new addressing plan

[davidk@isi.edu]

Pedro,

Pedro Marques writes:
> 
> Personaly i believe our present aproach is superior since it does not
> require any registry...

I agree that using the AS in someway is a very attractive way to go. The
nice thing is that you don't need any human judgement whether you can get
a prefix or not. However, the real world is not as simple as that.

In the end the decision has to be made who gets and AS and who not, which
will become increasingly difficult because we are giving more value to
the concept of an 'AS' then it had in the past. Furthermore, it is
already very common that an AS gets split up in several IPv6 spaces to
accomodate customers of the provider and other people to join the 6bone.
Doing this means that we need a (local) registry of some kind and it
makes sense to provide such a service for everybody at once instead of
letting all people to run their own database (I guess that most people on
the 6bone can spend their time better). Furthermore, we are supposed to
test things like aggregration and using an AS based structure will brake
that from the beginning.

Note that assigning the prefixes should not become a difficult thing. I
don't see much reason why this could not be done in a highly automated
fashion since we are dealing with temporary test addresses which are not
very attractive to hoard. We might even be able to use the registry for
renumbering purposes as a kind of global DCHP route slot server ...

David K.
---