Re: 6to4 relays [Re: WG Review: IPv6 Operations (v6ops)]

[Brian E Carpenter <brian@hursley.ibm.com>]

Pekka Savola wrote:
> 
> On Tue, 10 Sep 2002, Brian E Carpenter wrote:
> > Francis Dupont wrote:
> > >
> > >  In your previous mail you wrote:
> > >
> > >    > => I am not convinced by 6to4. For instance by the 6to4-relay concept.
> > >
> > >    6to4 relays cause some rather interesting problems, especially relating to
> > >    quality of service (too few of them at the moment, leading to huge
> > >    delays connecting to 6to4 nodes from some areas).
> >
> > If the worst problem with 6to4 is already a purely operational one,
> > I'm not at all distressed, and I wonder whether we can encourage
> > all XPs to operate a 6to4 relay?
> 
> I wonder if we can encourage Microsoft implementations not to assume bogus
> things about 6to4 relays.  Like it having the address
> '2002:<RELAY_IPv4>::<RELAY_IPv4>'.

The specification that the IETF actually approved was for router-based
6to4, not host-based, and it was assumed that router operators would
club together to configure and share relays. Clearly, a client-host 
based implementation such as the one you mention leads to quite different 
issues around relay deployment, which were not considered in the original 
scenario. We are still missing an RFC really analysing host-based 6to4,
and until that work has been done we don't know what is bogus.

   Brian