[6bone] semi-newbie Q on IPv6 address planning

[Nick Kraal]

I heard that this mailing-list holds the most experienced IPv6 prefix and IP
address allocation planners. I am in the middle of deploying our IPv6
network and in some mental block when it comes to IP address prefix
planning. Basically we have been allocated a /32 from APNIC and need some
advice/pointers in further allocating IPv6 addresses and prefixes.

Have read RFC2373/2374/3177 on this. Basically  we plan to allocate /48 for
end-customers and /40 for our pNLA customers. So basically for a /48
allocation I have the full 16 bits to play around with and for a /40
allocation only 8 bits leaving the last 8 bits in this field for the pNLA to
assign to their end customers. The end customers in both cases allocate
further networks in the SLA field. Reading on the web there are many methods
or allocating these bits ranging from allocating some bits:

a. differentiate core from customer networks e.g. between /40, /48, /64
b. geographical PoP sites and further bits to do point a. above

Have looked at the websites on this for SLAC, Stanford and Internet2,
Abilene. Are there any BCPs out there advising on this or are we on our own.
I have worked something out based on the information from the Internet -but
looks quite dodgy.

Do we have to stick to allocating in lots of 8 bits? Is a /44 allocation
valid? And what is a /126 allocation for a point-to-point link? I thought
the last /64 is reserved for the EUI-64 interface ID.

Would appreciate any pointers/hints/websites/etc on this.

-nick/