Re: DSTM

[Francis Dupont <Francis.Dupont@enst-bretagne.fr>]

 In your previous mail you wrote:

   > => "4over6" won't solve the same kind of problems, for instance
   > it couldn't help to communicate with an old IPv4-only-for-ever printer.
   
   So how does DSTM solve this?
   
=> yes, your remark that 4over6 is more oriented to direct
communication between dual-stack nodes is in the same kind of ideas...

   I was assuming there would be a v4-over-tunnel cloud and a native v4 
   cloud, and obviously some kind of connection between both clouds.
   
=> in DSTM there is no need to add this kind of gateways between
the two clouds, i.e., it is built in.

   In DSTM there is a single place where this happens: the gateway where 
   the tunnels terminate.

=> there is no requirement to have only one gateway. In fact the only
constraint is when the gateway gets automatically the IPv6/IPv4 address
mapping packets from the legacy IPv4 world come back through the same
gateway.

   "4over6" could be more flexible as packets don't 
   necessarily flow through a single gateway. We may even provide for some 
   kind of "proxy 4over6" tunneling so that IPv4-only hosts in otherwise 
   IPv6-only networks may enjoy connectivity.
   
=> I really liked 6over4 so I should like 4over6 too but I am afraid
the arguments against 6over4 apply too to 4over6. And the DSTM space
was already explored: there is nothing that 4over6 can do and DSTM can't.

Regards

Francis.Dupont@enst-bretagne.fr