Re: BGP-4+

[Brandon Black <[e-mail]>]

I'm foregoing quoting this whole thread again, we've all seen the other
mail on this subject so far.....


Now this may seem silly to some who have actually followed the v6
discussions closely, I have not.  Someone please tell me if/why this is
preposterous, or if any of my presumptions are wrong...

Since the v6 address format is heirachical, then why not consider a
certain topmost portion of the v6 address the AS or RD equivalent...

i.e., a provider using AS 123 may have the prefix 12:34:56 (that may not
be the right length or numbers, but this is just theory, I don't know the
specifics) on all v6 addresses that will be used in that AS.  We could
continue to use the 16-bit AS numbers we've been using until the
transition to v6 is complete (I think that it is feasible that 16 bits
will last us until that point).  Once the whole backbone is on v6, just
discard the AS numbers, and start using the address prefixes as "Routing
Domains", which is really what they are if v6 addresses are heirarchical.

During the transition period, if AS's and prefixes pretty much have a
one-to-one relationship, then I don't see any logistics problems with
this... although certainly BGP4+ or IDRP or whatever is chosen will need
to be modified to work this way.... but the modifications would be
simple... instead of storing or transmitting network addresses with an AS,
the network address by itself would be sufficient, since it contains the
RD (prefix)... and the increased AS size of BGP4+ would be the size of
that prefix......  

The big hole I already see in this is of course the little unimportant
company networks that want AS's because they want to be multihomed.... I
don't know enough about how v6 heirarchical addresses are supposed to work
in a multihomed solution to know whether this is a problem or how to solve
it...

Just food for thought for those who know this stuff better than I do...

Brandon