Let's continue discuss and will summarize after

2009/11/25 Durand, Alain <[email protected]>:
>
>
>
> On 11/24/09 11:25 AM, "Hui Deng" <[email protected]> wrote:
>
>>> You can use some form of header compression to mitigate this overhead. How
>>> to use header compression in that case can be defined in softwire
>> it is not that easy to do backward compatibility, and it's really hard
>> to use it today
>
> There are games we can play to "port" the relevant v4 header fields in the
> v6 header and remove the v4 header all together... As long as this is done
> on a link between two consenting nodes, this is fine.
==> what u suggest here is still same as the header compression.

>
>>> 4 could be a very straightforward extension of DS-lite. If you know the
>>> common IPv6 prefix shared by adjacent nodes, you can directly tunnel to
>>> them.
>> I don't understand here, are you suggesting all DS-Lite host has common IPv6
>> prefix? what kind of direct tunnel could be established?
>
> What I'm suggesting is to define a 'scope' in which nodes can talk directly
> using the shortcuts. It makes little sense to 'short cut' with a node on the
> other side of the country, but I understand why you'd want to do this
> between two nodes that are behind the same base station.
>
> Within the scope, each node has a unique v6 address and a unique v4 address.
> (again, either derived from the v6 or assigned via DHCP). Net 10 should be
> big enough for that.
>
> You pass both v6 and v4 prefixes via DHCPv6 to all nodes within that scope.
> That way, when a node sees that the IPv4 address returned by (say split) DNS
> is within the v4 prefix, it can use the v6 prefix to 'tunnel/route' the
> packets.
if the application generate IPv4 packet, and finally you will be use
IPv6 to transmit this packet,
isn't this a translation?

Secondly, when the receiver host got this packet, how could he know
this is normal IPv6 packet
or IPv4-translated-IPv6 packet?

thanks again

-Hui
>
>   - Alain.
>
>
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to