RE: Middleboxes [Was: Flow label versus Extension header - protocol itself]

["Christian Huitema" <huitema@windows.microsoft.com>]

> >> IMHO shim6 being able to be done in middleboxes is actually a 
> >> requirement if you really think this will every be deployed at all.
> 
> I don't agree. I think the best hope of deployment of 
> anything these days is stealth, and that means (in this case) 
> new versions of host TCP/IP stacks, without the people 
> operating the network having to change anything.

I am with Brian on that one. This sort of changes is typically done in
service packs, which would enable wide scale deployment of shim6 in a
couple of years after the standardisation.

Besides, the assumption that a middlebox sees all of the host
communications is often wrong today, and will likely become wrong even
more often in the future. For example, the middlebox does not see the
content of a VPN connection, and does not control the allocation of
addresses by the VPN server. The middlebox does not see the content of a
GPRS or UMTS connection. Even on a home network, you may well end up
with several middleboxes, e.g. one for the DSL network and one for the
cable network.

-- Christian Huitema