Pekka,
> Related to the debate on the mobile-ip list on whether
> MIPv6 should use Routing Header or some other mechanism
> to carry the home address to a mobile node that is
> away from home, I'd like to solicit for well grounded
> opinions on the intended semantics of the Routing Header.
>
> To properly understand what I am really asking about we
> must first make a distinction between two logical
> entities, endpoints and locations. In the current
> Internet architecture, both endpoints and locations are
> identified with a single mechanism, i.e. with IP addresses.
> However, they are conceptually different, as very well
> pointed out by Noel Chiappa in
> http://users.exis.net/~jnc/tech/endpoints.txt
> In essense, a communication endpoint is an active
> entity engadged in the communications, i.e. a party
> who consumes messages and generates new ones. A location,
> on the other hand, is a topological "place" within the
> routing fabric.
>
I think there is something missing in this analysis. A
communication endpoint requires more than just the
address to identify it, it also requires the port.
> Now, my question is fairly simple: Is the meaning of
> the routing header to allow a packet to be sent through
> a number of communicating hosts (end-points) or via a
> specific path, identified by a set of locations? Or
> is it both?
>
To me, it is fairly clear that the routing header is just
to route through a determined set of intermediate
forwarding nodes, i.e. routers, regardless of their
communication status. The header does not nor
should it require that a forwarding node be running
a specific protocol and thus support traffic
to a particular port in order to obtain forwarding
service.
> One way of pondering the question is to imagine that the
> endpoints and locations had different name spaces. For
> example, you could imagine that each host has a flat
> name tag (HIT in HIP terminology), and the locations
> are named according to the routing hierarchy as today.
> Under such an architecture, would the routing header
> contain addresses or these new name tags, or could
> it contain a mixture of both?
>
> The reason why I am asking this is the scenario, which
> Charlie Perkins has offered, where a packet is source
> routed through a Mobile Node that is away from home.
> According to his argumentation, in such case the
> routing header should have a route looking like
>
> .... - Care-of-Address - Home Address - ....
>
> where the Care-of-Address is the current location of
> the mobile node, and the Home Address is more like
> the end-point identifier of the mobile node. I am
> having hard time in clearly crasping the intended
> semantic meaning of this construction.
>
I've been loosely following this discussion, and
I think Charlie's argument has merit but I think
it is orthogonal to what the semantics of the
Routing Header should be. Currently, as
Charlie has pointed out, the two functions
are served by two header options. The MIP
case may be unique enough that perhaps
a new option is needed combining the
CoA and HA. I don't really see
this as being a general problem, though.
jak
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
