/127 between routers?

[Jim Burwell]

On 1/5/2010 22:25, Matsuzaki Yoshinobu wrote:
> Jim Burwell <jimb at jsbc.cc> wrote
>   
>> Makes sense to me.  It would seem to me that /126s were always the
>> logical equivalents of /30s in the IPv4 world for use on p-t-p links. 
>> /127s always seemed "wrong"  to me, since it uses the "magic" all-zeros
>> network identifier address as a host address.  :p 
>>
>> When I first started learning about IPv6, I wondered it the all-zeros or
>> all-ones host chunk of an IPv6 address held any special meaning as they
>> do in IPv4, and quickly figured out that all-ones doesn't, but all-zeros
>> still does.  A bit annoying in some ways since /127 would make a nice
>> XYZ::0,1 p-t-p interface address pairs, but oh well.  /126 it is.
>>     
> In case of /126s on p-to-p links, there is an unused all-ones address.
> I still worry about it.
>  - http://archive.apnic.net/meetings/26/program/apops/matsuzaki-ipv6-p2p.pdf
>
> A vendor said me the cost of special care for these unused address is
> too high - you need double lookup to prevent packet loops.  And they
> decided not to perform such double lookup.  So I prefer /127s on
> p-to-p links.  It's like an operational insurance to prevent packet
> loops.
> -----
> Matsuzaki Yoshinobu <maz at iij.ad.jp>
>  - IIJ/AS2497  INOC-DBA: 2497*629
>   
Interesting.  I'd hope most vendors would implement the RFC4443 special
case. 

Perhaps another solution in the case of a /126 would be to simply add
the all-ones address to one (or both) side, that way they simply respond
to that address as appropriate, and eliminate the forwarding loop
problem.  That is:

RT1 (2001:db8::1/126, 2001:db8::3/126) <--- ptp ---> (2001:db8::2/126) RT2

That way, the unused address doesn't result in the forwarding loop, and
no need for ACLs or anything like that.

Obviously this wouldn't work for /64s, so you'd use ACLs for that, or
the vendor would have to implement RFC4443, or perhaps some convention
could be used where all but the first two IPv6 host addresses on a p-t-p
link are dropped automatically (or alternatively accepted as if that
interface were the one addressed)?

Or maybe I'm too tired to be thinking about this right now.  :p