Re: 6man w.g. last call for <draft-ietf-6man-default-iids-11.txt>

[Fernando Gont <[e-mail]>]

On 05/18/2016 09:00 PM, Lorenzo Colitti wrote:
> On Wed, May 18, 2016 at 11:13 AM, Fernando Gont <[e-mail]
> <mailto:[e-mail]>> wrote:
> 
>     No. My take is that the concern is flawed. Please read
>     draft-gont-predictable-protocol-ids, and even RFC4941, which talks at
>     length about security and privacy issues regarding reusing identifiers
>     in different context, for different scopes, etc.
> 
> 
> Fernando, please stop arguing that a random MAC address is a
> "predictable ID". That statement is false and continuing to argue for it
> is not helping this document advance.

I simply referenced a draft. When I referred to embedding MAC addresses
in IPv6 IIDs, I said "improper" IDs, rather than "predictable".

Reusing IDs in different layers for different scopes, etc., is a bad
thing. It just takes some else making the same bad decision, an now an
attacker can correlate traffic because you and other guy thought that
embedding an identifier from a lower layer in your own protocol, in a
different context, with a different scope, for a different purpose.

Brian noted, just as an example, the ability to correlate IPv6 and IPX
traffic. But this is obviously just an example that shows why something
that might seem benign in some environment may show up as being a real
bad idea.

The fact that you'd be wasting 18 bits of entropy in your IIDs (if you
generate them by embedding a MAC address -- even if randomized) is yet
another datapoint that you're doing the wrong thing: an indication that
you're trying to combat the symptom, rather than the disease.

Thanks,
-- 
Fernando Gont
SI6 Networks
e-mail: [e-mail]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492