"Anthony Atkielski" <anthony(_at_)atkielski(_dot_)com> writes:
It isn't really that clever. For one thing, the
standards for any given category of addresses will vary
from one community to another,
Ah, no - the clever thing is partitioning the address
space at all. Whether or not this particular partition is
clever is something I do not address -- it is merely possible.
Moreover, it is possible with a trade-off in the Internet
between transport cost and efficiency versus extra
globally-visible routing information. There are some
added costs on the host side, but surely the folks behind
draft-ietf-ipngwg-default-addr-select-01.txt need to work
that out for the general case of multihoming anyway.
[many different partitions]
The number of required addresses spaces might be
infinite, and there would be no universal consensus on
the vast majority of them.
The IPv6 address space is pretty vast, or so we keep being told.
This surely should be considered a huge victory for
IPv6... We were forced to abandon the original structure
of (classful) IPv4 because of address space constraints,
and have been under pressure not to revise the current
CIDR structure, because of space-conservation policies.
Now with a much larger address space, and the allowance
for support for multiple per-interface addresses in hosts
and routers, we can explore totally different structures,
such as partitioning on boundaries other than "large
site", "medium site", "small site", and the CIDR
refinements thereto (/32-sized-site, /31-sized-site, etc.).
Sean.