IP adresses are not unique enough.

David Bolen db3l at fitlinxx.com
Fri Dec 22 19:20:07 EST 2000


Ben Hutchings <ben.hutchings at roundpoint.com> writes:

> gogo at bluedynamics.com (Georg Gogo. BERNHARD) writes:
> <snip>
> > The MAC adresses are unique, as I understood.
> <snip>
> 
> I have some anecdotal evidence of cases where a manufacturer assigned
> duplicate MAC addresses.  Someone found me a documented case, too:
> <http://puck.nether.net/lists/cisco-nsp/2731.html>.

The global MAC address space is designed to be unique, but given that
most devices depend on programmed information during the manufacturing
process to be assigned the MAC address, there's always the risk of
error - either user (programming the wrong base into a programmer) or
machine (a broken programmer).

There is also the possibility (the feasibility of this on PCs probably
depends on the card interface and OS) of programming a MAC to use a
portion of the local address space, in which case it's up to local
management to manage this.  This sort of thing was much more prevalent
on end stations I believe with Token Rings rather than Ethernets but
its possible.  And it's actually used to good effect in systems where
machines back up each other (like a pair of Cisco routers) where the
backup takes over the primary's MAC address in a fallover situation.

In a locally reprogrammed case, you're still pretty safe assuming
local uniqueness to the wire (or things wouldn't function too well),
but not necessarily global uniqueness.  I suppose you could detect a
local address (bit 1 of the first octet) and treat that specially or
something.

But aside from chip serial numbers (and the ruckus that raised), MAC
addresses from ethernet chipsets are probably the closest thing to a
unique computer stamp (while that chipset is present at least) as we
have.  But it's not without some risks of duplication nonetheless.

--
-- David
-- 
/-----------------------------------------------------------------------\
 \               David Bolen            \   E-mail: db3l at fitlinxx.com  /
  |             FitLinxx, Inc.            \  Phone: (203) 708-5192    |
 /  860 Canal Street, Stamford, CT  06902   \  Fax: (203) 316-5150     \
\-----------------------------------------------------------------------/



More information about the Python-list mailing list