Keith Moore wrote:

> in that case, I think the solution is the null set.  because none of:
> 
> - the fact that traffic from a client appears to come from an address
> - the fact that the client appears to respond to traffic sent to
>   an address
> - the fact that a client claims to be authorized to make assertions
>   about an address
> 
> are suitable as verification that a client is authorized to make
> assertions about that address.  such authorizations are *inherently*
> made by third parties - i.e. network administrators who are responsible
> for assigning addresses for use by clients.  the *only* way to verify
> client assertions about address bindings is to verify that the client 
> has been given the authority


I think I disagree about the third party part.

First, it turns out that return routability _is_ strong enough to
correspond roughly to the (current) security of the IPv6 Internet. This
would imply that that traffic coming from an address and apparently
responding to an address _can_ be used to make decisions about whether
a binding can be accepted or not. Similarly, CGA is somewhat stronger
than this and interestingly the nodes give an authority to themselves,
in a manner that others can verify this. (Subject to bidding down attacks
of course if we don't know what scheme to follow.)

Read more from http://www.piuha.net/~jarkko/publications/mipv6/Residual_Threats.txt

Jari

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to