RIPE 64

Chair: Sander Steffann, Gert Doering
Scribe: Ann Barcomb (RIPE NCC)

• Session I - Wednesday, 18 April, 09:00 - 10:30
• Session II - Thursday, 19 April, 09:00 - 10:30

Session I - Wednesday, 18 April, 09:00 - 10:30

A. Administrative Matters

• Welcome
• Thanking the scribe
• Approving the minutes from RIPE 63
• Agenda bashing

Gert Doering (WG chair) opened the session and introduced the APWG co-Chairs.The RIPE 63 APWG draft minutes were approved and the WG chair thanked the scribe. Gert then presented the agenda.

There were no further changes to the agenda.

B. Current Policy Topics

- Emilio Madaio (RIPE NCC)

Emilio Madaio gave an overview on the common policy topics being discussed in all regions and a summary of the policy proposals in the RIPE NCC service region.He also listed and gave an update on Policy Development Office (PDO) activities such as the Cosmetic Surgery Project.

The presentation is available for download at: https://ripe64.ripe.net/presentations/128-CurrPolTop-RIPE64.pdf

Wilfried Woeber (Univie-ACOnet) asked if there were any efforts to make inter-RIR transfer policies mutually compatible.

Emilio Madaio thanked Wilfried Woeber for bringing up the point and stated that the issue would be addressed later.

D. Feedback From NCC Registration Services

- Alex le Heux (RIPE NCC)

Alex le Heux (RIPE NCC) gave a report of the RIPE NCC's response on action points from RIPE 63.The questions being addressed were whether an inbound transfer policy is needed, and why LIRs might need multiple routable IPv6 blocks.

The presentation is available for download at: https://ripe64.ripe.net/presentations/131-RIPE64_APWG.pdf

No questions were asked.

F. Discussion of open policy proposals

Gert Doering gave an overview on how the Address Policy session works: discussions take place during the working group, but final decisions take place on the mailing list.He reminded attendees that the session was being recorded and that people were participating remotely via chat.

• 2011-04 Extension of the Minimum Size for IPv6 Initial Allocations (current status, Last Call) - Gert Doering (WG Chair)

Gert Doering gave an update.He stated that the proposal was in Last Call and the deadline is 2 May.

Aaron Hughes (6connect) asked why the proposal used /29 as a limit instead of /28.Gert Doering stated that this was due to the fact that early allocations by the RIPE NCC used /29, and this number was chosen so that new allocations would be similar to existing allocations.

• 2011-05 Safeguarding Future IXPs with IPv4 Space (current status, Last Call) - Gert Doering (WG Chair)

Gert Doering gave an update.He stated that the proposal was in Last Call and the deadline is 9 May.

The presentation covering 2011-04 and 2011-05 is available for download at: https://ripe64.ripe.net/presentations/123-wg.pdf

No questions were asked.

• 2012-01 Inter-RIR IPv4 Address Transfers

Sandra Brown (IPv4 Market Group) gave an initial presentation on the status of proposal 2012-01.In response to the feedback that has been received, she offered the following way forward: withdrawal of 2012-01 and introduction of proposals on Inter-RIR transfers (likely to become 2012-02) and timeline amendments for all transfers (likely to become 2012-03).

The presentation is available for download at: https://ripe64.ripe.net/presentations/95-RIPE-2012-01_Preso.pdf

Gert Doering suggested that any questions pertaining to the discussion be saved until all presentations on the topic had been given. He invited questions to clarify Sandra Brown's presentation.

No questions were asked.

Remco van Mook (Equinix) gave a summary of inter-RIR transfers (likely to become 2012-02).

The presentation is available for download at: https://ripe64.ripe.net/presentations/113-inter-rir_transfers.pdf

Geoff Huston (APNIC) requested clarification on whether the proposal covered the policies of the originating region as well as the policies of the receiving RIR.

Remco identified the rule that the transfer must be in line with the originating policy as addressing this concern.

Geoff Huston remarked that the wording was less clear and suggested changes to the text.

Paul Wilson (APNIC) described the proposal as similar to the related APNIC policy. He expressed the view that they would interact as intended, and that this was an example of the sort of coordinated policies Wilfried Woeber had asked about earlier.

Randy Whitney (Verizon) asked about specifying the need-basis requirements more clearly. He added that in the ARIN region this was an important criterion.

Remco van Mook replied that the need basis requirement was a reference to the transfers within the region. He wanted to keep the proposal simple and in sync as much as possible with just a reference, so that if the need-basis will be dropped in the transfer within region, it will also go away in the inter-RIR transfers.

Sander Steffann added that if they put more need-basis constraints in the inter-RIR transfers and than the intra-RIR transfer changes, there would be a bigger mess.

Randy stated that ARIN could have a problem in accepting a transfer with a region that does not have explicitly the need basis in its transfer policy.

Remco responded that he would be more than happy to explain this to ARIN.

Paul Wilson clarified that there is a similar situation in APNIC: they have the need-basis in their inter-RIR transfer policy and recently added the need-basis also in the intra-RIR transfers. As per the ARIN requirement it is intended to allow transfers only when there is a compatible need-basis address management system in the recipient RIR. He concluded saying that APNIC satisfies this requirement.

Remco van Mook then briefly described the unintended side-effects on inter-RIR transfers stemming from the intra-RIR transfer proposal, which would require amending the Inter-RIR Transfer Policy (likely to become 2012-03).

The presentation is available for download at: https://ripe64.ripe.net/presentations/114-Updated_transfer_policy.pdf

Niall O'Reilly summed up what was stated: run-out-fairly would only apply to additional allocations, whereas transfers would be handled by the classic rules.

Remco confirmed.

Niall supported the proposal's scope.

Andy Davidson (LONAP/Hurricane Electric) seconded Niall.

Gert Doering opened the floor to discussion of the roadmap put forward in Sandra Brown's presentation.

Nina Bargisen (TDC A/S) expressed opposition to having different requirements for purchase as opposed to new allocation, which she described as promoting the trade of IP addresses. She asked that if the policy is accepted it include the limitation that this would apply only after we ran out completely of IPv4 allocation and not as long as people have the choice between transfer and additional allocation.

Sander noted that from a procedural point of view, very likely when and if the proposal will be accepted there won't be difference between the two options.

Daniel Karrenberg (RIPE NCC) wanted to know if "run out completely" included the last /8.He did not think that was the intention, but asked that this be clarified if text is added to the proposal.

Gert Doering remarked that the current text that references the additional allocation policy means that the transfer policy will become ineffective after the last /8 is reached.

Nigel Titley (Chairman of the RIPE NCC Executive Board) described the process as "rearranging the deck chairs on the Titanic" for the benefit of people with financial interests in IPv4 transfer.

Eliot Lear, from remote participation, was concerned that the proposal would place the burden of IPv6 transition on small businesses.

Remco van Mook disagreed that this would be a consequence of the proposal.

Carlos Martinez-Cagnazzo (LACNIC) noted that no inter-RIR transfer proposal has thus been put forward in the LACNIC region. He also asked for a clarification: the Inter-RIR Transfer Proposal refers to the Intra-RIR Transfer Proposal, whereas the text of the Intra-RIR Transfer Proposal explicitly states that address space may only be re-allocated to another LIR which is a RIPE NCC member.

Remco van Mook clarified that this was due to the fact that the current transfer policy was an intra-RIR transfer policy, and the intention was to slightly adapt it rather than to completely revisit the entire transfer policy.

Nick Hilliard (INEX) responded to Nigel Titley's comment. He commented that this effort could be described as concerning the allocation of life jackets on the Titanic because it is not feasible to run IPv6-only services.He stated that he liked the proposal and saw it as trying to facilitate a functional market. He also said that as a technical person he was not qualified to comment on the economics, but that the text did create a situation with very few market constraints.The possibility of price fixing and hoarding concerned him.

Remco van Mook noted that he had only described the part of the transfer policy that he considered limiting, rather than the entire transfer policy.

Sander Steffann asked if Nick Hilliard's comment concerned the existing transfer policy or the proposal.

Nick Hilliard described it as a general concern about technical people attempting to define the terms of an economic market.

Gert Doering returned the discussion to Sandra Brown's presentation.He polled the audience to determine the general feeling about going forward with dropping 2012-01 and continuing with the two proposals Remco van Mook had described, noting that this was not a question of support for the proposals, only the approach.From his perspective as chair, he did not see a problem with following the proposed path.

Gert Doering thanked the audience for attending and closed the session.

Session II - Thursday, 19 April, 09:00 - 10:30

Gert Doering (WG chair) opened the session and went over the agenda for the day.

The presentation is available for download at: https://ripe64.ripe.net/presentations/176-wg3.pdf

E. What Is Consensus? Lessons learned from past policy proposals

- Sander Steffann (WG chair)

Sander Steffann described the process by which consensus is determined within the working group.He stressed that consensus is not the same as unanimity, and explained that the collective of working group chairs provides a check that the procedure was correctly followed.

The presentation is available for download at: https://ripe64.ripe.net/presentations/140-RIPE64_-_What_is_consensus.pdf

Piotr Strzyzewski (Silesian University of Technology)remarked that he liked the current approach of judging whether something qualifies as an argument.

Janos Zsako (ISZT) stated that it was beneficial to discuss this issue.

Sander Steffann thanked him for the feedback.

Peter Koch asked why "I don't like this" gets less weight than "I like this."

Sander Steffann replied that the statement "I like this" refers to the backing documents that you approve of, whereas "I don't like this" provides no information to enable people to improve the proposal.

Gert Doering added that specific arguments also allow people to follow the line of thought.

Peter Koch said that he believed the interpretation of the process was correct, but that the process is biased toward change. This, he noted, has the benefit of overcoming inertia but provides no safety belt.

Wilfried Woeber (Univie-ACOnet) responded that it was good that this was being discussed, but that he saw no need for absolute symmetry in responses because both positions required evidence.

Ruediger Volk (Deutsche Telekom Technik) noted that if approval of a proposal was seen as implicitly endorsing the arguments in favour of the proposal, it is necessary to ensure that these arguments are confirmed and clear.

Kurtis Lindqvist (Netnod) thanked Sander Steffann for the presentation and for clarifying what consensus means.He remarked that change should be positively supported and that it should be possible for someone to just express that a proposal is bad policy without having detailed counter-arguments. He added that the process should find a work around to the real problem of too few active participants.

Nurani Nimpuno (Netnod) noted that this discussion takes place periodically, but is nonetheless worthwhile because new people enter the community. She stated that Peter Koch's points are valid, but that weight is not the main point with consensus. She said that the Chairs play an important role in pushing people to expand on their positions because the purpose of the process is to bring the discussion forward rather than to create deadlocks.

Matija Grabnar (go6 Expert Council) remarked that the situation is balanced because only the first person opposed to a proposal needs to provide arguments and subsequent objections can be based on this argument.

K. IPv4 Maintenance Policy

- Rob Blokzijl (RIPE Chair)

Rob Blokzijl suggested restructuring IPv4 policies to create a single document which would outlive exhaustion.

The presentation is available for download at: https://ripe64.ripe.net/presentations/164-IPv4_Maintenance_Policy_v1.pdf

Gert Doering thanked Rob Blokzijl for his effort, and Rob Blokzijl thanked

RIPE NCC staff who assisted him.

Niall O'Reilly thanked Rob Blokzijl for leading the initiative. He stated that this task should have been undertaken earlier and there is a lot of work to be done.

Rob Blokzijl described his plan of one of improving documentation rather than creating a new policy.

Nick Hilliard (INEX) said that consolidation would simplify things. He expressed concern that the summary document removed some of the IPv4 allocation and assignment policy history and said that it was important that the document does not change policies.

Rob Blokzijl responded that he had sometimes let his personal opinion drift into the text, and that he would step back to ensure that all accepted current policies would be completely included.He noted that this would reveal existing inconsistencies.

Ruediger Volk agreed that the intentions were good but he felt that the approach would not add clarity.He called for analysis of what policy or administrative questions haven't been addressed.

Rob Blokzijl clarified that he is not writing a summary but compiling a collection of existing policies.

Ruediger Volk then asked for paragraph references, which Rob Blokzijl agreed could be added to the working document.

Daniel Karrenberg stated that things will be substantially different when the available IPv4 pool runs out, and that existing policies will no longer make sense.He proposed looking at the goals, and then making new policies to meet those goals: a radical fresh start rather than building on the existing system.

Gert Doering requested that the discussion be stopped in order to leave time for the remaining agenda items.

Ruediger Volk requested clarification from Daniel Karrenberg.

Daniel Karrenberg confirmed that he believes that collecting historic information can be a contribution but it will not be the solution.

Rob Blokzijl stated that his view is similar to Daniel Karrenberg's and that he wants to focus on the role of the registry because it will be relevant for many years to come.

Niall O'Reilly agreed that the focus should be on the registry.

Gert Doering advised everyone to read the document and continue the discussion on the mailing list.

L. On IPv6 PI/PA Unification- update, discussion

- Gert Doering (WG chair)

Gert Doering gave an update on the draft document he and Sander Steffann had written on building a unified IPv6 policy.The formal document was delayed because there were active proposals that affected IPv6 policies that he wanted resolved before he continued.Now that this has happened, he will resume work on the proposal.

The slides are part of an earlier presentation and are available for download at: https://ripe64.ripe.net/presentations/176-wg3.pdf

M. Discussion of open policy proposals, part II

There were no further open policy proposals to be discussed.

N. Revisit IPv6 "additional allocation" policy

- Discussion led by Jan Zorz (go6)

Jan Zorz started a discussion on the HD ratio mechanism by asking who in the audience understood it.There were only a few hands raised.

The presentation is available for download at: https://ripe64.ripe.net/presentations/188-zorz-HD_discussion.pdf

Ruediger Volk (Deutsche Telekom Technik) asked if understanding the ratio was important, so long as there was a formula that provides consistency.

Jan Zorz showed an example case and then posed the question of whether the HD ratio should be less restrictive, and if it should exist at all.

Geoff Huston (APNIC), a co-author of the HD ratio revision explained how the ratio was arrived at and also pointed out how the calculations should be done. In this regard, a subnet dedicated to transition technologies like 6rd is 100% in use for the HD ratio.

Sander Steffann pointed out that "in use" only refers to what is given to end sites.

Geoff Huston responded that if a prefix is used in transition it cannot be used for anything else, so the policy that defines "in use" is broken, but the HD ratio is not.

Randy Bush (IIJ) said that he appreciated the socks handed out by the RIPE NCC, and proposed that slide rules could be given out at the next RIPE meeting.

Aaron Hughes (6connect) noted that the issue had also come up in the ARIN region.He said that, in the early days, it should be easy for people to get additional allocations because lack of experience can lead to poor planning.In the long run the HD ratio makes sense.

Gert Doering said that 2011-04 allows any /32 holder to ask for a /31 or /29 without evidence--the restriction only came into play when the holder requested still more space.

Ragnar Anfinsen (Altibox AS) described making a common mistake of requesting a /32 and then discovering that it was too small. He asked for more leeway for early adopters.

Gert Doering said that the math is complex but that the table will display the answer.He said he didn't see anyone arguing that the HD ratio needed to be disposed of, or set to another limit.

Y. Open Policy Hour

Nick Hilliard (INEX) introduced his suggestion for PI space after IPv4 run-out.He described the current policy as biased toward RIPE members and LIRs and said he had submitted text to the working group chairs, which would soon be sent to the mailing list.The basic premise of the document is "maximum of /24 per End User" and otherwise it is the same in principle as the last /8 policy for LIRs.

Z. AOB

There was no additional business.

Gert Doering closed the session by thanking the audience for attending and inviting them to the next RIPE meeting in Amsterdam.