ARIN board accountability to network operators (was: RE: [arin-ppml] [arin-discuss] Term Limit Proposal)

john,

i think your attemt to move the discussion to the arin ppml list
exemplifies one core of the problem. this is not about address policy,
but arin thinks of itelf as a regulator not a registry.

contrast with the ripe community and the ncc, which is not nirvana but
is a hell of a lot better. among other key differences, the ncc is
engaged with the community through technical and business working
groups.

e.g. the database working group covers what you think of as whois and
the routing registry. the wg developed the darned irr definition and
continues to evolve it. consequence? the irr is actively used in two
regions in the world, europe and japan (which likes anything ocd:-).

the routing wg works with the ops to develop routing technology such as
route flap damping. there is a reason that serious ops attend ripe
meetings. yes, a whole lot of folk with enable are engaged.

for years there has been a wg on the global layer nine issues.

the dns wg deals with reverse delegation, root server ops, etc. and
guess what, all the dns heavy techs and ops are engaged.

there is a wg for discussing what services the ncc offers. the recent
simplification and opening of services to legacy and PI holders happened
in the ncc services wg, it was about services not addressing policy.

and this is aside from daniel's global measurement empire. not sure it
is a registry's job to do this, but it is a serious contribution to the
internet.

the ncc is engaged with its community on the subhects that actually
interest operators and affect our daily lives.

there is nothing of interest at an arin meeting, a bunch of junior
wannabe regulators and vigilantes making an embarrassing mess. i've
even taken to skipping nanog, if ras talks i can watch the recording.
all the cool kids will be in warsaw. ops vote with our feet.

randy

john,

i think your attemt to move the discussion to the arin ppml list
exemplifies one core of the problem.

Randy -

I offered ppml out of respect to the nanog subscribers, that is all...

/John

And I would welcome discussion of how ARIN (and nanog) can be more like RIPE - that is very much up to this community and its participation far more than ARIN..

/John

hi john,

i think your attemt to move the discussion to the arin ppml list
exemplifies one core of the problem.

I offered ppml out of respect to the nanog subscribers, that is all...

i will refrain from characterizing the ppml list. needless to say, i do
not subscribe.

my point is that what arin does should be of interest to nanog
subscribers. in theory, the ops are the arin community, the registry
serves operations. if it is not of interest to ops, it is not serving
the community.

[ get out of s'pore yet? drc got delayed a day with a missing part for
  his plane! ]

randy

And I would welcome discussion of how ARIN (and nanog) can be more like

RIPE - that is very much up to this community and its participation far
more than ARIN..

/John

How about we fold ARIN into RIPE? Why not? I agree with all of Randy's
points. I am sure RIPE can easily scale up to take on ARIN services, with
fees being reduced for all involved due to economies of scale.

CB

nanog is a separable game. it is currently very confused between form
and substance, making committees for everything. like the bcop thing.
two organizations, nanog and isoc, forming organizational structures to
create a document store. the ops' doc store is ripe's because the ripe
wgs produced work and someone realized they needed a place to stash it.
so now nanog and isoc need to flag-plant. the up-side is that it's a
great b-ark, keeps them from doing damage.

And I would welcome discussion of how ARIN (and nanog) can be more
like RIPE

i purposefully phrased it a bit differently, how can arin engage, get
real participation from, and serve its community, the operators. i was
stealing examples from ripe.

but, for concrete action, how about a half day session at the next nanog
meeting on, for example, arin database services, whois and irr. not to
try to reach hard conclusions or plans. but to open a dialog to explore
what the community gets and wants from these services and how they are
provided.

or pick another key service.

randy

i will refrain from characterizing the ppml list. needless to say, i do
not subscribe.

my point is that what arin does should be of interest to nanog
subscribers. in theory, the ops are the arin community, the registry
serves operations. if it is not of interest to ops, it is not serving
the community.

I fully agree, but also respect that this community has made some
conscience decisions regarding having ARIN be quite registry focused
and letting NANOG evolve as as a forum of the operators in the region.
I believe that several of the initiatives that you noted from the RIPE
region could easily be viewed as falling under either organization.

This community should not be disadvantaged by the structure of having
a distinct registry and distinct operator forum, but it does mean that
we need to be able to sort out _what_ the operators want and then where
it gets done.

Internet routing registries are a fine example; one could argue that
it should be integrated with the number resource registry, but we also
have examples of independent routing registries in active use (and I
can see some potential reasons why operators might even want there to
be a healthy separation between those functions.)

If the community has one mind of what routing registry capabilities is
wants here, including how it wants it governed and operated, I am quite
certain that ARIN will support the direction, regardless of where it ends
up being operated and how it ends up being governed. The lack I have
noted over the years is lack of clear direction from the community, but
that should not be something "ARIN" jumps in and tries to bring about -
it needs to be of interest to (and led by) the operators on this list.

We agree that ARIN needs to be relevant to the ops community, and I am
very open minded to any suggestions you have, but don't exactly think
that your examples from RIPE are necessarily where we want ARIN to go
as much as things we want to have happen, whether that's ARIN, NANOG,
or other associated organizations. On the other hand, your governance
examples from RIPE (e.g. "wg for discussing what services the ncc
offers") are directly on target, and I will share them on some other
lists that may defy characterization by you.

[ get out of s'pore yet? drc got delayed a day with a missing part for
his plane! ]

(Getting closer... the last plane was a fail due to fuel pump issues;
my dearest friends at United seemed have rerouted me through Hong Kong
but omitted a flight onward. Oh well.)

Thanks!
/John

Speaking for myself, only here:

  I'll be happy to let ARIN manage routability of assignments,
once they guarantee routability of said assignments.

  Cheers,

  --msa

My earlier message was sent before I saw this, but I think we converged
on the important point: ARIN needs to engage in a much better manner with
the ops community (more than just an ARIN update preso and registration
helpdesk); this should be closer to a "services wg" model.

Got it,
/John

John Curran
President and CEO
ARIN

I, for one, would not want to start having to pay RIPE-level fees.

ARIN fees are a much better deal than RIPE fees.

Owen

The RIPE IRR is used very widely in the Africa region, too.

Great toolset.

Mark.

there is the 'measurement analysis and tools' working group http://www.ripe.net/ripe/groups/wg/mat guiding this work, and it even has an 'out-of-area' co-chair to emphasize the *globalness* of our empire :wink: :slight_smile: :slight_smile: :slight_smile: :slight_smile: :slight_smile:

seriously: the ripe ncc was not conceived as a "registry" but as an association of operators where they can organise common activities that require neutrality, expertise and common funding. so whether it is a 'registry job' is irrelevant in our context as long as the community agrees it is useful and the membership of the association agrees to fund it with their fees. the huge overlap between community at large and paying membership keeps this consistent.

daniel

Hi Owen,

I, for one, would not want to start having to pay RIPE-level fees.

ARIN fees are a much better deal than RIPE fees.

Only up to Small... The RIPE NCC membership fee is €1750 (±$2400 currently) for everybody. The ARIN fees are between $500 and $32000, with category Small at $2000 and Medium at $4000. I personally am glad about this (although in ARIN I would probably be Small) because it doesn't give operators any financial incentive to stingy when giving their customers IPv6 prefixes.

If you want to give a million customers a /48 it is not going to cost you more then giving them a /60. IPv6 resources are not such a scarce resource compared to IPv4, so differentiating price based on the amount of integers you need doesn't make much sense in the current world anymore :slight_smile:

But: this is all RIPE NCC members/AGM stuff, independent of the RIPE community and its working groups. (well the RIPE NCC facilitates the RIPE meetings (note: RIPE meeting, not RIPE NCC meeting) and without the help of the NCC the RIPE community wouldn't have such well organised meetings. The NCC only facilitates though, it doesn't control or influence the RIPE working groups) and the structure of the RIPE working groups was what Randy was referring to.

Cheers,
Sander

Compare and contrast the costs of being a PI holding end-user in the RIPE region to those in the ARIN region and the difference becomes much more noticeable.

Owen

Hi Owen,

Compare and contrast the costs of being a PI holding end-user in the RIPE region to those in the ARIN region and the difference becomes much more noticeable.

Yeah, RIPE NCC is definitely much cheaper for PI: no initial registration fee of ≥$500. The maintenance cost is $100/year vs €100/year (±$137) so there is a little difference there. The $37 difference will take at least 13.5 years to make up for the $500 though. And that is just for up to a /22. The $4000 initial fee for a /16 PI would take you more than a hundred years :slight_smile:

So yes: for PI the difference is much more noticeable, in favour of the RIPE NCC :slight_smile:

Cheers,
Sander

€50 per PI assignment from the ripe ncc, no?

http://www.ripe.net/ripe/docs/ripe-591

Nick

Oops. /me was confused. €50 indeed!

Met vriendelijke groet,
Sander Steffann

nanog is a separable game. it is currently very confused between form
and substance, making committees for everything. like the bcop thing.
two organizations, nanog and isoc, forming organizational structures to
create a document store. the ops' doc store is ripe's because the ripe
wgs produced work and someone realized they needed a place to stash it.

I like this example, but not sure how it could apply here. Need a NANOG
document series? It wouldn't be an ARIN document series, would it? Or
did I miss the point of your example?

i purposefully phrased it a bit differently, how can arin engage, get
real participation from, and serve its community, the operators. i was
stealing examples from ripe.

but, for concrete action, how about a half day session at the next nanog
meeting on, for example, arin database services, whois and irr. not to
try to reach hard conclusions or plans. but to open a dialog to explore
what the community gets and wants from these services and how they are
provided.

I like this example.
I also appreciate the policy hour, where NANOG attendees get a few minutes
on ARIN proposals.

In another message you complimented the RIPE Atlas project. I like the
work from APNIC's labs, too. I also like LACNIC's development projects,
FRIDA, +RAICES, and education efforts. Would these kinds of efforts be in
scope for ARIN? Does ARIN need a Chief Scientist (a la Karrenberg or
Huston)? Or is that a NANOG role, since it might include things outside of
management of number resources?

I think North American operators are missing some advantages of the
closely coordinated RIR/NOG operations in other regions, and I would like
to see them closer together here. Unfortunately, it is not clear to me
that the examples above are in charter for either NANOG or ARIN. I'd be
happy to re-charter either, but that's probably a topic for NANOG-futures.

or pick another key service.

DNS? DNSsec?
Security?

randy

Lee

Yeah, RIPE NCC is definitely much cheaper for PI: no initial
registration fee of ≥$500. The maintenance cost is $100/year vs
€100/year (±$137) so there is a little difference there. The $37

€50 per PI assignment from the ripe ncc, no?
http://www.ripe.net/ripe/docs/ripe-591

guys, you are following an arin policy weenie's red herring. this was
not about fees. it was about arin's board being it's own governance
review committee and having no term limits, arin forcing folk to sign
contracts with clauses saying arin can change the Ts&Cs unilaterally and
arbitrarily, ...

randy

Exactly! -- Jay