Mobility observation

[Erik Nordmark <Erik.Nordmark@sun.com>]

Here is a food for throught how mobility can fit in with multihoming.

Assume that a multihoming solution allows the host to send its list of
locators to the peer.

In that case it is possible to provide hints about different characteristics
of the locators. Some might be long-term stable (what is a home address
in MIPv6) but might not be optimal for performance (due to the home agent
tunneling packets to the current location of the mobile node).
Others (care of addresses) might perform better but stop working
when the mobile node works.

Being able to provide such distinctions to the peer allows the peer's
destination locator selection to use different strategies e.g. when
packets to a locator no longer appear to make it to the mobile host.

Which such an approach one could continue to use the MIPv6 MN-HA
protocol for updating the home agent (this is independent of
the MN-CN aspects), and use the multihoming mechanism
to update correspondents with the set of locators.
A side effect of this is that a mobile node can provide multiple
care-of-locators to the correspondents to allow failover.

  Erik