Jari,
4. Do something else that is important for Shim6. What?
We would like to move forward with this in a couple of weeks. The WG
will either be closed or rechartered. Let us know if you are interested
and if so, what you believe should be done.
As you know, Sébastien Barré has written an implementation of shim6 that
is fairly complete and integrated in the Linux kernel. If there are
other implementations being developped, it would be interesting to do
interoperability tests among them to check that all implementors have
interpreted the shim6 specifications in the same manner.
Concerning the additional work that could be useful to pursue the
development of shim6, I would suggest the following :
- document the issues in supporting shim6 in firewalls. shim6 creates
new challenges for firewall vendors and documenting them would be useful
- network operators have argued that one drawback of shim6 from their
point of view is that host-based multihoming gives fewer tools for
network operators to engineer their network. I would say that operators
can have *different* tools to engineer their network. The solution being
developped within the ALTO WG could serve as a basis for a technique to
allow operators to influence the selection of the addresses by hosts in
order to traffic engineer their network. This could be considered as an
extension of the locator-pair-selection work
- another possibility could be to look at the interactions between shim6
and multiple tcp which have been discussed recently
Olivier