Subject: Re: "proxy arp" for ipv6?
To: None <tech-net@netbsd.org>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-net
Date: 05/03/2002 12:32:56
>> Now, the question is, how can I do this for IPv6?  If we assume I'm
>> using (say) 0300::/96 on the house LAN, how can I get G to respond
>> to neighbour sols for S's address as well as G's own?
> You'll never get a /96!

I've actually got a /56 (assigned out of a friendly site's /48), of
which I'm using a /96, with the other 40 bits presently all 0.

> The smallest you are likely ever have is a /48.  (in particular, you
> get a /48 for each IPv4 address that you own, see the 6to4 stuff).

I'm reminded unpleasantly of the early days of IPv4, when the address
space looked infinite.  Handing out /48s like candy worries me.

But that's neither here nor there at the moment.  I am in fact using
"prefixlen 96", so that's what I put in my message; even when I change
details, I like to change as few details as possible....

> If you can fill it... well... I'm not sure I really want ever atom in
> my house to be SNMP managemeable :-)

One could make similar remarks about the IPv4 address space.  The
reason we're in a v4 address space crunch is that there's structure to
the addresses.

I've got xxxx:xxxx:xxxx:xx00::/56 and I'm currently giving my machines
addresses out of xxxx:xxxx:xxxx:xx00::/96, with a trivial correlation
between v4 addresses and v6 addresses (x.x.x.N matches up with
xxxx:xxxx:xxxx:xx00::N:0).  Wanting to maintain this matching is why I
wanted something akin to proxy arp - proxy neighbour sol/adv, if you
will.  I certainly could put S in a different /96 and do ordinary
routing, but that ties addressing to topology unpleasantly.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B