Re: Newbie Question about addressing impacts

[Brian E Carpenter <brc@zurich.ibm.com>]

Tony Li wrote:
> > > Ideally, in an Internet architecture that we don't have today, there 
> > > would be N prefixes, one per ISP.  However, the host
> > > portion of the locator would be a constant regardless of prefix.  The 
> > > IGP need not know about the external prefixes, so there's
> > > no issue there.
> >
> >
> > Good point.
> >
> > But this implies modifying IGPs, right?
>
>
> Not seriously.  You would be modifying them to carry around a constant 
> local portion of the address and ignore the "global"
> prefix.  Simply picking a single constant prefix for use internally 
> would suffice.

Hence draft-ietf-ipv6-unique-local-addr-05.txt

I'd also like to observe that running with multiple simultaneous
prefixes (with its consequences for DNS and routing) has always
been part of IPv6; it's not a result of hypothetical multi6 solutions.

I also tend to agree that very large enterprises (such as I and Eric
happen to work for) may be best served by simply getting a /32, but
there are many medium size enterprises for which this is not true,
and that is exactly why we are running this WG.

    Brian

   Brian