>>>>> "Alexandru" == Alexandru Petrescu <alexandru.petre...@gmail.com> writes:
>> A different way would be for the IETF to instruct IANA to create a
>> NCN space, and designate the VIN-mapping as a part of it. This would
>> be harder for the IETF to do, but would I think, lead to a better
>> world.
Alexandru> NCN? Non-Connected Networks? That is another name for ULA
space no?
Possibly.
ULA-Random is one half of the space. No registration, no structure, no
ownership.
ULA-Central is one proposal on what to do with the other half of ULA
space, and it might involve registration, it might have structure, and
it might have ownership. ULA-Central could be a source of NCN space.
Or ICANNs could allocate a /12 for NCN space, or RIRs could each
allocate an /12 for for NCNs, or an RIR could allocate a /13 of an
existing /12, or some other more complex or subtle way.
>> you have confused globally routable with globally unique. These
>> prefixes are simply *NOT* going to be announced in the core of the
>> Internet.
Alexandru> Well, confusion may have crept in, sorry if so.
Alexandru> I understand ULA space is not to be announced in the core
Alexandru> of the Internet.
Alexandru> I understand PA Provider-Assigned addresses are to be announced
in the
Alexandru> core of the Internet.
You are now confusing address classes with policies encouraged by RIRs,
also what a sender does and what how a receiver acts. We can, at the
IETF never really "forbid" a sender from sending... rather we generally
write rules as to how a receiver acts when it receives valid or invalid
signals.
So, it's not that "ULA space is not to be announced", it's that
"announcements that include ULA space SHOULD be ignored".
It's a subtle distinction, because it explains that the decision about
what to do with address space is a decision made by mutually consenting
ISPs..
So, yes, unless things go horribly wrong, ULAs will be rejected by very
common policy, and we hope never to see them on the public default-free
zone. Given the size of IPv6 space, and the challenge of auditing where
leaked ULA traffic might be coming from, I hope that we will not see
ULAs in use in any place where there is administrator that could have
chosen something better.
PI (and even PA space) will generally be accepted by ISPs, provided a
number of other conditions are satisified; the SIDR WG has protocols to
cryptographically validate some of those conditions. (as an example
of PA space, I've announced a /48 of PA space to another ISP in order to
multihome an ISP)
There is no requirement for any set of addresses to be announced at all.
Second, there are in fact no rules what an ISP can and can not accept.
RIRs go to some lengths to make it clear that the address space that an
entity gets is not guaranteed to be routable...
--
] Never tell me the odds! | ipv6 mesh networks [
] Michael Richardson, Sandelman Software Works | network architect [
] m...@sandelman.ca http://www.sandelman.ca/ | ruby on rails [
pgpCOj37KFWZK.pgp
Description: PGP signature
-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
