On Mar 12, 2013, at 11:29 PM, David Conrad <[email protected]> wrote:
>
>> Each
>> region has its own approach and policies regarding provider-independent IPv6
>> space, and issuance of IPv6 space (particularly on a flat-basis) via a new
>> process is likely to raise questions which need discussion.
>
> And this regionalization is part of the issue: what would happen if the
> outcome of that "discussion" is that an RIR refuses (for whatever reason) to
> allow LISP experiments?
Indeterminate at this time.
By RFC 2860, the IAB has final say in these matters (e.g. is the EID prefix
an experimental assignment, etc.), and the IANA shall follow its direction.
By ICANN's mission and core values in coordination of IP addresses, it may
come down to a judgement call of whether affected parties have had adequate
opportunity to participate in the decision making in accordance with their
policy role, and whether the results would be in the "public interest"...
(and there are also some related obligations on ICANN due to both the USG
NTIA IANA Function contract and the ASO MOU signed with each of the RIRs.)
In any case, irreconcilable disagreement between IAB/IETF and ICANN on such
a matter "would be bad" (major understatement) for the Internet ecosystem.
The good news is that we rarely end up in such situations... it's not much
different than when AD's on the IESG have a disagreement, or when a global
address policy being discussed at the RIRs has different text; in all cases,
the result is more engagement and discussion to gain better understanding
of everyone's concerns, with the goal of addressing them where possible
and fully understanding them but deciding otherwise when not possible.
In fact, there's actually a very good track record here between the folks
in the IETF working groups and folks in the RIR communities; some examples
include RFC 3849, RFC 6598, etc.
>> If EID prefixes are allocated
>> on a simple flat basis, and might be used also as IPv6 blocks, then they do
>> not have the same property if expansion is needed for their IPv6 usage. Is
>> this a problem?
>
> Wouldn't this concern be trivially addressed by having the EID allocator hand
> out EIDs using the same "sparse-mode" algorithm the RIRs are using?
I imagine so, but presently there's no way to know if that being proposed or
not... Similarly, there is work in areas such as reverse DNS/DNSSEC, RPKI,
whois/crisp/weirds being done (protocols at the IETF, and implementation by
the RIRs and the IANA team at ICANN) for which an additional set of end-user
IPv6 assignments (via the EID prefixes) could have implications. I doubt there
are any real problems here, but the need for coordination is quite real.
>> I have no idea, but definitely want to make sure that the
>> ISPs who felt it was important in the various regional discussions have a
>> chance to consider this new potential with any EID prefix assignment plan.
>
> To be honest, it sounds like you mean "veto" instead of "consider".
Read above... "coordination" is probably the most appropriate term.
> This would be distressing since one of the long term promises of LISP would
> be routing system scalability. Of course, from an ISP's (that is, the folks
> you claim to a "liaison" for) perspective, LISP has the downside that it
> facilitates end site provider independence so there would be good reason to
> try to veto LISP experimentation. Not that I'm actually so cynical to think
> that's the reason for your concerns... :)
Personally, I am a huge advocate of exploring new approaches to managing
routing system growth - our indirect approach of relying primarily on
provider-based hierarchical address assignments as a control knob on
routing table size has performed adequately in the past, but we have no
assurance that it is up to the job going forward. I'll also note that
regional policy has generally tended towards more access to provider-
independent IPv6 assignments, but there are variations in the regions
and suddenly having them universally available in traditional IPv6
routing without some other feedback system on routing table growth is
not without risks.
LISP's loc/eid separation provides a rational way to segment the routing
problem and thus has allowed real focus on the scalable mapping problem.
(I speak firsthand, as I have been advocating for eid/loc split since back
in 1992 with draft-curran-tune-00, and as well looking at alternatives for
EID mapping back in LISP's early days with draft-curran-lisp-emacs-00...)
However, despite LISP's wonderful end-state and all of our advocacy for
it, we also have to make sure that the deployment of EID prefixes does
not cause problems for the existing operators, many of who have worked
diligently through the RIRs to adopt specific policies for how IPv6
prefixes are issued. I do not claim to know how EID prefixes (which
may also be used as IPv6 prefixes in the traditional routing system)
will impact these folks, but do believe that they have a right to be
told what is being proposed so that there is an ability to ponder the
implications and have their input considered in the discussion.
> More seriously, I agree that the experiment needs to be documented along with
> its potential implications and termination conditions and I'm more than happy
> to contribute.
While committing such information to paper does take a bit of effort,
it both allows for informed consultation with the greater community and
therefore is very important in avoiding surprises for existing operators.
FYI,
/John
Disclaimers: My views alone. Relations in this email are complex with
no user-serviceable parts inside - open with caution.
_______________________________________________
lisp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lisp
