Christian Huitema writes:
> I believe that the source address should be defined as "the address at
> which the source would like to receive replies to this message." In the
> case of a mobile host, there is value in getting replies either at your
> transient address or at the permanent address -- it basically depends on
> the circumstance, just like multi-homing, and things would be simpler if
> both were allowed.
Christian,
This is an excellent piece of insight. As it
stands right now in MIPv6, the mobile node sending
toward the correspondent node is supposed to put
in a Home Address destination option. This at its
base causes a number of problems, such as the
ones I've already pointed out, but also there are
yet more problems when you start to consider
mobile nodes behind mobile routers. Consider a CN
which got multiple binding updates from both the
MN and the MR: how would it know how to assemble
the routing headers in the proper order?
Perhaps we should take a lesson from SMTP and/or
SIP here. When I send a mail message, my From:
address is essentially constant. If I want the
return mail routed differently, I insert a
Reply-to: header. SIP extends this a bit more
and allows Proxies to add Record-Route to force
return signaling through intermediate nodes. There
is a very analogous situation with basic IP
routing, it seems, where routers are similar to
proxies. This should hardly be surprising.
What I'm thinking is that maybe we should consider
flipping the sense of current MIPv6 thinking:
instead of placing the CoA in the IP header,
you instead place your home address in the IP header
because it is, in fact, the way to reach you. If
you're away from home, however, or for any other
reason such as the case for multihoming, you would
place a Reply-via destination option into the
packet. The Reply-via is, in fact, the ordered set
of IP addresses that should be visited before the
home address; in other words, it's nothing more
than the explicit routing header stack. If
intermediate nodes, such as mobile routers, could
push their addresses onto the routing header
stack, then we will have solved the other
nettlesome problem and have done it in such a way
that we can keep unwitting stationary nodes behind
a mobile router unaware of their mobility.
I believe that the security considerations are the
same, however: if we use return routability as our
basic test for correspondent nodes, this has the
same considerations, I believe.
Mike
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
