Re: Preliminary minutes from the WG last week

"Vijay Devarapalli" <[email protected]> Tue, 4 Apr 2006 00:44:54 -0700
Newsgroups gmane.ietf.mobike
Message-ID <[email protected]>
> Personally, I am not as convinced as you are that they should never
> be used together to the same box. What if the client is
> on a v4 network? Or on two separate connections, including
> mixed v4/v6 and NATs. 

ok. for this you would use DS-MIPv6 to setup
an ESP protected IPv6-over-IPv4 tunnel 
between the mobile node and the home agent
with UDP encapsulation for the ESP packets.
DS-MIPv6 should be sufficient. the 'K' bit
would work for the DS-MIPv6 tunnel too. it
wouldn't matter if the outer source address
is IPv4.

> Popping up a level, it seems like
> some combination of IKE/MOBIKE/NAT-T, Mobile IPv6, Mobile
> IPv6 transition mechanisms, and MONAMI6 should be able
> to cover the following situations and combinations of them:
> 
> - Having a separate SGW before you can reach your HA
> - Integrated SGW/HA
> - IPv4 access networks
> - IPv6 access networks
> - NATs and firewalls
> - Multihomed mobile nodes
> - Mobile nodes that you cannot fully trust (e.g. might need
>   a RR before believing their new location)
> - Mobile IPv6 service that requires EAP authentication
> 
> What exact combination of solutions we should use
> is less clear to me. 

here I am afraid we are getting to system 
specifications. not protocols anymore.

Vijay