On Wed, 2005-03-16 at 04:41 -0500, Michael Banta wrote:
> I am aware of a full ip address, just figured I would spare you the full 
> address since it is not pertenant to the question.
> 
> I keep reading that with 6to4 addresses, they are supposed to start with 
> 2002: prefixes so that autoconfiguration can take place with the clients.

6to4 address are something quite different from a block provided through
a tunnel. a 2001 address is a "real" IPv6 address -- that is, a part of
the IPv6 Internet with no IPv4 dependencies.

6to4 addresses (i.e. those starting with 2002), on the other hand, are
part of the IPv4-to-IPv6 migration plan. If you have a globally
aggregatable IPv4 address (i.e. an IPv4 address that anyone on the IPv4
internet can send packets to, such as _not_ a part of the 192.168.0.0/24
blocks), you are, with 6to4, automatically given a /48 IPv6 subnet.

I know I haven't really explained this very well at this point, so I'll
try with an example. I'm using 6to4. I have a static IPv4 address,
82.182.133.20. Written in hexadecimal, that is 52.b6.85.14. Using that,
I can construct my automatic 6to4 subnet: 2002:52b6:8414::/48. I, too am
using a Linux router with radvd, and the computer I'm typing this from
has gotten the address2002:52b6:8514:200:20c:76ff:fe3b:a3f4. The nice
thing with this is that I need no tunnel provider. The bad thing is, of
course, that it depends on IPv4.

The way 6to4 works is that when my router detects an outgoing IPv6
packet, it first checks the destination address. If it starts with 2002,
it rolls the packet inside an IPv4 packet, checks bits 16 through 48 in
the destination address, and put those in the IPv4 destination address
field. For example, when communicating with my friend, who also uses
6to4 and has the IPv4 address 213.132.111.101, I send a packet to her
IPv6 address, 2002:d584:6f65::1. My router extracts d584:6f65, which is
213.132.111.101 in hexadecimal, and puts that in the IPv4 packet's
destination address field, puts the IPv6 packet as the IPv4 payload, and
sends the packet. When her computer picks it up, it unwraps the IPv6
packet and uses it.
When communicating with a non-6to4 address, my router sends it, again
wrapped in an IPv4 packet, to a IPv4-to-IPv6 router on the Internet.
Many ISPs support the anycast address 192.88.99.1, which always means
"the closest IPv4-to-IPv6 router". When a non-6to4 host wishes to send
v6 packets to me, it just sends them normally and the IPv6 Internet
backbone will route them to the closest IPv6-to-IPv4 router, which will
wrap their package in an IPv4 packet, check the IPv6 destination address
(2002:52b6:8514:X) and calculate the proper IPv4 destination address (my
82.182.133.20 address) from that, and send it to me over IPv4. My router
will then unwrap it when it gets it, and forward it over my internal
IPv6 network.

So as you see, 6to4 addresses are something quite different from the
2001::/48 block that you got from your tunnel provider.

> The /48 was given to me by the provider.  I am aware of the addresses 
> construction, just can't figure out how to get the clients to connect 
> through the router.

I still don't really understand what your actual problem is, after all
this.

Hope this helps.
Fredrik Tolf

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

Reply via email to