Le 08/11/2012 19:12, Michael Richardson a écrit :
{appologies if this is a resend}
"Alexandru" == Alexandru Petrescu
<alexandru.petre...@gmail.com> writes:
Alexandru> Well yes, let me try to understand what do you mean by
these big picture Alexandru> questions?
1) what use cases need topologically significant prefixes with
default routes
I am not sure why default routes are mentioned in this question of
topologically significant prefixes (i.e. using Internet global addresses)?
I think a default route (an entry in the routing table) needs to be
known at any host or router which gets a global address or prefix
(except maybe DFZ). This is so because it is impossible to store in a
node a route for each destination its application may run (there are
many applications).
One may question why a default route is needed in a smaller local domain
of ULA addressing were Internet is not available. I think that even in
that case a default route is useful for its relationship to hierarchical
addressing - because it offers smaller routing table sizes.
Alex
2) what use cases need to exchange routes to locally configured
Non-Connected Network prefixes (whether RIR NCN, or ULA)
Those two situations need to be seperated. They are not the same
thing, or the same problem.
In case #1, one needs to further split this into 1a) managed
situations where all systems are under the control of a single
administrator (aka "intra-AS")
1b) unmanaged situations where an individual lends another individual
some bandwidth.
(1b) seems really easy in today's multiply NAT'ed IPv4, because the
NAT erases all evidence that first individual might have violated an
AUP. (I disagree with those AUPs)
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
