This archive is retained to ensure existing URLs remain functional. It will not contain any emails sent to this mailing list after July 1, 2024. For all messages, including those sent before and after this date, please visit the new location of the archive at https://mailman.ripe.net/archives/list/address-policy-wg@ripe.net/
[address-policy-wg] RIPE 56: Address Policy WG minutes (DRAFT)
- Previous message (by thread): [address-policy-wg] 2008-03 Proposal Accepted (Global Policy for the Allocation of the Remaining IPv4 Address Space)
- Next message (by thread): [address-policy-wg] Agenda items for RIPE 57 in Dubai?
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Gert Doering
gert at space.net
Fri Sep 12 16:10:48 CEST 2008
Hi fellow APWG members, Susannah and Alex at the RIPE NCC have done a great job in writing the minutes for the last APWG meeting (at the RIPE meeting in Berlin). Below you'll find the *draft* minutes. Please send me your comments, if needed, so we can amend potential mistakes and declare the minutes final at the next RIPE meeting. regards, Gert Doering, APWG chair ----- Forwarded message from Susannah Gray <susannah at ripe.net> ----- RIPE 56 Address Poilicy Working Group Minutes Wednesday 7 June, 11:00-12:30 Thursday 8 June, 09:00-11:30 Chair: Gert Doering Scribe:Alex le Heux (RIPE NCC) Jabber: Kjell Leknes (RIPE NCC) Agenda ----------- A. Administrative Matters (selecting a scribe, approving the minutes, etc.) Approve Minutes ----------------------- Minutes from RIPE 55 were approved. There were no comments. Agenda Bashing There were no changes to the agenda. B. Overview of concluded proposals B.1 2007-07, "End Policy for IANA IPv4 allocations to RIRs" Withdrawn, replaced by 2008-03 B.2 2007-06, "Global Policy for the Allocation of the Remaining IPv4" Withdrawn, replaced by 2008-03 B.3 2005-08, "Proposal to Amend the IPv6 Assignment an Utilisatio" Change of /48 to "LIR's choice", /56, and HD-Ratio Accepted B.4 2007-03, "IPv4 Countdown Policy" Withdrawn, replaced by 2007-06/2007-07, replaced by 2008-03 B.5 2007-04, "IANA Policy for Allocation of ASN Blocks to RIRs" Accepted, currently in ASO AC review Gert asked an ASO AC representative to give an update. Wilfried Woeber (Address Council) said that the fact that the Policy Development Process (PDP) has been followed for global policies should be documented. He explained that this report has been constructed and was sent at end of April under the leadership of Fernando Obispo and that it is now in the editorial queue at the NRO, so we need to make sure the wording is compatible with the MoU. This should be done in a few days. He added that the content is done, now it needs editing and the chair will formally send it to ICANN. This will occur in a matter of days. C. New Proposals since RIPE 55 C.1 2008-0, "Assigning IPv6 PI to Every Inetnum Holder" [Presentation Url?] Lutz Donnerhacke Michael Dillon (BT) said that he was concerned that this could damage the the routing table by encouraging a large number of PI anouncements that otherwise would not be there. He added that no one should get addresses unless they agree to accept them, so database entries should not be made without some communications. When you apply for addresses the agreement is there, as an application has been submitted. There should be a mechanism to communicate with the end-user to make sure they understand. Lutz said that admin contacts could be e-mailed. Nick Hilliard (Inex) said that he was unclear on what problem is going to be solved. The current mechanism for LIR allocations is easy: send e-mail to hostmaster at ripe.net and ask, and they will allocate you a range. From this point of view proposal 2008-02 is pointless. Gert Doering stated that this is a discussion about 2008-01, not 2008-02. Nick continued that he agreed with Michael. There is no contract here. There is no indication that the end-user even wants it. He added that this will conflict with proposal 2007-01, which implements a payment system. He suggests that if IP addresses are handed to to end-users and then money demanded, it will put the RIPE NCC in a difficult position. Bill Manning (ARIN) said that Trudy Prins (RIPE NCC) and Randy Bush (IIJ) both talked about address ownership. He said that he saw the RIPE NCC handing property to people, property that becomes taxable. He said that he didn't want the liability of getting property that he doesn't want. Forcing this on people is something that the RIPE NCC shouldn't do. Lutz said there is PI space for IPv4, which can be consisdered as property. It's the same thing today. Marco Hogewonging (XS4ALL) asked if it had considered that auto assign interferes with IPv6 deployment? If we assign lots of blocks to everyone without people asking for it, we will end up with 0.0001% being visible and routed. Lutz said that it was already suggested to ask first, which is a good idea. Raymond Jetten (Elisa Oyj) asked about changing the proposal in this way: Give out IPv6 PI to the ones that have IPv4 PI? Lutz said that he does not like PI space at all and that we should tell customers to give up PI for PA space. He added that he did not want to give advantage to current IPv4 PI space holders. Raymond said that maybe people could get old PI back. Peter Koch (DENIC) said that at DENIC is a last resort registry and that he is acting as 'chief historian', guiding innocent colleagues though inetnums from the 80s. He continued that he appreciated the optimism about the accuracy of RIPE Database, but does not share it. There are a lot of errors in there. There might be a lot of dirty work to do. Proposal 2008-01 is thus useless from practical perspective. He added that it is being proposed to apply 1989 eligibility criteria of PI space to IPv6 PI space. Much of the old stuff isn't routed today and hasn't been for a long time. We should not make the same mistakes again that were made 18 years ago. Wilfried Woeber said that he did the last resort registry for Albania. He said he hasn't made up his mind about the core of the suggestion. He added that he has a problem with the argument in favour when you start pointing to headlines in the press. The PDP should not be influenced at all by anything that's put in the press. We know about quality of headlines, there is no direct relationship between that kind of publicity and the reality of the internet. James Rice (Jump Networks) said that this discussion may be more for the RIPE Routing Working Group rather than the Address Policy Working Group, but it is relevant here too. He asked if people were aware that a lot of DSL customers are registered as /30s? Did you consider 2.5M extra routes and the churn that this creates? The operational impact there sounds quite bad. Currently, with a few 100k prefixes, we see some hijacking, when you have 2.5 M there and try to keep contact information up to date, it sounds like administative nightmare. Lutz said that this is an interesting problem. He added that he did not think that many routes will appear and that he was surprised that no one in the audience is claiming that PI space is unnecessary. Gert said that no decisions are made at RIPE Meetings. The sessions are for comments from audience. There is no overwhealming consensus to go forward, so this will be discussed afterwards and we will decide how to proceed. C.2 2008-02 - Assigning IPv6 PA to Every LIR [URL?] Lutz said he would like to change this proposal, as was suggested on the mailing list, to not make assignments, but to send an e-mail to ask if they want an assignment. If they reply 'yes', they will get one. Gert said that there was a comment from Nick already about giving people things they don't ask for but will get a bill for later and that this is perhaps not a good idea. Marco Hoogewoning (XS4ALL) said that, as we can get IPv6 now, shouldn't we just abandon this and ask the RIPE NCC to mail everyone a big flyer or e-mail that the 200 rule is gone and it's much easier now instead of changing this policy? Gert said that this was an interesting plan and is proactive. Lutz said that this is not a policy change, this is for a one time operation. Marco said that what Lutz is saying is that everyone should get an e-mail to say that if you want your addresses, you can get them here. That's what current policy says, if you want your addresses, you can get them by sending an e-mail to hostmaster. Leo Vegoda (ICANN) said that he sent two mails to the list. The first asked about the billing implications. The RIPE NCC sent an e-mail. It said that everyone's count goes up, so there won't be an impact on billing. It's all proportional. The second is about the RIPE NCC inviting people to get an allocation, those that take them up then get one. That has billing implications. It's sort of like actively selling address space. Do we want to run that system this way? Try to make sales? Gert said that this is a good argument. We are caught between promoting and not promoting. Kurtis Lindqvist (Netnod) added that he thought people want IP address space because of need. There seems to be a slow uptake. He said that he failed to see what is gained by banging people on the head with an IPv6 block. People who haven't deployed IPv6 don't need it. The theshhold in the portal is low. Do you think removing this makes it really easier? Hans Petter Holen (Visma IT AS) asked if the policy needs to be changed in order to get more people to get addresses? There is no real barrier anymore. The question is what level of marketing should we do? He said that the group could spend all morning raising awareness. Is that raising awareness or marketing? It's the same thing. It is a good idea that we suggest to the RIPE NCC that they send more information to their members that there is IPv6 for them and how to get it. We do that at these meetings, and at trainings. Maybe direct mailings, newletters, maybe during normal contact with hostmasters. There are levels we can use here without making it marketing. James Rice (Jump Networks) added that he could sympathise with technical people at Local Internet Registries (LIRs) that cannot get permission becase they need management approval because of billing. How about making /32 free? Gert said that this is a decision for the RIPE NCC General Meeting. Rop Blokzijl (RIPE Chair) said that this is a solution, but what was the problem? The problem that is being suggested doesn't exist. They can get addresses. It all comes down to a glorified PR exercise. We should never give people what they didn't ask for. If people still think there is a high barrier, we are lacking in our information provisioning. Maybe the RIPE NCC Member Update could carry yet another article on how you get an IPv6 block. That is probably as effective as any verison of this proposal, without the dangerous side effect of handing out address blocks to people who don't want them. Gert stated that there was no consensus here either. We seem to have consensus on giving a message to the RIPE NCC to be more pro-active in telling people how to do it. C.3 2008-03 - Global Policy for the Allocation of the Remaining IPv4 Address Space IPv4 countdown policy, revisited [?] D. "Contracts and Certification session, part 1" D.1 2007-01 - Wrap up, next steps, comments from the NCC board Gert mentioned that he had some updates on the PDP. He said there was a last call that ended on Monday this week. Formally, it is now out of the hands of the AP WG. It is now in the hands of the collective WG Chairs to determine if there is consensus. It is being discussed, there is no final outcome yet and it will be discussed for a few more days. If it goes into a policy, which is likely, then the next thing is that the General Meeting will discuss it and the statutes will be changed. There is a General Meeting today, if you are RIPE NCC Member, please go there. It will come up. It won't be possible to make a decision today as this needs invitation letters. It can be discussed and voted on in Dubai. Parallel, the RIPE NCC can start planning the implementation. Michael Dillon (BT) said that BT was in favour of this proposal. He asked if the implementation could be seperated into two phases: future PI assignments immediately on contracts before all historic ones are done. Gert said that this is the plan. Rob Blokzijl |(RIPE Chair) asked if there is now a version that has these refinements in it. Nick Hilliard (Inex) said that the proposal is in final stages of Last Call. The current proposal doesn't have all the subtelties in, as those are implementation issues. The community expectation is that the RIPE NCC goes with the suggestions. Axel Pawlik (RIPE NCC) said that he agrees and that the RIPE NCC likes a challenge. This has has been one already and the Executive Board has discussed this. He added that the RIPE NCC can implement it as Nick has put it forward and it's good prectice to look at articles of association. The proposal can be implemented via Associate Membership of the RIPE NCC. ACTION ON RIPE NCC (1): Help to draft a contract framework for the LIR - End User. Because there are over 5000 LIRs, and they all have a lawyer, it's not useful to have them all start from scratch. Include things that must be in there and can be adapted to local law. Axel said that he would hesitate to provide contractual frameworks to so many different LIRs in different countries. He said he would like to provide principles. Gert said that this would be ok and something that he could give to a lawyer. Action on RIPE NCC (2): Formally bring ongoing policy issues to the General Meeting. D.2 proposal coming out of the Certification Authority (CA) TF Nigel Titley http://www.ripe.net/ripe/meetings/ripe-56/presentations/Titley-Initial_Certification_Policy_Proposal.pdf Gert said that the proposal says that an LIR can ask for a certificate for PA space. What is the plan for AS Numbers and PI space? Nigel said that this proposal only covers PA space. This proposal would either have to be extended for PI space and ASNs or a new one made. Gert said that as we will soon have a framework for End-User contracts, if a contract exists, one should be able to get certificate. Nigel said that If a contract exists, there is no reason not to have a contract. Hans Petter Holen (Visma IT AS) said that we could move ahead with this and modify it later when we have some experience. He said this shouldn't be stopped while waiting for other things. Gert said that if this is worded carefully, like "If there is a contact you can get certificate" then it would make it automatic. Antoin Verchuren (SIDN) asked what happens when there is an LIR with PI space. Nigel said that if there is a contract, there is no problem in issuing a certificate. Michael Dillon asked if the certificates are the same as the certificates used to allow/disallow BGP announcements. Nigel said they were. Michael said that if a substantial number of LIRs have these certificates, and their BGP is filtered based on certificates, and if an LIR's budget for being a RIPE NCC member is held in some unimportant corner of the company and they fail to pay on time, suddenly their internet access could get shut off. Nigel said that yes, this could happen. Michael asked if this was "securing the routing system". Nigel said that is also called "getting your payment system working". Michael asked if it was appropraite to make those running the LIR request certificates? Should this be higher up in the company, say legal council, CFO? Nigel said that this would be an internal business process. Michael added that there are well run, less well run and poorly run companies. Everyone's network should function. We do not want people to disappear because someone forgot to pay their bill. One might make the legal argument that the RIPE NCC is damaging companies by allowing them to shoot themselves in the foot. There is a potential downside. Nigel said that there is always a potential downside to not paying your bills. Gert said that he saw the point. There might be a problem, but we might want to encourage timely payment. The RIPE NCC is not so strict on this. Lutz Donnerhacke (IKS GmbH Jena) said that he had the same question about unroutable entities when the certificate is broken. Yes it's intentional, the routing system should withdraw routes that are not certified. Daniel Karrenberg (RIPE NCC) said that Michael's concerns are valid. When the RIPE NCC and the Certification Task Force looked into this it came up. It is clear that a lot of education needs to be done to tell people what it is that they are getting in to. That is part of the program. Nigel is also correct. How this is handled internally in companies is an internal business matter. We can make suggestions, but it is not up to the RIPE NCC to dictate how certificates are being used. If what Lutz wants is implemented, it only happens when ISPs want it. Daniel added that his personal opinion was that, after talking to ISPs about critical resource management, ISPs will be very careful with this. I see this evolving as certificates being used for provisioning first. Only when there is confidence they will be used for stuff which is closer to real time. Hans Petter said that routing policy between ISPs is not dictated by what we do here. Looking at this from the other side, when there are no certificates, anyone can announce anything and destroy business. On the other hand, I'm quite sure that all these telcos know how to cut of their customers if they don't pay their bills. Gert said that certificates give people the ability to perform route origin checking. If it is only used for ISPs towards their customers we already have a win situation. There are some concerns about the specific implementation, but that is not really part of the policy. He added that he thought we should go forward and make it a formal policy proposal. Nigel said that that the wording could be adjusted to apply to objects other than just PA space. Thursday, 09:00-11:30 ================== Proposals 2008-01 and 2008-02 have been withdrawn by the author. E.1 2006-01, "Provider Independent IPv6 Assignment for End User Organizations" Gert Doering summarised the discussions so far. He added that comments would be collected and then Jordi Palet Martinez (Consulintel) would adjust the proposal. James Rice (Jump Networks) said that IPv6 PI space is needed for multihoming. A /48 is a good size. It should be big enough so that they don't need a second one. A larger / could be given if justified. Wilfried Woeber said that he was not speaking in favour or against. He continued that first, there are differences in regions. It is not a global proposal, just regional. In the RIPE region, if we decide not to follow the others, things will be fine. He continued that, with his Address Council hat on, if there is a feeling in the community that this should be globally coordinated, then it should be a global proposal. Otherwise, it's fine that things are done here that other regions don't do or the other way around. The other observation is that almost all restrictions in the regular policy for getting IPv6 adress space have been removed. At the LIR/ISP level there is not much difference. It shows up in the routing table. There is also a special IX policy. He said that he did not really see in the long run why three, four or five special forms of distribution would be needed. Gert said that it's not a global policy. He explained that there has been feedback on lists that having fairly major differences between regions is annoying. For example, organisations could say "If we just had our headquarters in different regions we could do business in a very different way...". He added that it could be useful to try to align policy. Gert continued that an LIR could get a /32. For some enterprises this is hard, because being an LIR is expensive. For others it is not so hard, but there is a hurdle to declaring them as an ISP when they really are not. On the other hand, a routing table slot is just a routing table slot. But also, if they need /48, why give them more? Jordi said that when he started working on this in different regions he started with a /32. Then in several regions there were comments that a /32 was too big. The final policy approved was a /48 in AfriNIC. There has been some discussion in AfriNIC, and the the staff realised that all /48s get filtered and they can't sort it out. He said when the size is decided on, there may be the risk of the prefixes not routed correctly. He proposed that a small change is made in how the proposal is used in LACNIC and then split the proposal into two proposals: one for organisations that already have IPv4 PI space so that they can get IPv6 PI space as well; and one that asks for extra justifications for those that start new and do not have PI space. Gert said that he did not like this idea and that he would prefer an IPv6 policy that is not tied to IPv4, otherwise people might use IPv4 just to get IPv6. Marc Groeneweg (SIDN) said that, as the .nl registry cannot deploy IPv6 now, he was in favour of this proposal, so that there can be independence from ISPs. Gert said that SIDN could become an LIR and get a /32. Marc said that the main objective is to get .nl running and not be an ISP or an LIR. Marco Hoogewoning (XS4ALL) said that, regarding /48s being filtered, he thought that routing or other administrative purposes should not be a reason to get a bigger block and that there should not be special cases. He said that PI space is needed anyway. The /32s of today are the "Class A" of tomorrow. Gert said that, regarding special cases, a decent PI space policy is necessary. Wilfried said that with the changes that are on the way, the distinction between being LIR and not being LIR should be forgotten in regard to IPv4 PI space with contracts. He continued that the reality would be that, if you hold resources, you must have a contract with someone, with either an LIR or the RIPE NCC. He said he did not see what's so bad about being an LIR and that in the future you may have to have a formal relationships with the RIPE NCC anyway. He continued that what this is called doesn't matter and this was discussed that in the RIPE NCC General Meeting. Being an LIR is not a big workload. The requirement to have customers is already gone, you can just get the block of adddresses. Peter Wyatt (HP) said that a /48 wouldn't be big enough for HP. He said that he could document the need. He said that that HP does not want to register as an LIR and need a documented approach between PA space and PI space. Otherwise there will be organisations that have historic space that should not be LIR. Gert said that he had seen HP's prefixes and asked if he could second that ISPs are filtering your announcements or is it mostly working? Peter said that it could be argued that no organisation would need more than a /48. Gert asked if he had seen that /48s work? Or are they filtered? Peter Wyatt said that this works right now in terms of our ARIN and APNIC allocations. He said that HP wants to try to get a /43 PI space from the RIPE NCC as well. Gordon Lennox (European Commission) said that he hears the word "enterprise" but there is another community called "governments". They have no business models and need their own IPv6 space. Arien Veijn (AMSIX) said that AMSIX only has PI space for peering LAN and not for it's services. He said that they are a neutral organisation by nature but would like to have IPv6. Becoming an LIR just to get IPv6 space doesn't feel right. He said that AMSIX gets space through an LIR as the system was designed like that. It would be no problem to have a formal relationship with the RIPE NCC or an LIR. Michael Dillon (BT) said that he has helped companies get PI space in the RIPE and ARIN regions. There really is need, in some cases, for space that's registered in a global registry, such as for financial services for example. He said that when IP addressing and LIR creation began, it was clear that there were two types of need. One type is fixed need with no growth and the other need is ISPs with constant growth and expansion. With IPv6, there may be limits to growth. He continued that things have been divided so that the LIR is the receipient of the large block and is expected to grow. The PI space holder is not expected to grow so if we change the way we do things with IPv6 so that PI space is not really the small size and a /32 is not always for ISPs, he thinks there can be a rational system with more slots for an applicant to fit into. Gert asked if banks want global unique space with their name that is not routed? Michael said that banks do transactions between themselves and that these do not happen on the public Internet. A company like HP doesn't want /48s, /49s, /50s and so on, but HP is not an LIR but it functions like an ISP. He added that there are now three sizes: a /56 for really small users, /48 for any other customer of ISP and PI space and then there is a /32 for the big organisations like universities or ISPs. He suggested this could be extended to non-LIRs, on the same terms as for the smaller organisations. Ruediger Volk (DTAG) said that he was interested in keeping an eye on the effect on routing slots. He said that large scale uncontrolled PI space will create problems in the routing system and that this should be dealt with in the policy even though it is essentially the problem of the operators. He said that he saw three conditions where it doesn't pose a problem: it doesn't show up in routing system and problems with PI space that is mandatorily renumbered every two or three years can be minimised and then cleaned it up. Gert said that regarding unbounded growth, the contracts on the way. Ruediger agreed, and said that he wanted to state that everything mentioned that threatens to keep people from signing contracts is actually already a feature. He said that he was surprised that the EU Commission or HP or organisations of that size would feel threatened by the request to enter into a contract. Those large organisations have explicit IT service organsiations and it could be argued that they have an internal ISP. He said he did not want to see the notion that PI space is available in cases where it is needed but that it is not for everyone. Martin List-Petersen (Airwire) said that, for PI space, saying that everyone could sign up as LIR doesn't solve the routing table issue. He said that for PI space, he thinks this is needed. It has something to do with IPv4 stewardship and added that if IPv6 PI space comes, he is in favour of having contracts. Gert said that this is there. Martin said that from the routing table point of view, forcing organisations to become an LIR doesn't solve aything. James said that he got the impression that HP, I got the impression wants a /44 an asked if this was because of announcing /48s from different unconnected sites. It is important that when they make their request there is space for people to say how they will route it, so that can be taken in to account, and people can make filters. If they intend to announce /48s, it must come from that block and not one that has a /32 minimum allocation size. Marco said that from a global perspective, it's already there. If we formalise this, he can filter and other people can be educated about filtering. It will give a framework for filtering. He added that a way for people to get space without being an LIR should be found. Ruediger said that prefix length filtering will not meet the security requirements of a large scale IPv6 network. Gert said that he heard no fundamental opposition to the proposal and lots of voices that say there is a benifit. He said there was some concern about getting the mechanics right and, as decisions are not made at the meeting, there is enough input to start a new round with the document. He said that the document will be sent to the mailing list and input is needed. E.2 2006-05, "PI Assignment Size" Gert Doering (Chair) explained that the proposal will be sent back to the mailing list and discussion restarted and asked if anyone objected. No one objected. F. "End of IPv4" session F.4 2008-03, "Global Policy for the Allocation of the Remaining IPv4" http://www.ripe.net/ripe/meetings/ripe-56/presentations/Gagliano-Global_Policy_Allocation_of_Remaining_IPv4_Space.pdf /8 reservations to RIRs Roque Gagliano Michael Dillon(BT) said that he thought that this was a really bad idea and it moves the point at which IPv4 becomes hard to get closer to us in time. Most people are not concerned about IPv4 exhaustion, but abbout the effects. Gert Doering (Chair) said that for big regions the financial impact is only a few months, for small regions where they really do not have money to upgrade at all, the impact is big. Michael asked if Gert was trying to sell this as a global aid plan for less developed regions? Gert said that it is not his job to sell anything and that he just wants to point out that there are arguments for and against it. Michael commented that the general perception of this is of people playing around with numbers. Yours is the first that ties it to a real world concern. Roque Gagliano (LACNIC) said that the number of ISPs that consume a lot of addresses is very small. This policy helps because it brings certainty about when the last /8 will happen. At LACNIC there are now proposals about what to do with that last /8. The other thing is, it's a strong message to those outside the RIR community that it's not business as usual but that we are doing something special. Daniel Karrenberg (speaking as private citizen) said that people should keep in mind the actual playout of this and whether you can enforce the policies that you are making. He said he had concerns that if the only substantial amounts of IPv4 are located in smaller regions, someone from a larger region may find a policy-compliant way to obtain that address space. It's easy to find a legal entity. He continued that a similar thing applies to all policies that set aside space for special uses, like dual stack. Hans Petter Holen (speaking as private citizen) said that what we are facing is the end of something. There are two extremes: either we keep policies the same and hit the wall; or we try to do the sensible thing. He said that it's hard to figure out what the right thing to do is. Unless very strict regulation is enforced, it's going to be interesting to see what everyone is going to do. Gert said that this discussion should be taken to the mailing list. F.3 2007-09, "Cooperative distribution of the end of the IPv4 free pool" http://www.ripe.net/ripe/meetings/ripe-56/presentations/Hain-IPv4CooperativeDistribution.pdf Tony Hain IPv4 Cooperative Distribution Ruediger Volk (DTAG) said that there are RIRs that have national registries and asked what does theproposal mean about the pools that are sitting there? Should the RIRs that don't have national registries rush to create them? Tony said that this would be a defence mechanism. Wilfried Woeber asked why APNIC does not like this proposal. Tony Hain said that there was nothing concrete but that there was a specific instance where he was accused of racism and it was assumed that he was attacking specific regions. He said he was not. Wilfried said that he was surprised that a proposal did not get the proper treatment. Gert said that this case should not be discussed right now. Ray Palzak (ARIN) said that he wanted to comment about what happened at the ARIN Meeting and the status of this proposal in the ARIN region. He continued that there were 12 in favour, 25 against and that the Advisory Council recommended that this proposal is abandoned. Tony said that this was the original proposal and asked how many people were in favour of continuing to work on this. Ray said that this proposal was not alive in the ARIN region. Chris Morrow (Google) said that 'RIR shopping' happens today on a relatively large scale and said that he knew of at least two examples. Gert said that this is a more formalised way of doing things without people setting up subsidiaries all over the planet. An attendee said that large global corporations do this today. Chris said that there are organisations who have operations in California but have APNIC and RIPE NCC address space. Akinori Maemura (JPNIC/APNIC) said that he will check the situation but thought that the proposal did not have consensus at the time. APNIC has NIRs, but there is a shared pool at APNIC which serves the NIRs, so this is not a concern. Gert asked for a show of hands about the mechanics of the proposals. A few people thought that the general direction of the proposal was good. No one thought it was a bad idea. Gert commented that three-quarters of the room did not show their hands. Daniel said that it could be that they have no opinion. Gert concluded that there is some support, and some voices against and said that the proposal would be sent the mailing list for further discussion. F.1 Ongoing transfer policy http://www.ripe.net/ripe/meetings/ripe-56/presentations/Yilmaz-Others_Transfer_Proposals.pdf Discussions in the other RIR regions Observing Others: Transfer Proposals Filiz Yilmaz (RIPE NCC) This presentation was given to provide an overview of the transfer proposals in other regions and to facilitate discussion in the RIPE region. There were no questions. F.2 2007-08, "Enabling Methods for Reallocation of IPv4 Resources" http://www.ripe.net/ripe/meetings/ripe-56/presentations/van_Mook-2007-08_update.pdf "The Market" proposal Remco van Mook (Virtu) 2007-08 update Gert Doering (Chair) said that people are doing transfers today. Not doing this will not stop it. This is to help the RIPE NCC keep track. Michael Dillon (BT) said that the thought this is a waste of time. The period in time is that the IPv4 is used up, so that most of it is in use and is not available for transfer. Most of it is allocated to a few large ISPs. These are companies that are constantly growing, so they won't sell it, and even if they did, it would cost too much to validate and check and get approval to release the addresses,making sure the left hand doens't give away what right hand needs. He said he did not see a lot of transfers happening at all. Remco van Mook (Virtu) said that not all organisations are as big as BT and that he did not expect to see /11s transfered through this policy. There are actually quite a few people who would like to get their hands on a /20 and so there needs to be a mechanism to get that administered properly. Andy Davidson (LONAP) said that he supported this because it allows documentation of the changes that will happen. He asked if PA space must not have any assignments in it when it is transfered. Remco said that this proposal is only for allocated and unassigned space and that is because he thought it would be the easiest to address. He said that PI space could be included and allocated and assigned later. James Wilson (HEANet) said that he agreed that it would be difficult to free up space. Organisations need to move to IPv6, but in order to do that they still need some IPv4 space. This will take place at different rates in in diferent organisations. He continued that unless we do nothing, the only option is to provide the posibilties for one organisation to free it up quickly so another can use it. There is no other proposal besides this that can get us there. He stated that he supported this proposal. Tom Vest (RIPE NCC) asked if there had there been an evaluation of the legal implications of this. He said that if you take a common resource and transfer it to a private party, this is privitisation. It has huge tax and regulatory implications. Although this is an ambiguous case, there is a lot of economic doctrine about this. The substance of this is very close to privitisation and that probably merits careful review. Tom continued that when you associate the IPv4 resources with value and, if the value is substantial, you have the resource and you know it will be more valuable every day. It could create delaying migration incentives. Remco said that he thought it will be huge incentive for IPv6 to evolve. The entire IPv4 legacy space will be worthless eventually because everyone is on IPv6. It might take a while, but it will happen. If IPv6 is free and IPv4 is expensive, then there is automatic migration. Remco said that there is a legal and economic aspect, but that nothing can be done about that. People are doing transfers, we need to accept a framework. It is already a defacto standard. James Rice (Jump Networks) said that the minimum transfer is minimum allocation size, instead of minimum /8 allocation size and asked if Remco was sure he wanted that in the proposal? Remco said that by increasing the use, you will increase fragmentation. He said that we could stick to the minimum allocation size for the /8, but thought that there won't be a difference. He said that he would rather keep it simple and have a single number. Alain Bidron (France Telecom/ETNO) said that during the RIPE 55 Meeting a set of principles was presented and there was a followup. He said that there was a new common position on IPv4 exhaustion. This was published on the Address Policy mailing list. The main message is that before we support transfers, we have to carefully consider the impact of such a move. He said that he referred the audience to the position paper from ETNO and said that ETNO was not in a position to support this proposal. Remco said that it was fiction that this proposal creates a market. This is not true. This proposal will just document what is already there. Alain said that all impacts have to be carefully considered and analysed. It is a radical change. Remco said that if there were two more years to analyse it, we should do that, but we don't have two years left. He said that if we don't start implementing this it will be pointless. Therefore, we should move forward or abandon it. Daniel Karrenberg (RIPE NCC) said that Tom pointed out that if we privatise these resource it will have interesting consequences. It might actually increase those perverse incentives. He continued that If he was a big ISP and had lots of IPv4, it would prevent me from going to IPv4. If you look at ARIN/APNIC, at the differences, you might say is ARIN over-specifying. But what they are really trying to do is to keep the address space as public resource. You cannot transfer without the need being demonstrated beforehand. The APNIC and RIPE proposal says you can transfer but with one sentence: "the usual policies apply". He continued that, for example, the transferee does not have justified need and says "I sell you this address space" Do you think the RIR has a chance to go after them and say you don't need this. He said the thought this was fiction. Either come out and say "we privatise", otherwise it will bring the RIR in an untenable position. We can do this, and say this is just window dressing. He said that he was not saying that one is better than another but that we should be honest with ourselves. Roque Gagliano (LACNIC) said that this is also the view of the LACNIC Board: any transfer policy for legacy resources should be discussed amongst all RIR communities in a global basis. Gert concluded that the proposal would be sent to the mailing list for another round of discussion. Daniel said that we should listen to our LACNIC colleagues and ask if the regional policies should be better aligned then they are now. He said that maybe we should ask approporiate parts of PDPs to align themselves. Gert asked for a show of hands to see whether parts of the PDPs should be aligned across regions. There was a lot of support. ACTION ON RIPE NCC: Keep track of other regions PDPs and figure out if it's possible to align the proposals. Hand Petter Holen (Address Council) said that there is some work there for coordinating processes. He said he was not saying that the AC should do that, but there is some work there. Z. AOB ----- End forwarded message ----- -- Total number of prefixes smaller than registry allocations: 128645 SpaceNet AG Vorstand: Sebastian v. Bomhard Joseph-Dollinger-Bogen 14 Aufsichtsratsvors.: A. Grundner-Culemann D-80807 Muenchen HRB: 136055 (AG Muenchen) Tel: +49 (89) 32356-444 USt-IdNr.: DE813185279
- Previous message (by thread): [address-policy-wg] 2008-03 Proposal Accepted (Global Policy for the Allocation of the Remaining IPv4 Address Space)
- Next message (by thread): [address-policy-wg] Agenda items for RIPE 57 in Dubai?
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]