MIP Team wrote:

> If this is not the use then when will they be used and what for?
> I mean if they are meant for supporting IPv6 and IPv4 networks at the
> same time then they must have some welldefined use. If no use why even
> declare them?
 
IPv4-compatible IPv6 addresses are normal IPv6 addresses. On 
an IPv6 only host, they are treated as any other unicast address, i.e. send 
according to the default route of the host. The same happens with 6to4 
addresses.

On a dual-stack host, who is able to speak IPv4, it depends on it's routing 
table, where the packet goes. If the route points a router, it will be send 
as normal IPv6 datagram to that router. If the route points to a tunneling 
device of that host, the IPv6 packet will be encapsulated. But in contrast 
to static tunnels with pre-configured tunnel endpoints, the target IPv4 
address will depend on the IPv6 address, that's all.

IPv4-compatible addresses are intended for dual stack hosts, isolated in the 
IPv4 Internet, who want to speak IPv6.

The main problem with IPv4-compatible addresses is IMHO the missing well 
defined route to the native IPv6 Internet. 6to4 has it's anycast address 
192.88.99.1 for relay routers but RFC2373 misses that point. I think this 
has prevented widespread use of IPv4-compatible addresses. Nowadays, 
there's 6to4, which also allows one to connect single hosts, so
IPv4-compatible addresses are not needed anymore.

Mark.

P.S.: As already stated in this thread, and as far as I know, IPv4-mapped 
addresses should never appear on the wire. They are just there to tell the 
IPv6 socket API to use IPv4, not IPv6.

---------------------------------------------------------------------
The IPv6 Users Mailing List
Unsubscribe by sending "unsubscribe users" to [EMAIL PROTECTED]

Reply via email to