RIPE 46

RIPE Meeting: 46
Working Group: Address Policy
Status: Final
Revision Number: 2


Chair: Hans Petter Holen, Gert Doering, Andrea Borgato

Scribe Ingrid Wijte (RIPE NCC)
Wednesday 3 September 2003 09.00-12.30

Welcome word by Hans Petter Holen. Short introduction and reminder that the last time this WG had a different name. Short explanation of the difference between the two WGs.

Introduction of the 2 other chairs.

A. Administrativia

- Scribe
- List of Participants
- Circulate attendance sheet
- Agenda

- RIPE 45 Minutes

Some confusion as to whether the minutes were published. They were on the web site but had not been distributed by e-mail. They will be re-circulated to the  list.

- Actions

Andrea Borgato (AB) went through the open actions from RIPE 45
The updated list is available at from the RIPE NCC website.

B. Report from RIPE NCC Registration Services (Leo Vegoda)

Q&A? none

C. Internet Resource Status Report (Leo Vegoda)

HPH: Q&A? none

D. ICANN ASO Address Council Report (Sabine Jaume-Rajaonia (SJR)

She has a proposal to set up an open meeting at 17.30 for anybody interested in an open discussion?

Bernard Tuy (BT): Could be useful to know what you are doing or discussing?

SJR: Policy, what will be changed

HPH: interesting question, things are published, not a lot of input. Anyway, a room has been booked for an informal talk: St. Johns I at 17.30

RB: One comment on the slides, not RIPE NCC but RIPE after the names of the members of address council, it is open for everybody not just RIPE NCC Second comment, the address council has one important function which is to elect one ICANN borad member. For the rest do no harm.

HPH: Are we doing any good?

RB: Should be admired by their courage to take this position, as it is an invention of ICANN. ICANN is negotiating a new way of working with the RIRs; one of the outcomes will possibly be a revised Address Council.

HPH: Other questions? No. Meeting: we will be there in St. John's I at 17.30.

E. Presentation of ASO Address Council Candidates

HPH : Opportunity to present themselves to the community before the election in plenary WG. Only one candidate: SJR for re-election

F. Address Policy Charter

HPH: Feedback on the latest version of the Charter for this WG. Has anyone read it? The proposal is on the web:

"The Address Policy Working Group develops policies relating  to the management and registration of internet addresses and  routing identifiers (currently IPv4, IPv6 and ASN) by the  RIPE NCC and the LIRs within the RIPE NCC service Region.

"The WG co-ordinates its work with the appropriate bodies of the other RIRs.

"The Address Policy WG meets at RIPE meetings and has an open and publicly archives mailing list.

"Anyone with an interest in Internet numbering issues is  welcome to observe participate and contribute to the WG."

HPH: Do we agree with it?

ALL: Agreed. Will be updated on the website.

G. RIPE-152 Charging by local registries (HPH)

HPH: Document was developed a long time ago. Everyone should read the document. Relevant discussion were raised recently. Do we need to update the document If anyone wants to have the document revised contact HPH and we will set up a task force.

LV: The document says: "while registries may charge for their administrative and technical services, they may not charge for name  space or address space as such; no unit cost or price tag can be  attached to a domain name or to an IP address, public or private."  Do people still agree with this. Should domains be removed? Should  ASNs be added?

HPH: Should we remove the reference to domain names in this document?

Speaker: But remember reverse domains.

HPH: At least domain names should be removed. Is it ok to charge 5 euro extra for static IP?

GD: It's OK to offer different classes of product with static or  dynamic IPs. Just not by number of IPs. That's to stop someone  from buying a "Class B" address even when they don't need one.

HPH: and if I charge by PCs connected?

GD: It

Bill Woodcock: Sorry to add a note of levity but if someone can get  away with charging for private addres space - shouldn't we let  them? It's public address space that ought to be free.

Speaker: The document as it stands seems to say that we should subsidise domains in .com for any customer that wants one because  we are not allowed to charge for it. It costs us money so we need to charge the customers.

HPH: We clearly need to update the document. How can we update gather  a small editorial team to do that?

LV: Call to the list for editorial team to update the document.

Action Point: Call to the list for an editorial team to update the doc.

I. RIR-IANA relationship and procedures (Axel Pawlik)

Questions:

GD: Can you give us an update on what's been happening on  the IANA->RIR distribution in IPv6? How this is working?

AP: Not much has happened but we see a need for a similar doc for IPv6 (change of procedure).

HPH: Why different procedures for IPv6?

AP: Not very different, only numbers will be different.

GD: Hopefully more than just numbers will be different. The procedure will have to be different as well. We want allocations to last for 10 years. We don't want small blocks and fragmentation.

HPH: Repeat question. Why not do that for IPv4 as well?

AP: Good question.

HPH: Relationship between this new policy and RFC-2050?

AP: RFC-2050 needs to be updated

Raul Echeverria (RE): Opinion of LANIC: good policy, it clarifies criteria under which allocations will be made. Good also for conservation (not possible to stockpile address space for 10 years).

HPH: What should we do with RFC-2050 revision process? Should the task force continue?

Ray Plzak (RP): RFC-2050 does not deal with IANA/RIR relationship.

Currently applying the HD-ratio to IPv4: is being discussed in ARIN region. RFC-2050 sleeping group, better served to deal with current policy. Maybe in due time to make a comparison of RIRs to see the foundation points of RFC-2050

HPH: Will think about it and possibly make proposal. Perhaps the AP WG should be more involved

AP: Reminder of AGM!!! There are just a 100 registered out of 3000+ members.

*** Coffee break ***

HPH: Are we ready to start again? Anybody here that did not sign the attendee list yet? Will go round again. Next item.

H. Policy Development Process

HPH: When the changeover from classless/classful change over was made RIPE NCC has done lot of work on updating the doc etc. There is still no formal process on how to make policy. What do the other RIRs do? Do we want this for the RIPE NCC? So, do we need another process, does it need formalizing?

DFK: Speaking as a private internet citizen. The question is not so
much the process. Before we had a small group of ISPs who use
address space etc. They decided that it had to be distributed
fairly and they hired people to do this. Now 10 years later
there are many more ISPs and a need to serve them and fairly.
This has led to a bureaucracy. Which led to more and more a 'us
against them 'atmosphere. I am getting very cynical with all this
because it will continue to go this way. So, although very well
meaning, these bureaucrats are not ISPs. The ISPs complain but
do not come forward to do something about it. They are not helping
with the policy making process

HPH: Indeed, new policy document was circulated on the mailing list.
How many comments did we get???

LV: Only comments on the status attributes section. There was nothing
on the other parts of the doc.

HPH: So we have customers bitching about the policy process they
don't contribute to?

DFK: The complaints will continue to grow, the bureaucracy will be the
same, possibly other organizations stepping in. So, on a more
positive, we need to have only 5 or 6 people writing and
integrating these things, making the docs, must be from ISPs,
every year different people.

HPH: Agreed, that is what we tried last time, nothing happened.

DFK: Yep, but maybe people are afraid that they are not qualified. But
anyone that is an ISP is qualified. We need to encourage these
people because if we do not solve this problem it will kill us.

HPH: Tried with last PI doc, no comments/criticism until the doc was
published after help of the bureaucrats. Then comments arose.

JD: What is wrong with bureaucrats coming up with the text for the
policy? As long as there are minutes for every meeting, and all
is out in the open? The PI TF worked quite well. The problem is
that the PI problem is too complicated.

HPH: Latest policy document revision, see what happened, it is been on
the mailing list, hardly comments, so do people agree or not, we
do not know.

LV: Draft policy: one thing that is important is to know that we
updated the text, not made changes to the policy, the way it was
written down caused problems, policy was not changed.

HPH: That is good, that is what we need, but we also need people to
say if they agree or not. We need this involvement. Otherwise we
need 5 or 6 people to sit down and discuss it.

GD: One of the dangers in rewriting something that was unclear is
that you have to decide what was intended by the original text.
It is possible that the change could make something that is very
clear but not what was originally intended.

KL: Look at the contribution to the NCC Services WG mailing list and
the session yesterday. contribution to list. My interpretation is
that people think the RIPE NCC should do these things and they're
doing just fine.

DFK: Agreed but it worries me how things go. Six years ago policies
were written down, no revision since then. Six years is a long
time in Internet world. For example 6 yrs ago, high emphasis on
conservation, two conflicting goals: conservation -vs- aggregation,
if we haven't evolved from that, something is wrong.

WW: Two problems here:

1. we try to do micro management on all levels, from policies
down to procedures

2. while it is good to write everything down, too much will
lead to people back down, too much to read etc. The more you
will formalize things, the less will happen, less participation.
I would like concise effort to slim down and make things simpler.
e.g. new forms, could you give us a jpg showing etc, this is micro
management.

HPH: I was part of this, if it is in the policy, then it should be in
the procedure, thus in the form.

WW: But now we need to step back and rethink all this. We now have
IPv6.

DFK: You cannot blame bureaucrats to implement the policy. Therefore policies need to change. Someone else should take a turn, it does not have to be the same people every time. Let's have some people sit down this week and discuss

LV: It is possible to change the form, if people want us to do, we
can change it to "who are you; how much do you want?" but: you
need to tell us what you want. We need to be told what to do;
otherwise we'll just carry on doing what we've been asked to do.
Until we're asked to do something different

KL: The policy is not getting better than level of engagement

Speaker: the obvious path is RIPE NCC should go on writing policy, and
they are good at it. If they don't there is no guarantee someone
else will step in.

HPH: Good points

Neil O'Reilly: 'ships passing in the night' effect. RIPE NCC is always
at the leading edge of policy. LIRs only step in and out as they
choose. The RIPE NCC should more look at the responses they get,
which parts are not liked, should go off.

HPH: Indeed the latest forms are the result of this. We told them to
check the actual need but never said to which level do we want
them to do this.

DFK: The bureaucrats need to be checked and that is missing. The
bureaucrats have all the time of the world there must be a
pushback, this should come from the other side. The vision to
change policy

HPH: Maybe we need a council that reads through proposal to see if it
is good or not.

DFK: There is a lack of input, or people have too much trust in RIPE
NCC

HPH: Again, a council do we need this?

NOR: At one level there is a question of how the policy is interpreted
(by RIPE NCC) The problem is how to define/change policies. The
RIPE NCC should prompt the community to give feedback.

DFK: We miss the operational experience of the customers

LV: Impression there is a concern that the policy themselves are
wrong, we do try to get people involved, look at the docs, read
it, comment. And let us know if not, we will go along. So if you
do not like procedures, look at policy we base them on.

JD: I would like to participate on writing policies, but no time for
it, so you delegate, like ARIN, they have a group of people doing
this, volunteers that are elected.

Mark McFadden: Advisory council: they are facilitators and advisers to
Board of Trustees.

JD: Would it be a good idea to place trust in a group of people to do
this?

HPH: Maybe that will be a good idea, form a group that reads the draft
before it becomes policy, and advise. Not too much of an extra
step. A change should at least be agreed upon by this group.

DFK: Specify more: get volunteers to read and stir up the discussion.
And then get a vote.

HPH: We need fresh faces here as well as old hands. It should be made
more clear, when there is a policy proposal ongoing, in the action
points maybe .So it is easy to see what is under discussion. Specify
the agenda, these are small things to do without adding to the
bureaucracy.

LV: If the RIPE NCC staff were to send the draft IPv4 policy doc
again to the list, will people actually respond?

HPH: Good idea! Let's do that.

LV: We will update it and resend it and remember that you promised to
read it.

J. PI Address Policy & Initial IPv4 allocation size (Gert Doring)

Q&A:

HPH: Thank you. Which direction do we want to go? Get rid of PI, or
change the way we do this? How do we keep track of whom uses the
PI.

GD: If we make multihomed a requisite, linking it to multihoming is
dangerous because it puts stress on the AS number assignment policy.

Anne Lord (AL): In APNIC we have a small multyi-homing assignment policy.
They need to be multi-homed or about to become multi-homed. Anybody
can request an assignment of any size, either as a member or as a
non-member: one time fee and small maintenance fee (10% or original
fee is charged as an annual maintenance fee).

GD: There is a fee tied to it.

AL: Yes, there is a need, so there is a need for a service, we used
to do similar to the RIPE NCC but that way you do not know what
happens to the IPs, since introduction of policy not a big rush
to request for this.

GD: How is the size of the multihomed block determined? And what are
the criteria to get one?

AL: You give info, either of your upstreams at moment or which ones
they will be. Same criteria as normal, show you will use this %
of it within so much time.

GD: So, same no minimal size, so no guaranteed routability

AL: Indeed

KL: Like proposal, was on the PI taskforce. On the PA proposal, PI
is very popular, no easy solution Proposal is good. IMO only
difference between PI and PA is whether or not you have to be
member of RIPE NCC. So, not to feel the pain of requesting IPs.
So I think make membership obligatory.

WW: Kurtis said most of what I wanted to say. Technically there is no
difference between PI and PA. Just the label, so why add
multihoming as a requisite for PI, it is not for PA. As this is
just on the moment of requesting, contracts can be cancelled
after obtaining PI. So to simplify, you should be a member for PI
as well. Which means taking the administrative responsibility as
well.

GD: OK, maybe viable approach, as we are reducing the blocks. IMO
good idea to have cost connected to PI.

HPH: Maybe even smaller block than /21, if you need to be member.
Then do away with PI-PA altogether, as the distinction is
portable/non-portable and whether you can further assign or not.

GD: Maybe we should forget about worrying about conserving IPs,
radical but....

LV: One time service fee, must be presented to AGM for approval.
Keeping track of assignments is an issue. May want to use the
proposed org attribute for this. Or doing away with distinction,
radical change to policy, implies no need for status attribute any
more.

HPH: Outcome of this could be some people sitting down and make a
proposal.

Geoff Huston (GH): This is not radical. It's a replay, of IPv4 of 'one
time forever' allocations, the amount of dead space has gone up.
We do not know what has happened to it. Once and forever allocation
is not good as it is a replay of the past.

GD: The APNIC policy does not have this

GH: Indeed

HPH: Almost time for lunch.

K. Final Revised IPv4 policy

HPH: Already been dealt with: revised doc will be published again, new
deadline for comments.

X. AOB

GD: .de domain name people. For nameservers they need IP. Considering
to use Anycast. So they would like revise to PI policy to get a
small block of unicast IP for this.

JD: So no guarantee that this will be routed very far, will be
filtered.

GD: Agreed but we need to have some special policy for this. Tricky,
finding blocks that will be routed.

Ww: Before proposing changes, talk to the routing guys, they should
maybe implement some changes to the way it is filtered. Possibly
make the RIR maintain an updated list of small blocks that are
given out and should not be filtered out.

GD: Finding space should not be such a big problem for RIPE NCC.

JD: It was the routing guys that asked for that info.

Speaker: There is no such thing as golden networks, the routing guys
may change their minds. If you want to remain neutral do not
worry about the routing, focus on the need and use.

HPH: OK

Y: Open actions:

1. (NCC) Call on the AP WG list for an Editorial Committee to move
forward with a rewrite of ripe-152
2 (NCC) Update IPv4 policy draft with suggested status attribute changes.
Resubmit to AP WG list for two weeks for comments.
3. (Chairs) Clean up action list.