Re: Minutes / Notes

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Kurt;

> >> I guess Iljitsch was talking about stateless autoconfig RFC 2462
> >
> > RFC2462 does not give any useful definition of autoconfiguration.
> >
> > For example, in DNSOP WG, people, including those of IPv6 ones,
> > are discussing autoconfiguration with DHCP.
> 
> Yes. But in the meeting it was pointed out after your question that 
> this was referring to the features described in 2462.

And, then, it was pointed out that RFC 2462 is useless to define
autoconfiguration.

> > It is as secure as the Internet today with an address binded both to
> > an ID and an locator.
> 
> a) I don't think that maintaining the security level of todays Internet 
> is a goal

I have never seen such requirement nor proposal.

Note, for example, that HIP, having no cryptographic security on
initial identity, merely maintains the security level of todays
Internet.

> b) Introducing loc / id separation will require mapping, one way or the 
> other.

Wrong. The separation requires that a host know id and locators of
its peer with reasonable security.

> Even in LIN6 there is mapping between the layers.

It's LIN6 mobility.

> This introduces new bindings that needs to be secured.

The separation requires that a host know id and locators of its
peer with reasonable security. An initial packet of a connection
containing all of them is just secure.

LIN6 mobility, of course, needs other forms of security.

							Masataka Ohta