can see the transition
patterns, you get:
You should be viewing this in a fixed width font on your news tool. But
this shows that the complementary data pattern does NOT result in a
complementary pattern ON THE PLATTER. Here, I show an X where the data
changes from 'transition' to 'no transition' (or vice versa), and a dash
where there is no change.
So, half the magnetic domains are unchanged. Note that if you were to
map the domains (North verses South), this is only s _SHIFT_ of where
the changes occur.
>then another patter such as 11001100 and repeating this six times.
11 0011 0011 0011 0011 0011 0011
TNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN
0011 0011 0011 0011 0011 0011 0011
NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN NNNNTNNN
Now you may have a clue as to why this is for "Unclassified DoD Computer
Hard Drives" only. I repeat. If you are going to store stuff on a disk
that you don't want anyone to see, YOU MUST DESTROY THE PLATTERS TO
PREVENT ANY POSSIBLE DATA RECOVERY.