[AusNOG] My Predictions for the ISP Industry

[Paul Brooks]

On 16/03/2012 4:01 PM, Mattia Rossi wrote:
>>
>>   >  This means that the
>>   >  CPE is creating a 6to4 prefix out of it's public facing IPv4 address,
>>   >  which is then used to distribute v6 addresses to all his IPv6 devices in
>>   >  the house.
>>
>> More or less.  Or some other prefix.  But yes.
>>
>
> Some other prefix? E.g. fc00::/7 ?

The prefix distributed around the house is 2002:dcef:69e5:e472::/64 - which is a
globally unique prefix, derived from the globally unique external IPv4 address.

Local devices then use SLAAC to derive a unique address - which is globally unique

The address of my NAS is                     2002:dcef:69e5:e472:208:9bff:fec6:a050
The address of my desktop is              2002:dcef:69e5:e472:3813:960f:324d:73be
The address of my Android phone is   2002:dcef:69e5:e472:167d:c5ff:fea5:960d


>
>>   >  So far so good... but my question is: how do machines inside the house
>>   >  handle addresses of the 2002::/16 prefix?

Its a /64, and they do what every IP device does - if the network prefix part matches,
then try to reach the destination address over the local network. If the prefix is
different, go for the router

> So there's some form of NAT happening (1:1 NAT)? Link local in the LAN, and if the
> destination is outside the LAN, the CPE keeps the 64 bit host part of the source
> address intact, rewrites the 64 bit prefix part to the 6to4 prefix, and off they go?
> Interesting...

I'm not sure there is any NAT happening at IPv6 here. devices use their globally
unique address as the source address when trying to reach a destination outside the
LAN. The router acts like a..err..router (plus the 6to4 jiggery-pokery on the WAN side)


>
> Btw. mDNS is a real lifesaver when it comes to IPv6 and you don't have an
> authoritative DNS server for your LAN and possibly a real (or fake) domainname to
> use within it.
>
>>   >  I'm a bit surprised that this actually works...
>>
>> Aren't we all!
>>
>
> :-)

I didn't know I was supposed to be surprised it works - it seems to work in practice
though.
Anyway, thats enough debugging my home lab lads - thanks for the education - I'm off
for a pint or three :-)


Paul.