RE: ISATAP links connect *router* interfaces

["Templin, Fred L" <Fred.L.Templin@boeing.com>]

> I think I am missing something here. Why does ISATAP have a "model of 
> host" and "host variables" if it's meant only for router interfaces?

The introductory text of (RFC2461, Section 5) says:

   "This section describes a conceptual model of one possible data
   structure organization that hosts (and to some extent routers) will
   maintain in interacting with neighboring nodes."

I interpret this as routers may to some extent behave as
hosts and thus implement certain aspects of the host
specifications. Otherwise, how would ND work for links
that connect only routers (some of which need to be
discovered as "default" by others)? It seem to me it is
up to the individual routers to know which aspects of
the host specifications they need to implement.

> > But have all implementations done that? 6to4 was specifically
> > designed for router-only implementation; the first
> > widely shipped implementation was for host-based 6to4,
> > which has well-known issues.
> 
> I suspect I misunderstood the point. Otherwise, yes, indeed, several 
> implementations uses ISATAP as a last hop network, connecting several 
> hosts to an (upstream) router over a virtual link.

Maybe what I wrote above helps clarify the distinction
between hosts and routers somewhat. IMHO, this subtle
distinction has been a sticking point in our common
understanding for far too long.

Thanks - Fred
fred.l.templin@boeing.com
 
> -- 
> Rémi Denis-Courmont
> http://www.remlab.net/
>