This is not true. IPv6's TLA scheme has as its primary goal placing an
upper bound on the number of routing prefixes that are needed in the
Contrast that with today's IPv4 where the number of
prefixes that need to be maintained in the DFZ in order to have global
reachability is more-or-less unbounded, so some prefixes are not
reachable from everywhere.
As you know, this is not IPv6 magic. The underlying routing technology
between v4 and v6, namely CIDR, is identical. The only difference is that _by
convention_, the number of routing prefixes in v6 is limited. If we were to
create an IPv4 Internet' without the historical baggage of the existing IPv4
Internet, the same conventions could be applied.
Thus, traditional multihoming is still quite
possible, assuming the various ISPs that need to handle the routes
agree to do so.
I find it a bit strange that people seem to think the logic / incentives /
disincentives that drive multihoming in the v4 Internet will not apply in the