Due to the difficulties capturing a live speaker's words, it is possible this transcript may contain errors and mistranslations. APNIC accepts no liability for any event or action resulting from the transcripts.

THURSDAY 0900-1115

RANDY BUSH: Excuse us - we have a little technical excitement. A question: it would be nice if we had a volunteer to watch the Jabber chat room. And if somebody has a question in Jabber, that the person watching the Jabber chat room could speak for them. Owen? Would you like to volunteer?

OWEN DELONG: I don't speak Jabber

RANDY BUSH: Who can do Jabber? Somebody is going to help you do it web based. Sunny is going to come. Thanks, Owen.

For those of you who can read small print, there is the... is there some way to actually get the agenda? OK. My name is Randy Bush, I am from IIJ in Tokyo. I am chair. Jian, to Sam's right from CNNIC is co-chair. Thank you, Jian. Sam is not here, do not look behind the curtain.

If you can read it, that is today's agenda. We have two pressures on making the agenda work. The first one is that we have remote participants from Suva in Fiji and from Colombo in Sri Lanka. They have skewed timeframes and you may see them, the magic people up there who will throw things at me, when they need to speak we will switch to them.

The other pressure, we have a long and complex proposal set on transfer. And what we are going to try to do is, if you remember, there are at least three proposals here in the transfer area. But what we are going to do is not discuss those proposals in detail as proposals. Underneath those proposals are eight or so questions of substance. That is how we are going to approach it. So we are going to, we have an agenda, a detailed agenda on transfer that will discuss those eight. What we will try to do is cover them before lunch. At lunch, all the proposers will go crawl in a corner and try to merge everything between the proposals and what the community consensus of this room has been before lunch. Then come back after lunch with something we can talk about as a whole. There is Jabber participation, Owen DeLong has volunteered to represent those people by proxy. Owen, please stand up at the microphone, throw things at me etc, etc, when things come up. Do whatever it takes to get the attention of Jian and myself. If you are in the room and, for some reason would prefer to speak through Jabber, feel free to do so.

There is, if you go to the microphone and speak, please do state your name and your affiliation. Affiliation is code word for "who do you work for?". Be aware for most of the participants here, English is not their primary language. So try to speak reasonably for the audience, which means slowly, no idioms, as clearly as possible. Do not follow my examples.


One other thing, I am sorry to be rude but this is the Asia Pacific region. We do honour our guests from other regions but if you would please, as speakers, give priority and time consideration to those from the Asia Pacific region. And if anybody has any procedural questions, now would be the time to ask. Otherwise, Jian is going to take us through the overview of the policy development process, and we are going to do the summary of proposals, the traditional administrative things and then shortly, we will have the SIG co-chair election. Any issues? Thank you.

JIAN ZHANG: I got it.

RANDY BUSH: Just to introduce you while she is setting up, here are the people in Suva. Doesn't do much of it, there you are. They are the people in Suva. The people in Colombo haven't come on yet.

JIAN ZHANG: Good morning, everyone. First I am going to do some introduction, brief introduction, policy development process. Here is our policy SIG charter. Our charter is develop policies and procedures which relate to the management and use of Internet address resources by APNIC NIRs, and ISPs within the Asia Pacific region. Below, actually is two mailing lists we have. One is SIG policy, one is SIG policy chair. Our policy development process is open, transparent and bottom up. Anyone, by open, means anyone can propose policies and anyone can discuss policy proposals. Also, APNIC publicly documents all the policies discussions and the decisions. And the community drives policy development. So it is a bottom-up process.

This chart shows our policy development process. Four weeks before our APNIC meeting, you can submit a proposed policy or amendment to the APNIC Secretariat. SIG chairs will post the proposal to the SIG mailing list and the community will discuss the proposal on the SIG mailing list you just saw on the first page.

And then at our APNIC meeting during this week, we are going to do a face-to-face discussion in SIG session. Just like today. If at any point we get consensus, we are going to do report of SIG decision to AMM. If we get consensus AMM final call for public comments belongs on the SIG mailing list.

RANDY BUSH: Too much technology.

JIAN ZHANG: So finally if we reach the consensus, it will be endorsed by the Executive Council. And then, we're going to do the implementation so the process is going to take a minimum of three months. Remember, all proposals have been submitted with good intentions, but opinions may differ from people to people, but we all want the same thing, a strong and a healthy Internet, so play nice.

Next, Randy is going to go through the summary for the proposal we're going to discuss today.

RANDY BUSH: Now you're going to watch Randy not able to work a Windows system! She's cheating, today I put up an agenda. This is an overview of the proposals today and we'll go through them quickly and what we have left from previous APNICs. That's prop-050 on address transfers. Prop-060 on the change in criteria for the recognition of NIRs. In other words, changing how NIRs get established. There's also prop-063 reducing the timeframe of IPv4 allocations from 12 to six months. There's a bunch of new proposals this time. Prop-067 is a simplified transfer proposal and when we discuss it, the discussion is merged with prop-050. There's prop-068 which is inter-RIR transfer policy that will also be merged into the general transfer discussion.

And prop-069 is the global policy for the allocation of IPv4 blocks to Regional Internet Registries, this is the secret proposal by all the fenced and important people, important people at all the RIRs for global policy and hopefully they'll tell us what the purpose is. And then there's the change in the maximum IPv4 allocation size.

So prop-050, OK, we're merging the transfer proposals, all three of them. So we'll discuss them as a group. We need to go through with David who suggested that we do it this way and that the minimum size is one of the items, whether their use is justified, whether transfers can happen between RIRs and whether there's somebody in APNIC can get something from the RIPE region. Whether an NIR member can transfer to APNIC and presumably therefore to another RIR. Whether NIRs can transfer between each other. Whether the seller must be a full member of APNIC and whether it goes through from the list, and there's a suggestion that the seller may not be allowed to get more space once they've sold the space that they have. And then, whether or not APNIC should maintain a historical log of all transfers so you can see the history of a particular bit of address space. As I said, there's the change in recognised criteria in the recognition of RIRs, changing the timeframe allocation. A global policy for the allocation of IPv4 blocks and minimum allocation size.

For some reason, I'm supposed to do this in detail. The prop-060 says that a new NIR should be recognised if it has the - oh no, the current policy says that the new NIR must be recognised by both the community and the Government of the economy where the NIR is proposed. So therefore, what's seen is that if it is a possible, the NIR can be done by Government interest. The solution is to just have the community do it and all the APNIC members in the community and there would be lilting of the Government position on NIR boards.

On the mailing list, version 1 was sent out in May of 2008. Version 2 came out in August and there's been posting from various people. There's prop-063 which addresses a problem that as we run out of IPv4 space, as we get near the end, somebody could come in with a great big request, some big telco or whatever could come in with a big request and swallow all the rest of the IP space towards the end and that's because they can make a projection for 12 months. So this proposal says - start chopping it and making it smaller. OK. You can see when this proposal was made and said and generated and it was discussed on the mailing list. For our region, fairly extensively, not many people but that's life. There's a global policy proposal for the allocation of v4 blocks to the registries.

Once the free pool is depleted, the current policy says that it is not going to be useful because it stays within the RIR. So there's no policy for how space that's been claimed or returned can be shifted between RIRs. So this proposal says there's a new global pool so that if return space gets given to IANA and then can come out to the various registries, there has been some discussion on it on the list, there's been a new revision. OK, the prop-070 on maximum v4 allocation size - it's possible that a holder could get a large chunk at the end. This is again addressing the same problem that a previous proposal on the time window to reduce it from 12 months to six months, this addresses the same problem is that the people sitting at the table at the end of the meal could go and get it all.

The solution is to create a maximum allocation size that decreases the closer we get to the end of the v4 pool. There is only one post from one person on the list that hasn't really been discussed, so please discuss it here.

That's the proposals we have.

I bet if we go here, I'll get the agenda and next is the co-chair collection. We don't have a foil for that. We have two people who are volunteers for co-chair and you will find on the website or in the e-mail, their statements and qualifications. So, could you speak for just a moment and explain how wonderful and helpful you're going to be as co-chair? Please. I'm just doing it in the order that they volunteer. It also happens to be alphabetical.

RANDY BUSH: In Japan, in the Kibuku theatre there are people dressed in black with black shields on their hips and they come out and they move props, etc and they're visible on the stage during the performance, but you're not supposed to see them. And so you cannot see Sunny when he goes and helps and you cannot see Sam. They are invisible!

CHING-HENG KU: Good morning, everybody. My name is Ching-Heng Ku from TWNIC and I am the director of the IP department. I work in the area of networking for ten years. I am an active participant at APNIC meetings since APNIC. I hope to have the chance to contribute to APNIC and I have been a chair of the NIR workshop, NIR host master workshop. The co-chair of the database SIG and the co-chair of the NIR SIG. I hope to have support for me to serve you in the Policy SIG co-chair. Thank you very much.

RANDY BUSH: Thank you.

DAVID WOODGATE: Good morning, my name is David Woodgate and I work for Telstra administrator. I've been working in the Internet industry now since 1990 during which time I have - was participating in the IETF during the 1990s and more recently you may have noticed me at recent APNIC meetings participating in the policy development. I have a great admiration for the policy development process that is used in APNIC. And I would certainly like to offer my assistance to maintain the very good management that has been displayed in this policy development.

Two things which may be - something to consider - whether they would be issues for people who are about to vote. One of which is that I have had strong opinions on the mailing list. I expect that I will continue to have some strong opinions at times. I believe I have the capability to separate the duties of co-chair from being that of a contributor. I feel that the precedent as to how to do that has been set well by Randy in recent times. So I hope I can also achieve that kind of level of distinction between the roles of chair and the participant.

The other thing to consider is that and standing for the EC of the meeting, and if that is seen to be a conflict of interest in any way, then please make any doubts known, and if on either of the points, whether my strong contributions to the list or my nomination for the EC causes any substantial concerns, please let me know and I will cheerfully remove my nomination. Otherwise, I hope that I may have the honour of serving the APNIC community as co-chair and thank you for your time.

RANDY BUSH: OK. Just to give you... just in case for those of you, some of you who are new or don't know the position of chair and co-chair, etc. As you can see, we run the process and run the meeting and try to help with the mailing list and try to get people to participate and express their opinions, etc. So that's essentially the job. There are two candidates for it. At this point, I think we're just going to ask the room and the remote participants and are people from the Secretariat actually going to help with judging? The remote participants for us?

SRINIVAS CHENDI: We can switch them if we want to.

RANDY BUSH: Though I don't see anybody there. Are they there? Colombo will join us at tea time. Wise people. So here's the folk from Suva. Let me be tactless and just ask how many people would vote for Ching? That was a call for hands being raised to support Ching. Hold them up, please, for the people who are actually trying to count. Suva, you are welcome to vote. There are people actually counting. Suva, can you hear us?


RANDY BUSH: OK, just checking. Are the counters happy? OK, those for David please raise your hands. You're now stuck! You get to give the next presentation! OK, thank you all for participating.

Jian is now going to review the open action items. What we mean by the open action items is last time we passed some proposals or asked the Secretariat to do something, etc, and, so we're just going to run quickly through the list of action items from the previous meeting and report what the status and the completion etc is for them. And I'll try to bring that up if I can figure out how to operate a Windows PC. I bet you one of these things says how to do it right.

JIAN ZHANG: The open action item will be first is prop-026 version 1. The author of the proposal changed in the criteria for the recognition of NIRs in the APNIC region to revise the proposal to address questions raised on the mailing list and at the APNIC 26 Policy SIG. So the update, the current update on the agenda is for the Policy SIG.

Next one, prop-026... so, our next open item is pending...

Yes, it's the prop-064 pending approval at each remaining stage of policy proposal process. APNIC Secretariat to implement proposal prop-064 - "Change to Assignment policy for AS numbers."

The current update is currently being implemented. The third one will be pending approval at each remaining stage of policy proposal process. APNIC Secretariat to implement proposal prop-065, format for delegation and recording of the 4-byte AS numbers. The current update is actioned advised by the IETF published as RFC 5396.

Fourth action item is pending approval at each remaining stage of the policy proposal process. APNIC Secretariat to implement the revised proposal prop-062, 32-bit ASN numbers for documentation purposes, that 4-byte AS numbers from both 16-bit and 32-bit AS pools to be reserved for documentation purposes. So the current update is actioned by the IETF published as RFC 5398.

Fifth, open action item will be pending approval at each remaining stage of policy proposal process. APNIC Secretariat to implement the revised proposal prop-062, use of final /8. Revisions to the proposal include making the size of allocations reflects the minimum allocation size at the time of the last /8 is reached. This proposal not be dependant on the successful adoption of prop-055, global policy for the allocation of the remaining v4 address space. The update is implemented.

Sixth one is pending approval at each remaining stage of the policy proposal process. APNIC Secretariat to implement the proposal prop-055. Global policy for the allocation of the remaining v4 address space. Update is pending completion of the global policy process, currently in the last call before going to the ICANN Board of Directors.

Seventh will be prop-063. Reducing timeframe of v4 allocations from 12-6 months, returned to the Policy SIG mailing list for further discussion. The update is, updated version to be presented in this SIG. We're going to do it today.

So the eighth will be pending approval at each remaining stage of the policy proposal process. APNIC Secretariat to implement the proposal prop-066 ensuring efficient use of historical v4 resources. The update is implemented.

Ninth will be prop-050. IPv4 address transfers to be returned to the author and the mailing list for further discussion. The update is transfer proposals to be presented in this SIG. We're going to do it right after this. So that's all for the open items.

RANDY BUSH: Anybody have worries about something we forgot? OK.

The next item on the agenda... we're going to do transfer. There are a couple of presentations and then we're going to go through the discussion items individually. Jian is going to run the agenda and I'm going to try to help co-ordinate the speaking for each of them. Just a second. You want something else on the screen over there. Yes, I will put my presentation on the screen when I make it. Except that it just went backwards on me.

This presentation is actually not my presentation. This presentation is made at the RIPE in Dubai last October. And the point is that why we should be doing transfer. The point of this is, you know, this a classic diagram and well, A is the way things are. B is the way things will be and the intersection of the two is smaller than you think. That is, things are going to be very different. And that's because if we run out of v4 space and we're going to a different world. The current policy fits very well. We have the light green, we have a bunch of proposals that we've been dealing with for how to handle as we approach the end of IPv4, and that includes some things called sun set proposals which for us is the shorter time period for allocation. The smaller allocations, etc that we've seen in the various proposals that we'll be speaking about today.

Then v4 is going to become not available for some people. So, as Remco said, the solution is "get IPv6". So, this is what Remco is really saying, get IPv6.

This is a different market, but it is the same problem, we're going to be crying. Remco wants to say that no addresses were hard in this discussion or hijacked. This is actually a real - this is what I'm about to show you from Remco is real, no MDAs were violated, it was a real transaction. This is a real message that Martin received which is a solicitation to sell the note to buy address space. It's a real message, somebody actually sent him a message saying - can I buy address space?

This person, they then held a meeting with this person. This person would provide all legal services, would buy as many /16s as possible. They were not worried about ICANN, IANA, RIRs or IETF. They were not worried about what we call the quality of the address face, filtering, registration agreements, routeability or reachability. It was a real company, a real individual, it was not a scam. And they had the process all there, the way they'd been doing business for completion in less than 60 days. This is a broker. This isn't an end-user, this is a broker. The address black market is now big enough that there are middle men in the market. The address market and transfer are there. OK, which is I think Remco's point.

The process of going through is to execute a non-disclosure agreement. A holding company would be set up. A transfer would be made to the holding company by selling the transfer to the holding company and then the holding company would be sold to Dubai. The funds would be wired. They were offering about $175,000 US for a /16. That's about $4.23 for a host.

What's the point of this? The market exists whether it's legal or not. Legacy address blocks are being transferred while we debate transfer and who is there, etc are suffering. OK, money is changing hands and Remco is saying, if we just do transfer, then everything would be simpler. Any questions?

JIAN ZHANG: Next I'm going to talk in more detail about this transfer proposal. First, we're going to talk about what does the transfer proposal do? Actually, Randy already did some introduction of the transfer proposal. The transfer proposal allows APNIC to reflect records of transfers undertaken by APNIC account holders. And today, we're going to discuss upon each element of the following.

First one is - should we allow address transfers between APNIC account holders? So, that's a question about - should we allow this transfer to happen?

Then the second one would be - what should be the conditions of the transfer. Then more detail would be... was a minimum size, could it be transferred and should be there, any justification. And then seller may or may not get more space for some time period for the future after the transfer. And then APNIC to maintain a public log if APNIC need to maintain a public log of all the transfers.

Next question I'm going to ask you, your opinion is, what should be as a timeline of the implementation. Should the way it is implemented right after it is passed and should we wait until IPv4 runs out?

And the next question will be - what are the parties that are recognised to do transfer and APNIC policy? So that's about scope, what kind of scope we're going to implement this policy.

First question is, should APNIC recognise transfers between APNIC current account holders? APNIC will reflect the records of the transfer with the request from both the source and the recipient. Current APNIC account holders means current APNIC members and the member and non-member account holders.

Second question will be as a condition, was a minimum size should be a minimum size of address space to be allowed to transfer. One opinion is that we should set the minimum size of slash /24. The reason is likely that the slash /24 is likely to reflect the reality of the size of address space to be transferred.

Another opinion is that it is consistent with minimum allocation size or portable assignment size. The reason for that is consideration for routing table growth so the routing table wouldn't get too big.

Another question is - the justification of use. Should we request the recipient to justify address requirements at the time of transfer? One opinion is, should justify address requirements to APNIC. That will maintain the current practice. But on the other side, another opinion is, no justification should be required, so we just stick to the role as registry once the address pool runs out.

Next condition we're going to discuss is restricting allocations. Should we add restrictions to the source of address transfer to several subsequent allocations from APNIC? Well, opinions are that there should be restrictions added, for example, no allocation or assignment from APNIC for the first two years in the future. That will prevent people from requesting space from APNIC for the sake of making transfers to other organisations. But some others are saying that we shouldn't add any restrictions because it really doesn't make any difference.

Next one will be - should APNIC publically provide a log of address transfers? One address is yes, a public log should be provided. The reason for that is that that will help in the assessment of risk and workload required to make the address useful. But on the other side, some people say no, a public log is not necessary because it is not particularly useful. And then it's a timeline of implementation. When should this policy take effect. One opinion is implemented as soon as it's ready. The reason for that is that will make the problem worse if we wait too long. And then another opinion is to wait around the time that the IANA/APNIC pool runs out. The reason for that is that if we implement it too early, will, it may encourage transfer and create other problems if it is implemented too early.

And the last one will be the scope of transfer. What extent should be recognised to transfer between different parties? The first option is that the scope will be only within APNIC account holder. The second option will be between APNIC account holders and account holders of other Internet registries. So that means that the transfer between account holders of RIRs or the transfer between the NIRs.

So this is a summary we're going to have of the questions we're going to ask you and your opinion today. And that's where it goes through this list one by one and we'll ask for your opinion and go through for the SIG consensus. And that's pretty much it for the transfer.

RANDY BUSH: So, we're going to take the questions she just enumerated in order. I'd first like to ask a simple question. How many people here have read and have been keeping track of the mailing list? OK, and you've seen discussion on the list and in the interest of openness, a number of us have been meeting and discussing and trying to clarify some things between us so that we can understand better what the positions and feelings are.

And I think you're going to see that in the first item under discussion, it essentially says - should transfers be allowed at all between APNIC members? All right, and I think that's partially what Remco's presentation spoke of. Would anybody else care to speak to - should transfers be allowed? There could be some specific people if you tilt back your chairs and pretend you don't care.

JOHN SCHNIZLEIN: John Schnizlein from the Internet Society. Going through with the phrasing a bit, it may be the right question but Internet transfers of addresses are happening. The real question is, should we make sure that we record them all? My concern is with the integrity of the routing infrastructure. And there are real problems today, knowing where address prefixes are legitimately injected into the global route tables. This strikes me as essential to the preservation of the Internet, and so it should have priority over many other valuable reasons. You have to record it.

IZUMI OKUTANI: On behalf of the JPNIC address community. I think that opinion is very similar to the early figure, that you know, it's very important to keep uniqueness and to do that after the exhaustion, then we must make sure that we accurately record the transfers of address space, so there was general support for the concept of the transfer.

RANDY BUSH: Let me interrupt for a second and let Owen go first. People should understand that there are two people here who we know well, Izumi-San who generally speaks for APNIC and you should understand that in fact, the APNIC community, the membership speaks and we push it upstairs to the EC and they approve it. In the JPNIC community, there's the community who speaks, but there are also the board of directors who are responsible for running a business. And so, we're bottom-up. They're both bottom-up and top-down. So Izumi speaks for the bottom-up. And Maemuri speaks for the top-down and both are for the good. Just a moment because I think we had somebody from Jabber.

OWEN DELONG: OK, actually. So there are two people on Jabber. Nesh says the /24 transfer is an OK size but the justification should be provided.

Terry M says, "Yes, transfers are occurring now, let's recognise them".

RANDY BUSH: OK, the first one, we haven't gotten to that subset yet.

AKINORI MAEMURA: From JPNIC and I need to correct what Randy said. I am not speaking something from the top-down. JPNIC is still a member of the APNIC and also the broader APNIC community.

JPNIC as the member for the APNIC communities, it's a bit strange or rare and usually the APNIC members, usually members are LIRs who provide a service to the end-user by holding the IP address location. But JPNIC is a National Internet Registry which it is serving to the LIR. And JPNIC as the Secretariat or incorporation needs to think about the service provision for our members. I mean the JPNIC members, and actually, we discussed a lot for the transfer proposal in internal JPNIC Secretariat, but still, we are not confident we can successfully or safely provide a service to our JPNIC members. I wrote an article to the SIG proposal mailing list, it was lengthy, I'm sorry, and we have the transfer transaction can be made in quite a safe manner.

If there is no risk which the member or the stakeholders will be exposed and so on, we would need... you think that JPNIC, we need to work with such a certain point before the transfer moves ahead. There is something different in the positions with that that the JPNIC community has, because as Izumi said, the JPNIC community in general is in favour to have the transfer mechanism. But the JPNIC Secretariat's opinion is still uncertain.

RANDY BUSH: Do not leave the microphone. Two things - JPNIC may not want - if JPNIC has not developed enough process within Japan to be happy with doing the transfer, do they have a problem if APNIC does it for them? And the further evolution of this question is of course, JPNIC members who want to do transfers, which in APNIC join JPNIC which probably isn't good.

AKINORI MAEMURA: Yeah, that's one of the concerns. If the APNIC has a transfer mechanism. Actually, in the LIRs in there have the choice to go to JPNIC or go to APNIC, it's like, "I don't like this word but it is like a competition" and then if APNIC decide on the transfer mechanism, then JPNIC need to conserve very much about starting the transfer mechanism also in the JPNIC. Does that answer your question? Thank you.

GEOFF HUSTON: Geoff Huston. I'm the author of one of the proposals.

RANDY BUSH: Affiliation please.

GEOFF HUSTON: So for the purposes of today, you can assume that I've resigned from APNIC. We are where we are today because for the last ten years, this entire industry has said, this v6 stuff is really uncertain, so we need more time. So we had this transfer policy proposal first introduced into this community in September of 2007. And it was noted at this point that this was very complex and had a lot of implications and the folk would need some time. So we said - we'll come back again in six months and see how you feel. So we came back again in March of 2008 and I had said - how do you feel, you've had some time. And the general response was, well, we need some more time. So we came back again in Christchurch in September, or was it August? September of 2008 and said - how do you feel now?

It's probably time that we did something and the response that we got - as far as I understand - was that we need more time. So, what's going to happen on April 1, 2011? Do you need more time? How about on June 30, 2012. Do you need more time? Because life is uncertain. And the place that we've got to is incredibly uncertain, and what we don't have is time. We actually have a very limited amount of time. It is inexorable and you negotiate with it and there are a lot of people who need to do things between now and when the current distribution for IPv4 addresses comes to an inevitable conclusion. We have to prepare for whatever we need to do after that inevitable conclusion.

I think it is actually a little bit selfish of us to claim all of it for ourselves and allow no-one else to have some of that time, because other folk actually need to make their plans based on what they perceive to be the registry infrastructure that's going to be provided after the current distribution system comes to its inevitable conclusion, and they can not prepare, cannot make plans, cannot do their work, cannot have their share of the time if we steal it all. So at some point, this community needs to understand that we're in the way. And that other folk can't do their necessary work without understanding what the registration framework will look like. And that means that we have to make a decision that actually is made with some uncertainties. There is some risk. It won't be known completely the implications of everything that we do here, but if we take the time and steal the time from others, a perfect outcome too late is a lousy outcome.

And what I suppose I'm saying here is that in response to what I hear from JPNIC, I would have to say with deep respect, that what they're asking for is actually something that we can not give them. But we need to share what little time we have with everyone else who actually needs to work with the outcomes, and that means actually primed to understand the wisest form of action here is to actually lay down some frame working foundations and allow others in the industry to then start doing their preparatory work, because the distribution will come to a halt.

RANDY BUSH: Excuse me, I'd like to try to summarise or clarify. From my talking this week with many people, my understanding about the issue of time is that the time, the people who are there, who need time are not saying that we need more people to decide whether or not to do transfer. What we need time for is once we have passed this proposal, this policy proposal with its details before doing The transfer, then many of us, including the APNIC Secretariat, including JPNIC etc need to go home to their various businesses and economies and work through the implementation details. That's my impression, so I'd appreciate it if anybody in the next few minutes who thinks I'm wrong there... thinks I've summarised incorrectly.

GEOFF HUSTON: You've given the implication and let me respond. The question I thought at this particular junction was discussion of whether or not. And I heard the contribution from Akinori-San on behalf of JPNIC the organisation, responding to that particular question. So my response was again around that particular question. Should we or should we not? Thank you.

RANDY BUSH: Jonny was first and then you can go.

JONNY MARTIN: Jonny Martin from New Zealand. Should address transfers be allowed? No, no, and no which is quite the opposite of what I said at the last meeting. I had the chance to talk to a number of people. Now, John from ISOC makes a really good point, we need to preserve the registry, but that's not going to stop the illicit transfers from happening. The illicit transfers will have just as fatal a result, so let's not think that by recording as many transfers as we can, we keep the registry intact. Geoff is right, we shouldn't think that we're the lucky ones and we give the address space a go. The big problem of what to do now is setting a precedent for every protocol to come, so as soon as we start trading addresses and that's what transfers requires, governments and other institutions are going to get involved.

It doesn't matter what we say or what we try to do to change that. Do we want that happening in v6 as well? That's a no, the answer to that one. So yeah, that's it.

AKINORI MAEMURA: Akinori Maemura from JPNIC. What I am going to say is almost - it is already said but Randy, thank you very much. But I just raise a simple question. What we need is not the time but the preparation for anticipation, but I know that such preparation and anticipation takes some time. But the important thing is that we need preparation.

RANDY BUSH: Need preparation for implementation?


RANDY BUSH: Right, but JPNIC the organisation is OK with passing policy as long as we all agree on time and co-ordination for implementation. And that we will work together on how we implement it. Question.

AKINORI MAEMURA: It is still possible to have the meaning other than the transfer mechanism. In the sense that I don't necessarily like to have the transfer mechanism to be accepted as a consensus, but the most important thing is the preparation before that reinforcement. So in that sense, you are right.

IZUMI OKUTANI: I'm not speaking for JPNIC but for our community. Actually, we had quite a bit of discussions and open policy meetings in July, and at that time, a very similar concern that Jonny expressed was expressed within the community, that if we do this and we have a lot of implementations outside of the address problems and then we shouldn't create the problems. But then afterwards we had another meeting which Randy and Geoff joined us in the discussion and it was clarified that you know, regardless of whether we like it or not, we can't avoid these things happening, and then of course we can't solve all the problems that are - you know that we are anticipating. So you know, we should do what we can do in terms of what we can do as a registry and try to solve it within our role.

And, of course, other issues continue to be an issue, but then it is important that we get this going and then at least do what we can at this stage. That was the general feeling of our community in Japan.

JAMES SPENCELEY: From Vocus. This is the AP community. I thought what I might like to do is call for a show of hands for anyone who has actually been offered the sale of IP address space. Just as of interest, to see what the actual problem that we're trying to solve here is? So could anyone raise their hands who has been offered address space for sale? Anyone in the room been offered IPv4 address space for sale?

RANDY BUSH: Behind you.

JAMES SPENCELEY: It doesn't seem like this problem is happening and I think we need to be very careful creating a market which has a distinct and known value on a resource where the resource is achievable on a subsidised price. I agree transfers will be necessary when we don't necessarily have that subsidised before, but what we need at all costs that places value as we will have certain unknown effects because of that. So I encourage people to prepare to approve this proposal with the exhaustion of the v4.

RANDY BUSH: Sorry, just a second, James, let me clarify. Again, you're not specifically against transfers, you're speaking to when?


RANDY BUSH: OK, that's another item on the agenda. David, which of you two was first, sorry. David was first, sorry.

DAVID WOODGATE: From Telstra. Just responding to Jonny's comments about precedent, that I guess I haven't looked at the websites myself. And I certainly would invite members of RIPE or ARIN to correct any statements I make now, but my understanding is that the precedent has now been set within the world at least in terms of the RIPE now do have an active policy on transfers and there is at least being recommendations to proceed with the proposal on transfers on the ARIN region. As I say, I haven't checked the details so I may be in error on this. So, I realise that the Asia-Pacific region does have its own needs and considerations and it is very important that we address that going into anything, especially with the situation with the NIRs, nevertheless, I do feel that a precedence has been set and we do appear to be heading towards a framework of transfers anyway.

MASATO YAMANISHI: Since my company is a Japanese company, so I basically understand the situation in JPNIC. They are... I think JPNIC need to make sure of the stable service and also the address transfers they may have in the domestic migration. And also understand many NIRs have similar concerns. However, to resolve these concerns, we need to move forward to figure out what is the concrete issues in implementation phase on, maybe in the old address transfer phase. But the reason why I want to support address transfer. Thank you.

PHILIP SMITH: Philip Smith from Cisco. OK. My microphone is... so just to address the transfer, I'm actually one of the co-authors for one of the proposals. I think it's - you know it's nice and easy to say that no, we should not allow transfers to happen. We're being very idealistic, we can all say that. The reality is that these transfers have happened and we've seen in Remco's presentation and there were plenty of other situations where address space has moved around, so I want to echo what John said right at the start, it is really important that we document who is holding address space. ISPs or with the ISPs running the businesses properly, check the registries and the allocation address space to be April to be announced.

If we no longer know who's fault, how are we going to be able to announce? It is very important whatever comes out of this that we have to be able to carry on documenting who is holding address space.

EINAR BOHLIN: Einar Bohlin from ARIN. And in the end, the board has adopted the transfer and they have not instructed staff on how to implement it yet. They have details on how to pass it and there should be an announcement coming out on that in mid March.

RANDY BUSH: To be... I think I don't need to say anything, there's Filiz.

FILIZ YILMAZ: In the RIPE region, the transfer policy has been accepted back in December 2008 and implemented already in early January, so the are in effect -- so the transfers are in effect in the region. I must say that it is not the same, but as of the question for win, it is in place in the RIPE region.

SEIICHI KAW AMURA: We're an ISP with a lot of broadband customers and we are an JPNIC member. In two or three years time that IPv4 address space will be exhausted or depleted or whatever you want to call it, I think I would still have customers that would come to me and would want an IPv4 address. And if I can't provide it to them, it would be really nice to have an option and maybe go and buy addresses to give to the customers. If I had the option, I wouldn't have to go there and say, "No, I'm sorry, it's all sold out" which is kind of bad from the customer's point of view. And so, I am in support of the transfer proposal and I think about my customers. Of course, like Jonny said, we're going to have legal issues and that's going to be a pain for all of us. But we do have legal issues already and I think there would be a lot more pain on the ISP side could be handled By the ISP. That's my opinion.

RANDY BUSH: OK, Geoff and John.

ROQUE GAGLIANO: From LACNIC. We don't have right now an active discussion in the transfer policy in the LACNIC region. We did have some discussion in the last policy forum in January discussions, but not about the proposal. However, in order for the transparency, we did receive a proposal in the last 24 hours and in the process, we go through the translation in different languages so you can receive that on the mailing list. That's just a heads up.


GEOFF HUSTON: Geoff Huston same affiliation - or lack there of. I would like to respond to Jonny. I think what he's saying is a complete and total aggregation of the responsibility and role and quite frankly that worries me a lot. The time to talk about what you would like to happen, what you would prefer to happen, what would be the thing to happen, what would be the best outcome was probably in 1995 when we were trying to figure this out. Because then we had choice and time. Now we have neither. It's actually a really nasty situation because this industry has wedged itself really badly. We're addicted to the continuing supply of a resource that we know is going to run out before our addiction stops. Bad things are going to happen. What's APNIC's role? Close our eyes and walk away?

Or understand that realistically, what we used to do was two things, not one. We used to - and still do for a little while, hand out addresses in IPv4 on the basis of justified need from the great big unallocated address pool in the sky. And just as importantly, secondly, maintain a registry of where those addresses are so that those who have addresses can effectively demonstrate that they are the unique holder of those addresses. And secondly, as well, with those who feel they need to have that information in order to run their routers properly and make the packets go where they were expected to go. In other words, Jonny, make the Internet work, need to have that information. If we walk away from this as you are proposing, we walk away from the Internet.

I mean not just walk away from it, we kill it, we kill it. I cannot as an individual inside this community watch that happen and say that's an outcome that I personally think we should do. It's wrong.

As APNIC, we should do what we can and no more. Let's not run a market, let's not run a trading house, let's not try to do any of that. Let's try to run a registry that across that rather difficult period, at least tracks who has what so that routing will work and packets Will flow, because that's responsible, that's doing what we can do and not overstepping our responsibility, but doing What we can do. Others will have to do things too. Yes, there's uncertainties, yes. But within what we are, a registry, we can at least do that. Or we can listen to what you're advocating in which case we might as well find the next plane home.

JONNY MARTIN: Jonny Martin. Just a clarification before and speaking from a small member from New Zealand, not as New Zealand. I think we need to preserve the registry function, yes. But if we're seen to be sanctioning the transfer of addresses which are going to be bought and sold and governments are going to get interested, getting very interested, particularly if and as prices will go up and there will be no more addresses and I just think that at the moment, v6 is wide open. There's nothing limiting us to what we can do there. As soon as the governments do anything with v4 and think taxes, think property rights potentially, that sort of thing, the same happens for any numbering or naming space that we think of in the future. So yeah, we're stuck between a rock and a hard place. Yeah, I think transfers are going to cause more issues than we think.

RANDY BUSH: Jonny, don't leave the mic. Two questions I have there. What is - I believe, the issue of regulation within economies on what my local Government is going to have to say in negotiating with people from the Government is the intent behind Maemura-San's desire to be able to work his internal Japanese process and implementation. OK. And I think we have many economies represented in the region. APNIC trying to solve everybody's problem isn't going to scale, so everybody is going to need to go home after this and spend the next while working it out with their local justice is dictions, and I believe what I'm hearing Geoff saying - excuse me, I'm just trying to sum up discussion to get a little clarity. Geoff doesn't like me speaking.

GEOFF HUSTON: I'm sure I can edit the summary up.

RANDY BUSH: I think what Geoff is saying is that yes, that's all true. So past transfers and policy and everybody has to go home and figure out how to implement it, just like APNIC does. Geoff has to go home and finish the RPKI stuff so he can do it technically, etc. The second point, Jonny and if, and I would like you to speak. Is that if you believe that transfers and the market place will bring Government action. The fact that there are transfers and it is a marketplace does not seem to change that. And in fact, my fear is that a black market is far worse politically than a white market. All right.

JONNY MARTIN: I guess the word you missed out there that I mentioned earlier was the APNIC sanction, or RIR sanctioned transfers. Now, if we can come to a point where we can maintain the registry function without sanctioning what appears to be a market creation, that is a position, but I'm not sure how we do that.

RANDY BUSH: Can we let the Jabber room inject?

OWEN DELONG: A person named ASJI says, "I strongly support the idea that the registry tracks what's going on. Transfers are taking place, the database is of little value as it gets more out of date as it will as transfers accelerate".

DAVID WOODGATE: Raising the issues that I specifically wanted to talk about here. And I think there's a lot of danger in equating the question of transfers with the creation of trading. I think it is certainly important to recognise that enabling transfers may indeed have a basis for trading and although, as many people are saying, there may well be a black market emerging as well. I have no personal experience of that. I do not know that. But I think it is very important to emphasise that the proposals before APNIC are not about trading. They do not, they do not believe that the proposals endorse, explicitly endorse the trading in any way, shape or form. That this is purely a question about our members entitled to transfer addresses between each other.

And whilst everybody is focused on the transfers, there may turn out to be other reasons why such a mechanism is beneficial. Therefore, we strongly believe that whilst as I said, it is up to each and every important in the room and on the mailing list and across Jabber, to think about whether some of the implications have got to arise. Nevertheless, I do not believe that adopting these proposals at APNIC is in any way, shape or form going to be endorsing explicitly the monetary markets associated with addresses.

JONNY MARTIN: Just to add to that then. Put another way, that doesn't refer to markets or anything. The only way we're going to move away from IPv4 is if we make IPv4 addresses irrelevant. We haven't had much luck in doing that in the past ten years, but allowing for transfers for them more so they can only take away from actually an IPv6-based Internet which is not a good thing.

RANDY BUSH: I'm sorry, Jonny, but I'm an ISP. I can not promote IPv6 by punishing my customers. My job is not to punish customers. My job is to deliver their packets. And you know how strong with IPv6.

JONNY MARTIN: How many IP addresses do you need in the next five years?

RANDY BUSH: You know, I'm going to do what I can to give the best service I can to my customers. I have to.

JONNY MARTIN: It's a finite resource we've got here.

RANDY BUSH: I understand. But... Martin.

MARTIN LEVY: Martin Levy, Hurricane Electric.

RANDY BUSH: Excuse us, by the way. We keep looking at the screen and having grim looks on the faces because we have the hardest time hearing the speakers in the room. It's not your fault, it is the sound system and how it is working.

MARTIN LEVY: I'm about to say what you just said. As an ISP, that is extremely focused on v6, we know full well we have to continue in a v4 part and, Jonny, I don't know how long that date is, but it will be longer than five years answer. But that said, Martin Levy speaking as a capitalist, it's a whole different story. If the RIR is a registry, and not a brokerage house, then the job of brokering addresses should be left for those people who are brokers. The job of the registering should be left to those people who are registers. And here's the hint - if you make the pricing the same as the registry charges - in other words take the bottom out of the market, you will find that more people will be willing, whether to use the ARIN methodology at the moment or maybe the right methodology, they will be willing to go through a non-black market method.

They will go through substantiation. They will go through a registry, and the brokers will actually end up with a price list that matches identically the RIR in which the address space is sitting. You are a registry, brokers are brokers. I'm a capitalist, I like to make money but I see places where that money is not easy to make.

RANDY BUSH: I think it is notable that there for instance have been proposals in the ARIN region where... this is what Martin is I think alluding to is the proposals in ARIN where the registry of the broker between the buyer and the seller and that proposal has not been made here.

MARTIN LEVY: I understand, so there may be a special case. Anybody can be a broker and it could also happen to be the RIR. So be it. But that's a different function than a register.

RANDY BUSH: I don't think anybody has been suggesting that APNIC act as a broker.

MARTIN LEVY: Right, I understand.

RANDY BUSH: So, let's let Jabber because they're disadvantaged, so they get first.

OWEN DELONG: OK. Terry M says that if APNIC is not seen to address the transfers and they're on track, that the RIR system holds, no-one will be able to stand and say, they have this IP and the governments will replace the bottom-up process. That is the risk. IPv4 addresses will never become irrelevant.

Nesh says stopping transfers will not always be possible. Allow it publically and this allows it to be followed with easier control and transparency.

TOM VEST: Tom Vest, consultant for a European registry. What I hear from Martin is that, does that mean that he's going to make his address space available to aspiring buyers. Actually, the reason I came to the microphone was to identify what I think is a common thing and what almost everybody has spoken about today is the essential that we must survive registration function, and I think - I don't think anybody has come to the microphone suggesting anything other than that fact. Suggesting anything other than the fact that that is their priority to preserve the registration and function.

The question simply is - is it possible to preserve that function? Is it conceivable to preserve that function under the reality which is created by the transfer market? I think the relevant analogy with the Internet routing registries, I think at one point in time can be perceived to be useful and if everyone continued to participate in a timely basis on an ongoing basis, they would still be incredibly valuable tools. But the decay of value of those tools was very extreme. It only takes a few people to decide to stop participating for a much larger number of people to decide on no value at all and this system is self regulating.

So it seemed to me listening to Jonny that he wasn't espousing any different goal than anyone else and that the registration function is paramount, the question is, how do you do that? In an environment in which people will do, just as you said, Randy, will squeeze and scrounge and do whatever is necessary in order to satisfy their customers in an environment where not everybody's customers can be satisfied with v4.

RANDY BUSH: Excuse the people at the mic. There's somebody from Suva, Fiji who wishes to speak. I'm told there's somebody from Suva who wishes to speak. Can you hear?


RANDY BUSH: I'm sorry, the problem is that we can't understand you here. I think there are two alternatives. Speak in a higher octave, speak high! That might make it through, or if there's Jabber or if there's a high woman's voice or something to try to get through with the technology.



SPEAKER FROM SUVA: Is that better. Hello.

RANDY BUSH: We can hear you a little better. Now, it's the woman we can hear. Can I go to Jabber.

OWEN DELONG: It sounds like what they're getting is long delay feed back, if they can turn down the speakers, it may help.

RANDY BUSH: Did you hear that.


RANDY BUSH: That sounds good.

SPEAKER FROM SUVA: Our content is that we would like to vote for the transfer of the v4 blocks the infrastructure in the v6 and if we have the transfer option, it will give us a bit more time to migrate our infrastructure to be ready for IPv6 and since most of our networks are small, the v4 addresses are quite adequate to cater for what we need. Plus the customers and the equipment are also limited to v4 and yeah, so the transfer option is quite useful for Fiji and I think for the Pacific region, especially the small Islands. Thank you.

RANDY BUSH: Thank you. I think, Philip.

PHILIP SMITH: Philip Smith, Cisco. I want to address Jonny's concern about Government involvement in transfers being of concern and so forth. Government is already watching what's going on here and other organisations are watching what's going on here. May I remind folks that we're kind of in some sort of an economic crisis, recession sort of thing because of the industry and paying attention to what's going on. So I think that the people here have a duty to watch what's and try to facilitate - not facilitate, but try to ensure at our level that we can still register or document the address space of the users. Tom's example of the Internet routing registry is slightly different because the IRR was a voluntary thing not something that had to be done to ensure that the Internet function you had.

So yes, the IRRs probably are less useful than they were when it initially started off. But the register of information for v4 address space, it is vitally important that that is maintained and we should work hard to ensure that it is maintained. I agree there are implications. There are implications whatever we do. There are implications for whatever we do not do. But we should try to stay at least with what we understand and acknowledge that the transfer is happening anyway and try to work within those particular boundaries.

KENNY HUANG: First of all, just trying to echo what Randy just mentioned, to support the proposal with the title of relevance and difference, so maybe just here to think about it, and secondly, just try to remind there, the significant address block either preserved with the address holder and without any legal binding with the RIR. And I think it is time to think harder to have some sort of sentiment to measure if we can unlock it.

Make sure that and go with some extent for another three or five years for the community. And as everybody probably knows, to deploy IPv6 and IPv4 it is mandatory. So basically, as long as we want to promote IPv6, we need a certain amount of IPv6 address blocks. Thank you.

RANDY BUSH: OK, the Jabber room.

OWEN DELONG: Nesh says, the Suva speaker's comments are very relevant from Fiji as we have only four ISPS.

NURANI NIMPUNO: I thought that I would comment on James Spenceley's comment about, this is not happening now. I think we've seen many examples of it actually happening now. I think the other comment I would like to make is that the ideas that we want to put in place with the frame work before chaos ensues, not after. And just following both Jonny and Philip's comments about Government interest, and someone who has been involved with the IGOV discussions and having to talk to governments, and who actually uses the RIR system as an example of a bottom-up transfer process that works. I think if the RIR fails its most basic responsibility of ensuring proper registration, the governments will be even more interested in taking over the process. So I think that that is - if we fail in that, then we fail as a system.

RANDY BUSH: We're living in a world where the administrative responsibility for the financial system has created a global disaster. Jonny.

JONNY MARTIN: If we know we need to preserve the registry function. If that's what we want to do, then we just allow transfers, but we should only be looking at one thing. Who gets the space now? Not do they deserve it? Should they be allowed to get more space later on?

RANDY BUSH: That's one of the items on the agenda.

JONNY MARTIN: Yeah, so maybe we'll wait until later questions to look at that. But I would be happy with a very, very simple transfer, a simple one there. But it's got to be one where we're only looking at the new home of the addresses and absolutely nothing else.

RANDY BUSH: We have a couple of hours to simplify it.


RANDY BUSH: I believe names and then... pardon me, I think Jason is first.

JASON SCHILLER: Verizon. I have four different points that I've been waiting to bring up. The first one I want to talk about is that there is the transfer policy. I'm wondering in the short period of time since that's been implemented, if you can tell us about how many transfers you've seen? I'll talk about some of my other points if you want to research that a little bit.

The second point I wanted to bring up was in response to James's comment about the question of how many people in here have been approached to - have been offered to buy addresses and I want to ask there, and ask how many people here would be interested in purchasing addresses if they were on the black market versus if they were on the white market. The concern is that if they're here, there are a number of legitimate companies that would be unwilling to participate in the black market because of obvious moral and ethical quandaries attached to it.

RANDY BUSH: I think I'd rephrase that question. Let me ask that just for the giggle which is - how many people here believe that they would not buy on the black market, but would buy on the legitimatised market? And some companies believe that they would.

JASON SCHILLER: So the point of that is that a white market might encourage transfers more than a black market would.

RANDY BUSH: Is that good or bad?

JASON SCHILLER: I just want to throw it out there. The second point is something that Geoff Huston talked about and he says today the RIRs perform two functions. One is making sure that there's justified need and two is the registry of documenting everything and keeping records accurate. And he said that we should still continue doing number two. I'm wondering if that seems to suggest that we shouldn't still continue to do number one, or are you just not addressing that point at all. I don't know if you want to comment on that. Geoff Huston, did you catch the question?

GEOFF HUSTON: Well my point was, you can't allocate IPv4 addresses on the basis of meeting folks' needs when we can't do that.

JASON SCHILLER: No, the question is, the point here is when transfers occur, we should document them because at least we can still perform the second function of the RIR, and a second function is should we still try to perform the first function and make sure that the receiver has justified.

GEOFF HUSTON: I'm going to do something with prop-050. Am I allowed, chairs? I'm going to do something naughty and refer to prop-050. Am I allowed to do that, chairs?

RANDY BUSH: Certainly. But I just want to inject one thing which is that the chairs, or mainly I am in gross violation of my responsibilities and we've drifted past tea time! And I don't think that the lines at the mics are going to get any shorter, so my suggestion... people are free to go and get tea and bring it back. And we'll just keep working through lunch. Sorry, Geoff.

GEOFF HUSTON: I was going to say with prop-050, there was no demonstration, but as Jonny Martin appears to be advocating that all that appears to be involved in the proposition was to maintain the registry function and not to get in the way of any other aspect.

JASON SCHILLER: And my final point is that in response to something that Jonny Martin has said. And basically what he said is, we should not support transfers because that would basically kill the IPv4 Internet. What we're saying is that, is it more important to - how can I phrase this? If we don't support transfer and we kill IPv4, that's bad. But if we do support transfer and we kill IPv4 and we extend the property rights to IPv6, that's worse. So, I'm surprised he hasn't said - we're faced with the proposition of killing IPv4 or the proposition of killing both IPv4 and IPv6.

RANDY BUSH: I'd like to interject, something from Jabber is that somebody has admitted to acquiring a /16 in exchange for a case of white wine. So that sets... that sets a price. He hasn't spoken to what kind of white wine of course. If it is some sweet German thing, I mean. I think it was James.

JAMES SPENCELEY: I just want to...

RANDY BUSH: Could people at the mics please help in keeping order and I would like stop the mic queues now. No more people at the mic queues because I'm getting beat up in Jabber says we should take a break and then hold and come back to the mic after the break.

JAMES SPENCELEY: I just want to comment, I understand that people are interested in transfers and there are cases of transfers, but what I'm hearing is that the sky is falling. That we can't trust a single address block out there. And basically, my concern and I have a number of systems is that as soon as you allow the white market you create the value, and those comments are that allowing transfers doesn't necessarily create a market. Absolutely it does. We know that there is demand there. We know that there is supply there. Legitimising it creating a value. Now what happens when the value becomes greater than an ISP's profit in a Third World developing country? Are they going to simply pack up stumps? Close the ISP and sell the address space?

What are the actual effects for the end-user of it. Let's look at registry function, but the end-user. Are we going to decrease competition. Are we going to crease ISP from functioning from having address space. These are my concerns.

SEICCHI KAWAMURA: I've been getting instant messages asking is the JPNIC community against this policy. So I think I needed to clarify. What Akinori Maemura was stating was that JPNIC has a lot of things to think about and to consider. But that doesn't mean that the JP community is against this proposal. And in asking that question, all I can say is that it is not a full consensus, but I think that a little bit more people are for this policy than against. I'm speaking for the JPNIC community, we had an open policy meeting a couple of weeks ago. And as a JP community member, I would like to have discussions and hear opinions about what other NIR members think about on this topic. Thank you.

RANDY BUSH: Yeah, I think the point is that this is going to be significant work. And that's just life. It's going to take a lot of work.

SEIICHI KAWAMURA: But that doesn't mean that we're against this policy.

RANDY BUSH: We have something from the Jabber room. Why is there a black market? Good things are not allowed by policy that are happening. Then why do we believe that when we have policy, that people will follow it?

GEOFF HUSTON: Geoff Huston. It would be nice to think that this room could determine the entire Internet and its fate. It would be nice to think that everyone does precisely what we say about the entire Internet. They don't. There are now 1.9 billion users of the Internet out there at the moment. The growth is phenomenal. We wanted to take on the telephone companies. We wanted to show them that there was a different way of doing communications globally. We wanted to say that it would be cheaper and more effective. We wanted their role. We have it. And there's a price. Because if anyone thinks that governments aren't involved, they're living in the past, they're living in an age that used to be not today. Because we are the global public communications infrastructure.

We support trillions of dollars, governments are involved every hour of every day. So is industry. So are you, so am I.

So everyone, you're now in the middle of this. You can't keep anyone out, they all have roles. And we alone don't determine the few. We play a role, not the role. What's the responsible thing to do? Not what we would like, what is the responsible thing to do? What is the right thing for a registry to do? What's our enforcement capability? What is natural? What is unnatural? Can we direct a multibillion dollar industry to make investments when their shareholders don't want them to? No, we cannot. What can we control? What can't we control? Ultimately when it works all the way down to the basics, we're in a policy meeting about a registry function. And what we're talking about is admission to that registry. What policies allow entries to go in. And my view is that we do the world the best job we can by registering reality, where those addresses are.

How they got to that stage, once we no longer are capable of handing them out is none of our business. Other people have a legitimate responsibility to worry about that question. But as a registry, we're making sure that once the parties have reached that point and that action has happened, reality has changed and that we as a registry record that reality. Why? Because as I said before, the Internet needs it. Packets need it. Routing needs it. And if they don't have it, you haven't got the network. And all those 1.9 billion people and all that money and all that value have a problem. The best we can do is not to be the problem. Don't overstretch, don't try to do too much, Jonny, just do what we need to do. Be the registry function. So when transfers happen, recognise them as reality.

Because realistically what I think this is about, there's a whole lot about what this is about, but underneath it all is what we need to do as a registry. Thank you.

MASATO YAM ANISHI: Just going back to the question of wanting to sell address space. I guess my answer for the question is, we don't want to sell address space. However, we want to transfer our address space to with the company and let me explain hour strategic example.

My company itself is provided a fixed access service that with our subsidiaries, providing a data centre service. Since these companies are different companies, so we only want to transfer address space from fixed access company. Another example is while the subsidiary is providing server phone, and it is slow compare to fixed access, we may want to transfer address space from to the Sera phone company. It is the big responsibility.

RANDY BUSH: From the --

OWEN DELONG: From the Jabber room. Johnson says it is a supply-and-demand problem. I support having the registry, just like the land registry in my country, you can get the address information. You can know who and what addresses trade and how often. Adjust the space and time for the frequent trader broker. Provide for the needs of the real users. Andy says, 'Geoff hits the nail on the head".


IZUMI OKUTANI: Izumi on behalf of the JP community. And I really apologise for the confusion that JPNIC and the community is causing for the people. But I want to emphasise that we really had strong support from our community to move this proposal forward. We had some discussions in November. We had an open policy meeting like this. We had about 100 people participating and about 80% of the people supported that we should go ahead with the transfers, so that's how much people in the community actually support this proposal and they can't all be here present, because it is, they also have language problems but then you really have quite a number of people who can't be present supporting this proposal and then the right to address some of the concerns that have been raised.

I think some of the people are asking, how many people are actually going to go ahead with the transfer proposal at the moment? And I assume the hint is that maybe the problem was not happening at the moment. Why do we do this right now? But I think it quite obvious from some of the comments that have been expressed from ISP that they actually feel the need to be able to obtain address spaces even after the APNIC pool runs out. So it is not apparent at this very moment. It is, we surely need it at some stage in time and also to make the transfer very smooth until we are able to deploy IPv6 properly.

FILIZ YILMAZ: From RIPE NCCC. I want to go back to the first question from Jason which was a while ago, but it is still open. The question was, as far as I know, and can remember, was that if in being the first Internet region and we have reached consensus on transfers. If any requests have been received so far, well December 2008, the proposal became a policy in our region. It was accepted. And early January, so more than more than a month now, it was implemented by then. As for the numbers, we have received no requests so far. So nobody approached us to be affiliated under the new policy for transfers.

MARTIN LEVY: Martin Levy at the microphone before tea break. My space is available for $4 a sentence because this is a limited resource. OK, after 2011 or whatever the date turns out to be, there will only be transfers. There won't be anything in the RIRs, so having a policy ahead of that date will make that date a little bit more smoother. Immaterial of black market, white market, ebay or whatever.

RANDY BUSH: OK, I think we should take a 15-minute break now. We're only three quarters of an hour late for it. And I do have a question - on this particular part of the agenda, are there more people who wish to speak? In other words, if we opened up the mic queue right now, are there people who would go to the mic? I see no hands. Well, then before we go to break, I would like to... oh! John, was I wrong? OK. Then before we go to break, I would actually like to see what the feeling of the room is on this subject. Unless somebody has a problem with my doing so.

OK, how many people in the room support the generic idea that APNIC policy should allow transfer of IPv4 space? Please raise your hands.

By the way, we're not counting.

How many people feel that APNIC should not allow transfer?

I think we move along. So we'll take a break.


By the way, there are a number of people in the Jabber room who put their virtual hands up!

(end of session)

Thursday 1130-1300

RANDY BUSH: We are going to make a start. Thank you for getting rid of the music. Jian? Jian? There is a missing co-chair. We have a referee time out. Ah. I guess we can resume. Yes. OK. woops. Of course, like to show you this. This is what we just finished. So we did have a consensus that we should allow transfers, which was amazingly brave of us, considering we haven't defined one. We will now go through the exercise of the remaining - I think it is seven points to try and focus on exactly what we need at the detail level. The first point, do you want to lead? The first point is what should be the minimum size of address space allowed to be transferred. Here are two examples. Anything down - that is a prefix that is no longer a /24. And the proposal has been whatever the minimum allocation size is in terms of policy, at the time of the transfer request. Those have been, those are the two wordings of a couple of the proposals. We are free to actually come to any consensus we believe. The microphones are open.

GEOFF HUSTON: Thanks, Randy. As one of the proposers here, I'd like to explain very quickly why I have proposed a /24 and why I would not go to what they call a minimum allocation size. My thinking, in drafting that, was very soon there will be no further allocations, they will be over. In five years' time, when we look back and say, "What is an allocation?", we have no idea what we will be talking about. So to tie the future against some of the constrictions that exist in the current address allocation system didn't appear, to me, to make an awful lot of sense. Then comes the issue of what should it be? My temptation is to say absolutely nothing. But I do observe, at least pragmatically that today what happens, what gets routed is a /24 or larger. So I put it in but if you said, "Geoff, would you die if it was changed to any other value?" My answer would be no. I don't hold any part of that prop-050 dear to my heart. I am not overly concerned one value or other would mean an awful lot at this juncture.

OWEN DELONG: Earlier, a Jabber participant said he felt the /24 was the appropriate size. I would reiterate that from Jabber.

PHILIP SMITH. As one of the co-authors of one of these proposals, I have no great feeling either way as to what the minimum should be. And /24, realistically today, is the minimum that is routable so it seems entirely reasonable to me to go over it. But the majority of the group wants something else, I am happy with that too.

OWEN DELONG: Terry M states, 'The minimum allocation should be what we use. When the last /8 is reached, then a /24 is fine'.

IZUMI OKUTANI: We discussed this in Japan, ISPs, I think it pretty much reflects the opinions of people who have spoken. Because the opinions are very much split. A lot of our operators preferred option B which is to stick with the minimum allocation size so that although, the actual size that is being routed and what is defined in the policy is not the same, we prefer to see what is going on and if it does cause problems maybe consider split it into /24s. That was spoken from the operator side. But there are others saying not a very realistic size. People prefer to make transfers in a much smaller size. If we really want to make this policy work, it is better to allow a /24 to start with. Opinions were split. I am just sharing the details of why people are taking each position, just for your information.

TOM VEST: Curious about the rational for having any size restrictions at all, if there is a policy reason for having a larger rather than a smaller, then it will be interesting to have that. Articulate it clearly because it is important because the very question of restrictions like this seem to me to be at odds with the logic that dictated the last question, which is to say, with the co-chair, asserted that people will be scrounging and doing whatever is necessary to satisfy their customers. If it means doing something which is currently contrary to policy, doing transfers they will do it and ignore policy. I am not sure why the logic would not apply to any restrictions, for example, about address sizing including participation in the database itself. The points of clarification on those two matters would be useful, I think.

RANDY BUSH: I believe why people are discussing a minimum allocation size, a minimum transfer size, are because of hearing concerns about routing table explosion. Does that answer your question, Tom?

TOM VEST: It clarifies it but again it begs the question, routing table explosion is something that people generally perceive to be an ecological or environmental factor. They think that generally speaking, their routes are legitimate, important, valuable, important to the customers. The other guys, those routes are wasteful or misinformed. In fact, my routes are important and if my routes are gone in fact I need to have address space, I need to parser it up in smaller quantities and purchase it. Well, my needs are legitimate. It is a rational but it doesn't seem to me to suggest how any restrictions like this are going to have any effect.

RANDY BUSH: But I am not advocating for or against minimum size. I don't care. The - we have a very long - as in measured in decades - history of registries trying to balance allocation size versus need. Whether it is right or wrong, we have that history and that cultural momentum. I believe the people pushing this particular point are merely following history, they are not inventing something here. We may question it. Question whether it is reasonable as we run out of v4 etc. But I am saying I don't think they are insane, they are following history and as I said, I don't really care which way it comes out.

TOM VEST: That history was in the contest of a mechanism that made policy decisions possible. But it seems to me we are entering a different kind of phase. Thank you for the clarification.

JONNY MARTIN: I think if we want to follow through with changing the least, so simply maintaining the registry function we shouldn't be concerned at all with minimum size. It is something we just leave up to those who are transferring addresses to sort out. Of those two options I would go with a /24 but the third option is make no mention of it.

RANDY BUSH: I think Geoff was next.

GEOFF HUSTON: Only a microsecond. I would observe, over the last eight years in the routing table itself, the adherence of parties to advertise only what they receive from their RIR is not just optional. It is ignored universally. Part of my motivation is simply writing a /24, it is what people do. They just slice and dice their allocations into /24 and enjoy themselves. It seems to me if that is what folk do out there, then that is what folk do. But I keep saying, like Randy, I am not unduly fussed.

JASON SCHILLER: As a large ISP, I am concerned. The more specific of an allocation you allow to transfer, the greater the impact it is going to have on the routing system. With respect to deaggregation and the worst possible solution would be to allow even more specific /24s. If people manage to buy a /28 and then come to me and buy service and hold a lot of money over my head, my company will probably force me to put it in the routing table. If I want my peers to listen to that prefix, I probably have to listen to prefixes of that size from my peers. This takes us into a very bad direction, especially considering the fact the prefixes are likely to be sprinkled throughout the routing table and it will be very difficult to distinguish a /28 that is transpired. Because I can't get it to it because there is no covering aggregate as opposed to someone who wants to deaggregate slash 28s for traffic engine earring. It brings me to my second point legitimate routes versus illegitimate routes. It is important to distinguish between multiple non-contours deaggregates you can't role off. If you make the policy support small siders, you will end up with a lot of small aggregates. If I have a need for address space and people are selling /24s, I will have to buy a number of /24s and announce all of them because they are all aggregates rather than getting a /20 or a /19 or a /16.

RANDY BUSH: Jason, could you clarify because I am not sure I understand what your position is. You are clearly saying anything longer than a /24, please, please, no. Can you live with a /24? Or you prefer not.

JASON SCHILLER: I would just suggest the less specific the better. If there is a reason to change the minimum size locally for this policy and for allocations and assignments made by APNIC, I can certainly understand that. Having a different minimum is going to lower the bar of the routing table to whatever the least common denominator is. So the less specific the better.

TOM VEST: Quick point, transfers will create - on registries or allocators the output ratio will be flat rather than spaces. The possibility of aggregation resulting in this will be certainly likely to be like a majority of things which results in this. I am afraid aggregation was a non-starter. But a point of information, I think about insight about allocation size which is associated with an allocator which is bypassed and actual content of the router table, if there is a substantial gap between the allocation size and what is in the routing table now, some people have perceived that to be a criticism of the success of the system, but a criticism like that is say because people drive 10 or 15 miles above the speed limit, not only should the speed limits be abolished, they should never exist. But people would drive roughly the same speed, which is inherently false, even if the benchmark, which is established by a policy institution.

TOM VEST: May I speak. I apologise for having to do this again.

RANDY BUSH: We're up.

TOM VEST: There isn't always going to be some gap between a policy defined maximum or maximum deaggregation and what is observable on the routing table. The reason that the gap is the size that it is and not infinite, is not vastly larger than it is for days because the allocation mechanism we have now cannot be bypassed when people have to pay attention to the structures. When that ceases to be true, this goes back to the original question. When it ceases to be true, it is not clear to me what mechanism would make any minimum size have any bearing at all on the actual transactions. Actually with the behaviour of the people that want to sell it to the market and the people who especially, for example, who will be facing dramatically increased prices and, in fact, who may only be able to afford a /28 or a /30 in the market.

So I'm not really sure, I think that there's a general or a broader question of what do you call it, moral hazard. It's useful to be mindful of what can be enforced and if we're moving into the environment where there are no policies which are enforceable, then we should take that back.

RANDY BUSH: Tom, just a sec. In answer to this question, has simply yes or no.


RANDY BUSH: Are you advocating shorter prefixes than a /24?


RANDY BUSH: No, you are not?


RANDY BUSH: No, thank you. Jason?

JASON SCHILLER: Verizon Business and I apologise to continue to have this extended conversation with Tom. Tom, I hope you're listening. I have two points, the first point is yes, I agree with Tom that all of these transfers are going to be flat-routed. If you assume that there is a fixed amount of space available for transfer, is it better to have that fixed space flat-routed as /20s or /24s and that is a significant difference in the number in the routing table. The second point I wanted to make was with regard to more specific to /24 and although I don't know if it has any muteness at this point because it doesn't sound like Tom is advocating more specific than /24, but the point that I wanted to make was if the policies is to only go with policies up to /24 and no shorter, when someone comes to me with a prefix shorter than /24 asking me to route it, it makes it easier for me as an ISP to make a justification to my company that we should not do this and this is bad for the Internet and there are policies against this. So that makes it a little bit less difficult for me to avoid routing for specific than /24.


MASATO YAMANISHI: Before making my comment, let me clarify what the exact minimum size of the proposal is. I think that the 4.1, prop-067 says, I think that the option A and option B. And in option B, it says one or more areas of unallocated block is smaller than the current minimum APNIC allocation size. I think, I guess it means that class B of peer address still with be spread to multiple 2 and some of them can be transported. Is that the correct understanding?

RANDY BUSH: As one of the authors of the proposal, Philip I think you can speak.

PHILIP SMITH: So that's why you're looking at me, the answer is yes.

MASATO YAMANISHI: So I think that prop-067 still has some there, so I'm not sure that there is or was a big difference between option A and option B. So I support option A. Unless there is a significant input in that.

RANDY BUSH: Actually, those weren't meant as exclusive options. That was... but, the two proposals present - those two proposals present two different possible statements in the space. There are other possible statements, for instance, somebody could come to the microphone and advocate, you may only transfer an entire allocation. I'm not advocating that or saying it is crazy, those proposals are two points in the space. Philip and Jason can I interrupt for a second, this is not trying to receive consensus in control of the point, I'm just trying to see what people are think. I'm sorry, Jason.

JASON SCHILLER: Actually, I'm just going to ask the chair, it seems like maybe we're trying to ask a general question, should the transfer policy consider limiting in the deaggregation system? Is that what you're driving towards?

RANDY BUSH: Well, that's something that's good to ask. That is not the specific question I was thinking of. Because the problem is that, you know the people writing proposals and the people interested in it, you know Geoff, Philip and myself and whoever else wants to join this us up to one full table at lunch are going to try to come out with a combined proposal that lays it all out so that we can go to tomorrow's general member meeting and say - this is the proposal which reached consensus at the Policy SIG. And so it would be nice if we had a clear consensus on this point, and what you're proposing is that, you know it should try to minimise impact on the routing table. I'm not sure that that is something that everybody will think is clear enough. Philip, you were next.

PHILIP SMITH: Yes, it was just more about minimum size to and fro. And it is really up to ISP what they want to allow into the BGP table within the network, so sure, if someone transfers the /32 and wants it routed, I don't know how far they're going to get. And even I know some ISP, even if you announce a /24 to them, there's no guarantee that you're going to get it either. And so, it is really a path to say, look, we can't go telling ISP what they should and shouldn't accept within the network. I don't really want to try to do that. But we just have to realise that operationally, there are going to be some ISP that don't care at all and ISPs like Jason who are really, really worried about the size of the internal BGP that they're going to have to carry around in the global backbone.

MASATO YAMANISHI: About... and as everybody knows well, very big and traditional company has large PR address space, and while talking with one of them, they said that yes, certainly they are not using almost all peer address space, but however it is difficult to return whole address space and since they're a traditional company, so actually, they don't know exact status, which equipment is using which address.

So some equipment in factories may be using - I'm sorry, some equipment in factories may be using some address or registration. So I think that it is quite difficult to transfer whole allocated space, even and especially in PI address case, so I can not support either which limit to... I can't support the idea which means PI address can not be spread and transferred.

RANDY BUSH: Let me try this. This is not binding, this is not a decision. I'm just trying to understand if we could even come to a decision. I'm going to ask, how many people prefer /24? How many people prefer - I'm going to list what I'm going to ask so people can decide where they spend their money. How many people prefer current allocations size, ie a /22. If you can't break an allocation, if you have a /16, you have to sell the whole /16. And how many people think there's another possibility I have not thought of? Ah, here comes Geoff with other possibilities.



GEOFF HUSTON: Just one other, none.

RANDY BUSH: OK, none, /24, now current which is /22 and allocation size. Jason?

JASON SCHILLER: Verizon Business. Question of clarification. If I had an aggregate, say a /16 and a small piece of it was in use, would there be available space for me to number into in order to give up my aggregate? Because that could make a difference.

RANDY BUSH: Let me see if I can phrase it. Would the RIR give you an allocation because you wanted to sell the /16 on the private market?

GEOFF HUSTON: (Inaudible)

JASON SCHILLER: Let me rephrase it differently. Would some of the space in the last /8 be reserved to help people renumber and return large blocks of address space in exchange for small blocks?

RANDY BUSH: Good question which nobody has thought of. Somebody take that note. OK, I just want a feeling in the room. How many people think there should be no limits? Can sell anything? None. How many people advocate a /24. Maybe eight or ten or something. How many people think a /22ish size would be tasty? Only three or four. OK. And how many people think you should not be able to break up a block? Wow. So three. OK, so people seem to kind of think that it should be an aggregate. It should be - people kind of like about current policy. OK, let me ask another question. First, does anybody want to submit? David?

DAVID WOODGATE: Could you please restate your assessment of that.

RANDY BUSH: Your assessment of what you just saw? What I saw was small things for everything and no large support for any particular one, but maybe twice as much support for the current aggregation size, /22ish.

DAVID WOODGATE: Could you possibly recheck that?

RANDY BUSH: Redone, you think I'm remembering incorrectly.

DAVID WOODGATE: I think you might be.

RANDY BUSH: Thank you, let's go again. My Alzheimer's device is probably failed. How many people think there should be no restriction? How many people think it should be a /24?

OWEN DELONG: One from Jabber.

RANDY BUSH: David is right, my Alzheimer's device did.

OWEN DELONG: Just one from The Jabber.

RANDY BUSH: How many people think it should be current allocation size? And how many people think it should be no breaking blocks? OK, so indeed David is right, my Alzheimer's device did fail and more people seem to be in favour of a /24. Let me ask this question - is there somebody who - could the people who would be really unhappy if it was /24 please raise their hands. In other words, does somebody feel strongly that no, it should be something other than a /24? So we suggest that we move ahead with using a /24.

OK, I think we reached consensus on this that one. No, it wasn't the same, there was at least twice as many on the /24 as the other. I'm willing to call it again.

GEOFF HUSTON: I don't feel that we're seeking consensus on each and every point, I think we're trying to guide policy.

OWEN DELONG: So from Jabber, there was one additional vote from /24 that arrived late and a comment from Terry M. "Nope, not a killer from me". Barry Hall says, "OK, I give in."

RANDY BUSH: So Jian is objecting procedurally, and let me solve the problem by saying - we'll try again. /24, does anybody have a serious problem? Does anybody feel that we can not reach consensus on a /24?

Thank you, that is consensus on the /24.

Moving along - the question is, currently the RIR and host masters look at an allocation request and the person requesting the - the organisation requesting the address space has to justify its utilisation and there are procedures, forms and methods of evaluation for doing so. Should the recipient of a transfer have to go through that process? In other words, should APNIC with the help of the NIRs, etc judge whether the recipient should receive that address space because they have justified its use from an engineering and business view point? Or not? Should they just - you want to buy that space, this guy wants to sell that space, finish.

OK, would anybody care to speak to that question? Geoff Huston and David can race to the microphone. He's more agile, David.

GEOFF HUSTON: Closer. I would actually have to say that this is an issue that in drafting prop-050 I would say was a critically important issue and I deliberately felt that there was no justifiable need for putting in address requirements in a transfer environment. Because we're not allocating any more, and what's going on is the two parties out there are reaching some form of agreement to move an address resource. If APNIC attempts to qualify or moderate or attempt to regulate such behaviour, we are assuming the role that we have never assumed in the past and we actually have never done before. We're trying to be a market regulator. And I'm thinking a registry, a neutral title office, has any role in also trying to be a regulator?

I don't understand how there is an issue of need when there is no guarantee you're going to get the resources anyway. It then becomes an issue you have to work on the broader environment with the market and figure out when the price and your need equilibrium. So when we go to the supermarket every week to buy the goods, does somebody at the door say, "I'm sorry, you only have a justified need for one loaf of bread this week?"? They don't, what regulates your behaviour is your perceptions as a particular of need and price, and inside the environment we are about to enter because that's unfortunately where we are, then players will work out what they need and equilibrate on price. Having a registry like APNIC and tend to try to intermediate in that would guarantee failure and position the registry in a role where it is incompetent.

OWEN DELONG: So from Jabber, Terry M says, "Depending on when it is implemented, the exhaustion as to whether you have regulation or not."

DAVID WOODGATE: David Woodgate from Telstra. There's a few points that I would like to investigate. One is that I personally agree with Geoff's position that in trying to take APNIC back as a regulator on transfers, actually saying that the transfer can only have, can only take place once it's been reviewed by APNIC is going to open up all sorts of problems. However, having said that, I do note, I believe that the right policy that has been addressed I believe actually does and has gone with that position that it will be and transfers will be better provided by right. So, there are two sides to that point already. I do have a few other things, but they temporarily escape my attention, so I'll let Izumi speak.

RANDY BUSH: Izumi-San, can I just speak. The proposition proposal from Philip and I contained the suggestion for the rejustification, and the reason we did so was what we thought... by the way, I don't care either way and I don't think that Philip does either. The reason we put it in there was that we were hearing that people wanted the justification there to keep from having a speculative and hoarding market. That was their goal. Whether this is the best way of achieving it is an open question. I think that's what we heard and why it's there. There's another proposition here also to address that same point which is - if you sell space, you can't get more for some period of time. So there are a number of things that tried to address that question, whether this addresses that question well is, or addresses it in the best way, I don't know. I'm sorry, I just wanted to explain why we're discussing this.

IZUMI OKUTANI: Izumi. So, among ISPs, opinions were pretty much split, so just like Randy has just shared with us, some people feel that we should prevent people from obtaining space without really needing it and just hoarding or speculation, take advantage of this transfer system, and for those people who feel that way, they support APNIC or possibly NIRs and checking the utilisation of the address space at the time of transfer, but at the same time, we also had the same number of people supporting and not doing the checks for the reasons that have been very well expressed by Geoff and David already.

JAMES SPENCELEY: It's tantamount to handing as much...

RANDY BUSH: Sorry, you need to say who you are.

JAMES SPENCELEY: Sorry, James Spenceley, Vocus. I'll go back a few steps. Allowing this policy to go ahead today without the requirement for justification is like handing as much address space to the richest parties in the world. Is that what we want? If we implement this policy today, we need to implement it with justification. If we implement the policy after we've exhausted the free pool, then I don't believe we need this. The last thing we need to do is to provide the richest parties in the world an ability to gain as much address space at the disadvantage of others.

TOM VEST: Another point of information, just in response to Geoff's non-regulator and the position. Well, that's in fact the role that the RIRs have played since the establishment. It's why they were established. They were established to mediate conflicting claims on a finite, a sit of finite resources both with the routing table capacity and the address spaces and the conflicting claims were between large incumbent search providers who would like to enjoy ever-increasing economies of scale and new entrants who like to have the opportunity to enter the Internet service provision market.

The quasi-regulatory system we've had in place for the last decade or longer satisfied that requirement by the combination of making aggregation possible which made prefixing possible, made it possible for large service providers to continue growing, made it possible for the very concept of IP transit as a kind of a one-stop shopping for global Internet access to be physically possible, which was a very useful service for everyone. Providers and customers as well, but the externalisation of these functions, the address registry function and the allocation for the neutral third party made it possible for them to come into the market without making it to create a new competitor. And in situations where you don't have that neutral gateway into an industry, the alternative is a closed cartel.

They're usually not good candidates for self governance and good candidates for providing good service either. So I would point out that it is not really the case that trying to make some value judgments about what the community, what we would like to see this institution do, would not be some new precedent. It would be a continuation of - I think the very wise and very wise policies established over a decade ago in which fed the foundations and the reasons why we're all in this room.

DAVID WOODGATE: Just to comment on Tom's point. The change of environment that we're about to go into of course is that up until now, the reason why the RIRs have been in a position to perform that regulatory function as described is because they've had control of that resource. They've actually held it and it's been in their database and something that they can directly manage and they had a duty to regulate and how they distributed it. Given that they're talking about a situation where basically after the exhaustion of that pool of the RIRs, the RIRs won't have that resource to control any further and then it becomes a question about - do the RIRs have an act to go and impose themselves upon the activities of the members directly, and that's perhaps the open question.

I'm conscious that for the allocations which have been made by the RIRs to members, that they've already embedded in theory, the address space has been handed out on need and demand. With respect to historical space, that obviously has had a different perspective and there may be some scope there, but I think there's going to be... there needs to be an assumption that we're not sure that there will be wholesale earning of transfers, just because I think the people who the addresses are now may lead them no many circumstances. So, I'm not sure that I believe in a Dr Evil. We thought that APNIC is 'Austin Powers'. I'm not sure that there is somebody who is going to go around the world buying up addresses and trying to hoard them, especially since I think we all believe that IPv6 is still there and we saw any activities of that sort, which I suspect in necessity are an invention and would stimulate us to drive forward more rapidly with it to avoid such issues. Thank you.

OWEN DELONG: From Jabber. I'm going to display my ignorance of the Vietnamese pronunciation, ThuThu says, 'I prefer that APNIC and NIRs only log the transfer and ensure the exact information in the Who Is database rather than become the regulator for the transfer.'

RANDY BUSH: I have received a private Jabber. That there seem to be three options. One - address requirements to be justified before and after the run-out of the free pool. Justification is required only until the free pool runs out, but not after. And three, justification is not required at any time.

GEOFF HUSTON: This is Geoff Huston. I've already addressed my comments about justifications required before, under and after, and I heard a view that said, "Well you only needed it before exhaustion but you don't need it after", and I believe the case was made therefore that you require justification before in order to ensure that some broke who would otherwise not be able to transfer would then have access to it. But before exhaustion, aren't we still doing an allocation for IPv4 spaces. If so, can you go to APNIC and have the addresses need. And I believe, unless someone in the room says no, I believe that resoundingly it is our job and until it gets exhausted, that's what we do. So even if we pass this today, it is extreme and do a transfer tomorrow because in general, it's probably going to be easier and cheaper to use the existing allocations.

The only question that I see here in this instance is, why do transfers at all and the issue about preparation and timing and signalling. Put it in place now. And get the thing yourself and if there are some folk who want to try out for real, fine. It won't be really serious because generally if you have need, you'll be able to get it through to the allocation process, but at that point, if you can justify it and if you can't justify it, why are you putting your money there in the market? So the issue becomes, if there are players there, it's a natural behaviour anyway. So what's really going on here is - why do we worry about applying history to the future when this particular case, allocations are different from what happens afterwards. We can't really stand in the way of - you know me going into the supermarket and buying bread and saying, no Geoff, you can only have one loaf.

It doesn't make any sense in that kind of environment. So this is why I would say - start as we mean to continue in transfers. Simply register what happened. And stop trying to say - well, that happened but I'm not going to register it because - that would mean that we get into the other space of being incapable of cracking reality, and that's a bad thing I think for a registry.

JOHN SCHNIZLEIN: Two points I would like to make. One has to do with what I have seen in several places, the idea that it is a minimal change to the current policy to continue justification of the size of the transfer. And I believe that that is simply misreading the reality. Today, the RIR has two functions, the recording function and allocation from a pool held from the RIR. Those can be seen as transfers from RIR to a party. That is completely different and in that case, the RIR has reasons to hoard for the as-yet-unreceived requests coming in the future. That's the way the game was played when we had lots of future in the address space. That's not what's happening today or soon. What's formally different is when there are two different parties, one of whom holds address space which needs to be transferred to another. It is simply not the case that it is continuing the RIR's function of reviewing that, because this is a third party imposition. This is what I think Geoff correctly describes as a "regulatory function".

My second point is that regulatory functions do happen in the world. There has been less regulation in the Internet world, largely because it hasn't mattered, it hasn't been necessary. It is a big mistake to take on a regulatory function when there are real regulators, because they jealously guard their power. You can tell real regulators from pretend ones. Real regulators have armies or national banks. If APNIC wants to get into the role of being a regulator, show me your army!

JONNY MARTIN: If we are going to go ahead with the transfers and stick to the role as registry, no justification should be required. Even what looks like we're about to do here, you know Geoff, the less we do the better. Any justification requirements we do put in firmly establishes the RIR's rules, roles as one of facilitating the market which I don't think it should be.

OWEN DELONG: OK, so from Barry Hall we have, "I am strangely going to agree with David Woodgate". Stop!

LORENZO COLLITTI: Just a note to an early point. If it is true of the IPv6 option, we must think carefully about allowing transfers now. Will our speculative behaviour that will place people who have lots of v4 address space in a position where they prefer to delay IPv6 adoption because they stand to make lots of money?

TOM VEST: I just want to say one more thing. OK, so clearly something now has to change. Merely accepting or rejecting v4 transfers. Neither of those outcomes will solve anything with respect to the inflection of the point we're about to reach. Regulators and regulations are just a word for co-ordination imposed by a third party. It is definitely the case that in the absence of some means to co-ordinate effectively internally when something goes wrong with a critical surface like the one that we now provide to many, many people. If internal co-ordination is not possible, someone will help us out with that. It is an unfortunate fact that some demands of the market cannot be sustainably met. OK, I think we're looking at a manifestation of that fact in the broader context in the global financial sector right now.

It may well be the case that within the industry, that this particular change seems reasonable, appropriate, necessary, useful, profitable and whatever. That does not necessarily mean that it is sustainable. A lot of the same things were said about all of the very exotic financial products which basically have - all of which had a similar level of opacity and insusceptibility of co-ordination and even to observation by third parties, so no-one really knew how bad it was until we discovered that the industry is globally destroyed as a result. So it seems to me just a real open question as to whether or not the industry as it is currently structured can accommodate this wish.

JASON SCHILLER: Verizon Business. I wanted to echo one of Tom's points which is - isn't it better for the industry to self regulate, particularly given that we have a community-driven approach, rather than leave the door open for someone else to come in and regulate on our behalf? Possibly in a top-down approach and possibly not having the technical expertise to understand the ramifications of the decisions they're making. That's my first point.

The second point I wanted to make is that I wanted to respond to something Geoff Huston said about when you go to the store to buy things, do people stand at the entrance and say, 'You don't have a need, you can't buy'. And I think, yes, that does happen when resources are constrained. And I'm thinking about in the United States in the 1970s when there was a gas crisis. In addition to the price of gasoline or petrol going up, we also had limitations on how much you could buy. That's pretty much what we're going to look at as we face IPv4 depletion. It is going to be a scarce resource.

The third point I wanted to make was to respond to Geoff Huston saying people aren't going to do transfers until after depletion occurs. And I think empirical evidence shows that transfers are happening on the black market and there is a reason for people who need space to go outside of the RIR system. I could speculate that maybe that is because filling out the paper is too difficult. They don't have the current space in order to show that they're using the space, but they clearly need more. There's any number of reasons, so I think it is unlikely to conclude that people will not do transfers until depletion occurs.

JAMES SPENCELEY: If I could borrow Geoff's analogy again, we have today a workshop that does regulate and you can go and buy what you do need. What we're talking about today is opening up a second workshop where you can go and buy as much bread as you can afford. And at the end of the day, the person with the most bread is going to live the longest. That's as simple as analogy as I can make it. This is a dangerous, dangerous option we're considering here.

RANDY BUSH: OK. I just want a general feeling. How many people - I'm going to ask, how many people feel there should be some form of justification and how many people feel there should be no form of justification, and then we can say - sorry Owen, we have a Jabber.

OWEN DELONG: From Jabber, Nesh says he agrees with Jason.

RANDY BUSH: And then if there's feeling for, feeling for justification, then we can try to figure out how. Does that seem reasonable? OK. How many people believe that there should be no requirement for justification to receive address space in the transfer? How many people feel that there should be justification to receive space in the transfer? Do you have a Jabber?

OWEN DELONG: Well, I have one more Jabber for no justification and two for justification.

RANDY BUSH: OK. Actually, 1/3 I'm seeing on Jabber.


RANDY BUSH: 3/4 justification, one against. So more people thought there should be justification than there shouldn't. And not a lot - it seemed a lot of people didn't care. No shock.

TOMOYA YOSHIDA: I have one suggestion that this suggestion is on the timeline.

RANDY BUSH: That's the Jabber thing that the Jabber thing could be just before and after justification and they didn't suggest justification just after. I'd like to have lunch, but I'd like to make it through this one at least before lunch.

How many people - I'm going to ask the - how many can live with question. How many people could live with - could not live... that's a double negative. If the decision was no justification, how many people would be really unhappy? If the decision was justification required, how many people would be really unhappy?

OWEN DELONG: The comment from Andy is that we can keep referring the decision until the address space runs out and then the point will be moot.

RANDY BUSH: There were enough of our friends who will be really unhappy in either case. Does anyone see a way out of this dilemma other than what Andy says which is that we can wait forever. James?

JAMES SPENCELEY: The option of justification up to exhaustion and then no justification after that. All bets are off.

RANDY BUSH: Um, OK, then let me ask this. Then let me ask... how many people think... I'm going to ask, justified before and after the end of the free pool. Justification only until the end of the free pool. Where the end of the free pool means that we're on our last /8. And justification is never required. OK.

How many people, if the decision was - the requirements to be justified before and after the end of the free pool, how many people are for that position? You have to justify it, period? How many people advocate justification until the free pool runs out and then no justification? And how many people believe justification should never be required? OK. Then let me ask, how many people would be exceedingly unhappy? For the first choice, how many people would be exceedingly unhappy if it was justified? And we know the answer to that before asked it. OK.

How many people would be exceedingly unhappy if it was required only until the free pool ran out? Oh, we may have a compromise!

How many people would be exceedingly unhappy if justification was never required? OK. I believe... let's try that second point again to be sure. Justification... I did not see anybody strongly unhappy if justification was required from now until the end of the last /8, until we're on the last /8 from prop-055. And then not afterwards. No justification afterwards. So, would anybody be very unhappy if you were required to justify now and when we're on that last prop-055 /8, no justifications are no longer required? It's time for lunch.

But I declare consensus, by the way.

Number two, justification is required until the free pool runs out and then not after. This is not easy stuff.

DAVID WOODGATE: Thank you for that.

RANDY BUSH: Oh, Suva, we appreciated you in the remote participation. Apologies for the technical difficulties and thank you very much for participating. I think given that I believe we still have a long way to go downhill economically, I think we will see more of this in the future.

SRINIVAS CHENDI: And APNIC thanks the host for hosting the session in Fiji, thank you.

RANDY BUSH: We'll come back for lunch. Is one hour enough? Let's come back at 2:00 as scheduled.

By the way, I would like to thank those people who are willing to compromise and say - it is hard when people raise and say I'm going to compromise in this. Thank you.

(End of session)

Thursday 1400-1600

RANDY BUSH: Could those people who aren't here, please raise their hands. What we are trying to do is get up on the screen, what we are doing is in the next part of the session is trying to come up with the wording to consolidate the proposal representing consensus with the process we have gone through. So at the end of the day, we can say, "This is the proposal as it stands." And try to get consensus, yes, to the membership meeting tomorrow. They are trying to put it in Microsoft Word form. Wish them luck.

SRINIVAS CHENDI: Welcome back to the policy SIG, while Sam is fixing that I would like to welcome our participants from Colombo, you can see them on the screen. They just joined us and will be through us all through this afternoon. Suva got disconnected in the lunch time because of the time difference, it is now 6pm there. They were in the morning session. But Colombo is hosted by Dialog Telecom and we very much appreciate it. We have our APNIC staff, Champika over there. He is helping the participants in the policy process. We would also like to thank CNNIC for sponsoring the policy SIG. I forgot to do that in the morning. I apologise but we very much appreciate your sponsorship. As you all know, they are the next host for APNIC meeting in Beijing. Please give them a round of applause.


Thank you.

RANDY BUSH: OK, welcome back. This is, as I said, what we are trying to do is put together a real-time proposal with the help of Sam and Jason and a whole bunch of folk and Geoff. So this is not the exact wording, we will have the exact wording at the end of the day. But essentially, this is what we have come up with. The minimum transfer size accepted be a /24. Use of the space, it says until the final /8 is being allocated that the recipient of transfer will justify use of the space using the allocation and assignment procedures in force at the time. After that, no justification is needed. The next item we had before us, is 4.3. Seller may not... boom! So should somebody who just sold that allocated space be able to get more from APNIC? The idea behind this was that somebody who is making money by selling space will, we shouldn't give them more from the public group. And I think we already have people who want to speak to the point.

DAVID WOODGATE: David Woodgate, Telstra. Just special terminology, I think 'seller' may be the wrong word to be used in this context, it should be 'transfer' or something. Something like that. Again, going back to the fact there is no transfer policy has no concept of, saleable spaces associated with it.

JAMES SPENCELEY: I think if you sell your resources you shouldn't be entitled to request more.


JAMES SPENCELEY: It is Dr No, James Spenceley, Vocus.

RANDY BUSH: Clarify?

JAMES SPENCELEY: They should not be able to request more space.


JAMES SPENCELEY: Just make it two years.

RANDY BUSH: No, I am just, did somebody want to discuss? The two dimensions to this would seem to be whether they can get more space from APNIC and if they may not, is there a time period for that?

DAVID WOODGATE: I agree that there shouldn't be an open-ended ability to come back and request further address spaces. Some of these just managed to transfer, I think that it is definitely a case that we are assuming that somebody has sufficient address capacity to be able to transfer addresses, they have sufficient planning to avoid emergencies like giving away addresses and then thinking about, "I need some more." So I support the concept that a two-year break or some sort of break, at least. Or several months, two years. It should go along with the concept of any transfer.

RANDY BUSH: Philip was next. We should have a tall mic and a short one.

PHILIP SMITH. Philip Smith from Cisco, as one of the coauthors of this proposal, it is something I feel reasonably strongly about there should be restrictions as on the screen. Because otherwise, what is stopping somebody going back to APNIC every day and getting another piece of space and selling it onwards? So I really would like to see point A put in place, in other words, some kind of restriction coming back to APNIC, maybe two years time, because in two years time, will APNIC have much v4 address space in any case?

JONNY MARTIN: Jonny Martin, small ISP in New Zealand. Yes, there should be restrictions but in keeping with the stance earlier, enabling transfers sticking to the bear minimum registry function for APNIC, I think the restriction should be taking care of in the vetting process. If somebody applies for space and it is justified on the basis of using it to number infrastructure, for a particular job or a particular growth of customers, if they come back again, using the same justification for more space, then obviously, they shouldn't get any.

RANDY BUSH: Jonny, I would note, the decision if I remember correctly and I am sure I will be corrected, just before lunch, was that there would be evaluation by APNIC host matters until we were on the last /8. So therefore, I always see the failure of point, somebody could do a transfer two month before then and you are on the last /8 and there is no evaluation. So therefore they could get

RANDY BUSH: I do have a question that I was going to insert anyway. Would somebody care to speak against this? In other words, we're hearing a lot of fors. If nobody comes up and speaks against, I'm going to try to get consensus real quickly and move along. So, is anybody against?

DAVID WOODGATE: Wasn't Andy Linton's comment really an against - against people who might make a mistake in the transfer?

RANDY BUSH: Andy was mildly against and said that the host masters would catch it in the evaluation. Does that represent what he said?


DAVID WOODGATE: And I suppose if we think of an against, I could foresee a situation where perhaps one company, there was some sort of change with respect to the structure, a company or two companies and part of one company went to another and needed to take that across, but that's probably already there and keep that address space because their network is already numbered that way, but that's probably already covered by the existing policy anyway, isn't it? So yeah, that won't be part of what we've been talking about here.

RANDY BUSH: The real world often comes up with ways to work around engineering searches. They'll form shell companies. Owen, with Jabber?

OWEN DELONG: From Jabber. Lina says, "So there's discussion about people buying and selling IP space. It may be happening. With businesses going down with the economic downturn, it may be likely such purchases are made without anyone being aware of it."

JONNY MARTIN: Sorry, just to rephrase my previous point. I'm against restrictions being written into the transfer policy. I would rather see them taking care of the existing host master vetting process.

RANDY BUSH: Let me see if I understand correctly. You believe the need for justification will handle that problem of somebody going left-hand and right-hand continually?


RANDY BUSH: Then let me ask - nerd mind again - what is to prevent me from selling something and being able to justify receiving more, getting more, selling it? I still have the need.

JONNY MARTIN: What's to stop you from obtaining or selling on the black market and... I see the problem exists either way we go and the simpler approach is to just not mention that in the transfer.


OWEN DELONG: Ron says, 'Lost video'. Terry says, 'Lost transcripts and video.'

RANDY BUSH: Is engineering on the case?

OWEN DELONG: Somebody lost audio too.

RANDY BUSH: OK. I am anticipating that the mass of people will not vote in one direction, so I probably am going to do again, asking generally how everybody feels and then see if anybody really objects one way or another. So how many people prefer the situation where they're there that a seller may not come back to APNIC for more address space from the free pool for two years. How many people prefer that solution?

How many people prefer no restriction on coming back to APNIC?

OK. How many people can live with - sorry, how many people would be extremely unhappy if there were - if the two-year restriction was there?

How many people would be extremely unhappy if there was a restriction?

I believe there is consensus on there being the two-year restriction. Is anyone going to throw rocks at me?

Should APNIC publically provide a log of all transfers? This is not keeping who is correct, this is not keeping who was - as the joke goes. This is a publically available log of every transfer. A transaction log. That is my understanding.

OK, Geoff, you were first.

GEOFF HUSTON: Lucky me, Geoff Huston. Wanted to make a comment here, normally I would not even put that in a policy proposal because it seems to be a conventional component of what a registry normally does, but oddly enough, that is not what we conventionally do. We in effect publish a snapshot of what is and as things happen, we keep on updating the "what is" but we never actually publish the changes between what was yesterday and what is today. And in the particular case of transfers, it seemed to me that there would be a social good for the participants in a transfer. So at least to have visibility on what and who had made the transfers. So that there is some suspicion by some parties that other parties are doing something in transfers and everything that APNIC did is accessible in a public log.

As I said, normally it would be there anyway, but APNIC doesn't conventionally publish a log, so I thought it was a minor technicality, but in this case, a positive technicality that should be explicitly noted in the policy saying that the membership would like this to happen. Thank you.


IZUMI OKUTANI: Izumi. I think this is very important for ISPs. A little bit different from the perspective of what Geoff has explained, but for an ISP operator to be able to judge what would be the risks involved or what kind of workload that they would need to make that particular address space that they purchased or obtained to make it useful. So there's a strong request in the community as a condition of accepting the proposal that APNIC provides the pass holders. Not just the previous one, but then the record of all the holders who have held address space in the past.

DAVID WOODGATE: I agree basically with what Izumi has just said and Geoff as well. They want to go further and say that the actual way to ensure that there is traceability back to the allocation from IANA logged back in the public log. As a chain of liability that some of the transferred addresses can understand the authority with which, that the authority has been dually delegated so that the transfers can be reasonably perceived.

JONNY MARTIN: No, the public log is not necessary. We stick to the registry function, update the registry with who has them now. There are going to be people who track that and create logs anyway. But don't do that within APNIC.

DAVID WOODGATE: I'm responding to Jonny. I think that is very much part of the registry function to trap such things, just ask anybody from the Alanti office.

PAUL WILSON: I'll just mention that we do quite regularly receive enquiries as APNIC that ask us to provide historical records of previous, Whois data. We do comply with the requests and I would generally regard it as going without saying that that is critical registry data that we're expecting to have a track of. I think that the community would be extremely surprised if we were to say that the current Whois records are the only ones we have and we have no idea what happened in the past.

OWEN DELONG: So, Terry M says he's fine for the log, provided it is limited to the necessary information only.

IZUMI OKUTANI: I have a question to Jonny. Well, as I said earlier, the ISPs, what is really necessary and it may be that we don't need it at the moment because you know the transcripts are not happening at this stage, but certainly in the future that we would be needing once this is happening and what is the strong concern that we have about APNIC providing this kind of log.

JONNY MARTIN: Maybe we're getting into some academic arguments here, but a log of Whois data, sure thing. But a specific transfer log, I'm not so keen on. I guess what I'm trying to avoid here is the policy that we're creating at the moment is setting APNIC up as the market regulator. We're putting all these clauses in as we go through each one. The more restrictions, and that sort of thing that we place on it, that's pretty much the outcome. I don't think we want that. Well, I don't want that.

JAMES SPENCELEY: James Spenceley from Vocus. I'm a transit provider and I get requests to route all the time and relatively often, annoyingly often, I get requests to deroute. I would find it useful to have a log of transfers to refer to. It would certainly assist in these disputes.

DAVID WOODGATE: Again responding to Jonny. I strongly don't feel that putting in a log encourages APNIC to take a regulatory stance and I think it is fully consistent and really, it is part of the duties of any registrar in fact.

JONNY MARTIN: To start with all of these little things, it's happened already today. What started out as a reasonably easy line of questions, what were we at, number four?

IZUMI OKUTANI: My point is pretty much the same as David's. I think keeping the record of the log is - you know within the role of the registry and I don't think that you would put APNIC into the position of market facilitator or involve APNIC in any kind of market exchanges of address space.

RANDY BUSH: Does anybody new wish to speak? How many people are in favour of a transaction log, publically available transaction log of all transfers? How many people are against it? Thank you. That was pretty clear.

I'm going to skip this one for the moment because... they're not specifically called out here.

OK, let's use this one. We're talking about timeline for implementation. We're assuming that we have a policy that allows transfer. There are many legal jurisdictions within our region. Five of them are respected by NIRs. OK, they have to go home and deal with governments and those of us who are individual members have our governments in strange places and regulatory regimes, etc. So, as far as implementation goes, discussion over the last week has suggested that NIRs... oh, one more thing. The current way we do things in the APNIC region is that APNIC policy automatically by default applies to NIR and NIR members. Therefore, decisions we make here on transfer by default immediately apply to the NIR members.

There are NIRs concerned that they have to go home and explain this to complicated Government and regulatory environments, work out procedures and processes and essentially build consensus within their environment at home. The solution that has been discussed this week and is not directly on this, but was alluded to in Geoff's proposal, prop-050 is that even though by default, APNIC policies apply to NIRs, this one does not, until that NIR chooses to opt in. So that would allow the NIR to go home, work things out internally and say - now we are ready. Or never say that they are ready. OK, so let's presume this foil, point B, point 1 says that. And I would like to try to gain consensus over that as a specific item. And I would specifically like to place the NIR to which I am a member in the - excuse the American idiom "hot seat" - and Maemura-San, does that address your concerns of the NIR management?

AKINORI MAEMURA: Akinori Maemura from JPNIC. In the assumption transfer mechanism to be enforced, I would prefer to have point B, point number one as accepted.

But, I would prefer to have it... APNIC is just allow the transfer from 2 in the NIR neighbours, but it doesn't mean that the NIR is forced to have that transfer mechanism, because the NIR has its own contract, policies which is to be applied to the NIR member. Then, once the NIR enforces the transfer mechanism into their rule, then the transfer between the NIR members to the APNIC member would be happy. That's my understanding.

RANDY BUSH: Thank you. Geoff? I was going to say before Owen sneaks up behind you.

OWEN DELONG: Terry M says he is fine with B1 provided the NIR members have a choice in passing non-playing NIR.

RANDY BUSH: They do by definition.

GEOFF HUSTON: They do by definition. I was actually going to clarify the intent here, because normally the NIRs are bound by their agreement with APNIC then they form that they would abide by all APNIC policies. So unless we explicitly say otherwise about the policy, unless we do something, then this policy will apply to NIRs, because when they became an NIR, they said - we will agree to all of the prevailing APNIC policies. So an APNIC policy happens, the NIRs are bound to adopt it. But I thought this one had particular implications that made it special, and transfers do have some issues on just the registry function. And NIRs operate within the legal regime, a national regime, which are in this case slightly different. And I think because of that, they actually have more work to do in understanding the implication within their legislative frame work, the regulatory framework that doesn't necessarily apply to APNIC as a whole, as an international body.

So in this case, I thought it was worthwhile allowing NIRs the ability to - within their own communities and frameworks - understand what they would like to do as an NIR community and adopt what they believe is appropriate. So this was actually trying to make sure that all parts of APNIC - including the NIR members - have maximum flexibility here to do what is necessary within their own environment.

SEIICHI KAWAMURA: I have a comment about this NIR issues. Generally I agree with the concept of transfer of policy, and secondly, also we understand you know and eventually the transfer between NIR members and APNIC members and other NIR members which is not available. So we need a policy for the situations, but this time, I will with hold my answers on this item because, frankly, I didn't have time. Because since the new issue at this time - and frankly I do not have enough time to discuss with the members - and like the form of proposal, which whether applied with the NIR - and we feel that this policy will affect NIR policy too. So I said before the transfer between the NIR members and other members will happen, so I would also like to develop our own NIR policy which is consistent with APNIC policy. So in order to do that, I needed time to consult with my members. So anyway, if the policy is passed, I would suggest that we put one condition there to allow the members to develop the policy within a reasonable amount of time.

KO WEI WU: As we know, the NIRs have their own member and if their member is doing the transfer under the NIR, we in APNIC and policy point of view, are treated as a member. But if the member of the NIR tried to do the change to a non-NIR member, that means that it is current to the APNIC member, then it might be better if the policy is there to design the official policy, because this is kind of strange. So I would prefer to have a very special policy applied in this kind of a transfer between the NIR member and non-NIR member.

JONNY MARTIN: I don't think it makes sense if we are having a transfer policy to not have it work across the entire IR eco-system. You're only going to help out the black market if you're going to try to stop it between subsets of NIRs or between RIRs. So if you're going to take - if you're going to accept transfers, then accept that they need to happen across the entire RIR system.

RANDY BUSH: Jonny, I don't believe the intent here, and someone will correct me, is for the NIRs forever. They're just trying to get the ability to be asynchronous, because when they go home, they have different levels of requirements to pass through in their local environments.

JONNY MARTIN: Yes, if it is a timing issue, then no opposition. But you know, it's still an opportunity. That timing could be abused. It's going to take two years to implement something and that's as good as having an opt-out clause of the transfer model.

RANDY BUSH: OK, let me see if I can phrase this: address transfers should be permitted between APNIC account holders and NIR members if and when the individual NIR opts in to the transfer policy. Sam, you got that? I'm rephrasing the question. Address transfers should be permitted between APNIC account holders and NIR members if and when the NIR implements the transfer policy.

How many people support that?

How many people do not support it?

Thank you, so consensus is for.

The next one I've got is between APNIC account holders and account holders of the other NIRs. RIRs, my apologies. On the condition that the seller meets the conditions of the selling registry and the buyer meets the conditions of their resident registry. In other words, somebody from RIPE can sell to somebody in APNIC, but the APNIC person must, according to a decision being made earlier today, justify the utilisation if the last /8 is there. So it is to try to allow inter-region transfers within an environment where the policies are not the same in every region, because the regions are not trying to get a single global policy on this one.

DAVID WOODGATE: I support what you just said, I just object to the use of the term "sellers and buyers" so long as that's not going to be a part of the wording.

RANDY BUSH: I will continue to use 'seller' and 'buyer', because for non-English speakers, 'transferee' and 'transferor' is a little difficult. Seller and buyer are clear.

DAVID WOODGATE: I appreciate that, but I would prefer that not to be the representation in the final wording of the proposal.

RANDY BUSH: That will not be the final wording. God knows what the final wording will be, but it won't be mine.

IZUMI OKUTANI: Regarding the inter-RIR transfer, I did hear some voices that said - OK, we shouldn't just allow everything at once, so take steps. First, allow it within the APNIC region only and then afterwards, as a second step maybe we can also start considering the inter-RIR transfer. However, even those who say that, they do agree that at some point in time, that it's good to have inter-RIR transfer, so not just to restrict the transfers within the region, but then able to transfer in any areas - with other RIRs, that also implements the transfer policy.

MASATO YAMANISKI: Actually, some regions have already had address space. Unfortunately, in some other regions they don't have the address space. Also the ratio is not the same ratio as the progression. I think it is a problem, but it is clear and present condition of us, so if we will limit on the... sorry, if we will not allow inter-region transition, I think we cannot resolve this issue. That is my concern, so I would prefer to allow inter-region transition.


JASON SCHILLER: Earlier this week, there were some discussions with regard to transfers between RIRs, and discussions with buyers and sellers which we have addressed and other people have addressed restrictions. Is that something that should be addressing? Are we overlooking it?

RANDY BUSH: I'm not understanding.

JASON SCHILLER: So, in other forms of the transfer policy, some people put restrictions, for example can't be more than a /24. So that's a restriction.

RANDY BUSH: There have been interesting restrictions placed.

JASON SCHILLER: Does that apply...

RANDY BUSH: Though these are the restrictions we're talking about. In other words, if you in ARIN want to sell a /25 to me and /25s are allowed in ARIN, I cannot buy it because I'm in APNIC. And we do not allow /25s.

JASON SCHILLER: OK, I understand what you said and I think that's addressed. He didn't believe it was addressed based on how the conversation is going, but it is clear to me that it is. Thank you.

RANDY BUSH: That's the intent. The seller must comply with the selling rules of their region. The buyer must comply with the receiving rules of their region.

GEOFF HUSTON: Geoff Huston, for those of you who either didn't quite do your homework before you came here or would like to read the gory details. The wording that might be appropriate is in prop-068 that describes in some detail how you would - how the disposer would meet the local requirements. The requirer would meet the local requirements and the block to be transferred would have to meet both sets of requirements. So in other words, no RIR's policy would be diluted or weakened. That was all that was the intent.

RANDY BUSH: There are at least two proposals that describe this.

GEOFF HUSTON: Thank you.

JONNY MARTIN: This is a question more to Geoff. So, you're after free and open transfers, that's correct? That's the whole idea of the transfer policies is to leave it up to the free market? People can dispose of and obtain IP addresses as they need them? Sorry, as they want them. So why are we putting all these constraints on it?

GEOFF HUSTON: I was asked if this was a constraint, and I believe that allowing the folk from other regions to interact with folk in APNIC is not a constraint. It is saying - yes, fine. And from APNIC's perspective, all it says is that the APNIC side of the transaction works within the APNIC policy framework. I'm not sure that you could get fewer constraints than that, Jonny.

JONNY MARTIN: Weren't we talking about both parties meeting the requirement of the other?

GEOFF HUSTON: Oh, sorry, maybe again my English and what I say at the microphone doesn't precisely match the exact words in prop-068. And in which case, what I'm saying is wrong then prop-068 you should read and hopefully you're an English language reader and can understand it well. Thank you.

OWEN DELONG: So from Jabber, they're still not getting the transcript and I'm doing my best to summarise for them in the meantime. But Andy Linton says, "If you allow transfers to users based on what was discussed this morning, that is you document this morning and then transfers between RIRs are likely to happen just as soon as APNIC transfers, you can't dictate who will want to transfer what to whom".

RANDY BUSH: I keep seeing motion but none of it is directly towards the microphone. OK. So the wording - oops, I have something in Jabber.

So how many people support the policy that the, in the transfer, the person sell - yes, David, your objection is noted. Must meet the selling requirements of their region and the person buying must meet the receiving requirements of their region? How many people support that proposal?

How many people are against it? OK, there aren't very many people who feel about it, but it was consensus for that restriction. You look concerned, Mr Wilson?

PAUL WILSON: I just thought of a question that might need rephrasing.

RANDY BUSH: Do you have a suggestion for rephrasing?


RANDY BUSH: OK. What other delightful questions are we left with? Sam?


RANDY BUSH: Ah, yes. We can now address the timeline given the NIR out of the way. Let me see if I can get the wording.

OK, there was concern that the transfer policy should not take effect immediately. Where immediately is defined as when APNIC has figured out the technical procedural regions and any other process needed to do it. We have agreed on a point which allows the NIRs to set their own timeline, so we're now down to the timeline only for APNIC members and APNIC. And the question seemed to be - between allowing it as soon as APNIC was prepared to process transfers, or postponing until we're at the end of the free pool. So do those concerned agree that that represents the question properly?

GEOFF HUSTON: Yes, says Geoff!

One of the attributes of this process is that sometimes good ideas come up unexpectedly. And between the time that all these questions were drafted, and now, there was a good idea that we had and the good idea was that we would change the criteria of assessment when the free pool ran out. And it seems to me that that overtook this question. So I suppose I'm now sitting here thinking - for those who were worried about this as an issue, and wanted to defer it, is the already agreed point of using demonstrated need until the free pool runs out sufficient and adequate?

In other words, is what we've done already enough to answer this and we can simply adopt it when we're capable now, because we have the rest of the procedure in place? As I said, good ideas seem to come up sometimes without planning for them!

IZUMI OKUTANI: I think, well, regarding the point that Geoff has made, I think the people who actually expressed concerns, and requested to wait for the implementation weren't just concerned about the part about devaluation, but I think they were concerned about other things such as implications on trade market and things like that. But having said that, what I wanted to say was that quite a number of people in Japan supported the idea of implementing this proposal before the IANA or APNIC pool running out, and actually they feel that there will be less confusion by doing this, because people will be more prepared to and used to the process of transfer rather than facing it at the last minute once the pool runs out, and then there will be less panic that oh, I have to get hold of the address space.

So there was support expressed for Option A.

RANDY BUSH: So, what you're saying is that this would be better than this?


RANDY BUSH: Good luck with that, scribes! Sign wave versus square wave, if you want to get into it. Jason and Philip, fight it out.

PHILIP SMITH: Philip Smith. Jason is saying I should go first. Haven't we already talked about this at length this morning?

RANDY BUSH: No, in a different way.

PHILIP SMITH: In a different way, but I think we agreed to implement the policy and there was restriction on the recipient until we have only the last /8 left.

RANDY BUSH: As Geoff said, does that point overrule this one now? Geoff asked that question. Can I ask to go a different way? Would anybody now care to speak for delaying implementation? Jason, did you wish to speak for delaying implementation? Well then, I don't think that we have much discussion. But you have something to say.


RANDY BUSH: Go for it.

JASON SCHILLER: I just wanted to capture a point that Izumi-San brought up which was with regard to timeline that we might want to treat the inter-RIR transfers on a different timeline and that when you pose the questions we were going to talk about, that was not one of the questions listed and I was just wondering if you intend to deal with that question later or... with this?

RANDY BUSH: OK, let me ask another question. So, there will be the time that APNIC region is willing to conduct transfers. There will be the time that LACNIC is willing to conduct transfers. Can you give a reason for delaying transfers between LACNIC and APNIC members beyond that time?

IZUMI OKUTANI: Let me try to answer it. I suppose the people who felt that we should take steps in allowing inter-RIR transfer would be - maybe there are more things that you should work out in the procedural manners than just allowing transfers within the region. But if these things are clear, then maybe there are no strong reasons to wait, but then I think people feel that maybe they are more likely to have more issues with inter-RIR transfers than just transfers within the region.

RANDY BUSH: Owen with a Jabber.

OWEN DELONG: I do indeed. Thutoy - would somebody correct her name if I am wrong - says, "You would like to add a point to this discussion. We've just got consensus that sources entity should not contain more address space from APNIC in two years, but I think that is not enough. To prevent the ability of selling again and again v4 space, I propose to add some routes to have the source entity to be the recipient for others for two years after the transfer."

RANDY BUSH: OK, that is a new and different point we can take up after this one.

MASATO YAMANISHI: I agree that there are more issues between inter-region traditions. However, I think this experience, I mean the tradition in the inter-region can be applied to transition for the APNIC account and the NIR account because the transition is a different payment registry. So I support the transition in the inter-region and also we must start as soon as possible because it will be helpful for the implications in the NIR I believe.

SEIICHI KAWAMURA: I understand the point between APNIC and RIR and APNIC and ARIN. RIR... are, I hope I'm saying them correctly and APNIC and ARIN are totally different, especially the reverse delegation of the DNS. So I think there's a little bit more to be concerned about.

MASATO YAMANISHI: OK, at this stage, I cannot list a full list of issues, but for example, of course APNIC and each NIR are using different things. I think that it is one of... let me say it is one of difference of inter-registry, internal registries. Maybe there are many differences.

GEOFF HUSTON: Could I help out here. If we talk about the technicalities of the interaction, then there is a suspicion that it is slightly easier between NIRs and APNIC and slightly harder with RIRs. But as a policy group, I'm not sure that the slight issues should be a matter of concern about policy. If that's the policy, it can be made to happen. So I wouldn't worry too much about it. It's more the issue about - are you really concerned about the principle of the inter-RIR and that the APNIC side of things is consistent with the framework and from the APNIC side, there's no additional constraints and no additional constraints. It's just the transfer policy. Is there any reason not to just do it is really the question here and what we're trying to do as a group about policy. So technically yes, slightly different, doesn't matter.

SEIICHI KAWAMURA: If APNIC and JPNIC will tell me that I can feel comfortable with this inter-region transfer, OK, I can agree on that.

RANDY BUSH: I'm looking at you, Paul. There's consultation going on. I have one personal comment on Geoff's previous which is that I do think that we take engineering constraints into consideration. Just, we also believe that the implementation departments of the various registries are competent, but we do try to take care of our own routers.

PAUL WILSON: I'm sorry for that pause. But look, we have implemented numerous transfers of registration records between RIRs, between APNIC and NIRs in the past and it's certainly something that we can do in a routine and reliable manner. The thing that's uncertain about this entire scenario I think is the number of - is the amount of traffic, the number of transfers that we might end up seeing, but I think the likelihood of a very rapid - if we take RIPE NCC's example so far, there clearly is not an enormous pent-up demand here and the likelihood of an extremely rapid growth in the activity seems small. There is also issues I suggest of cost recovery through transfer fees which is out of scope with the policy discussion. That may need to be considered by the APNIC EC if it turns out that there are requirements that are more substantial than we can cope with in terms of the normal migrations.

RANDY BUSH: To clarify for those who don't know, the EC sets fees. The Policy SIG does not. Owen on Jabber.

OWEN DELONG: With Jabber, "With address space getting squares and the uptake is different for different regions. Won't facilitating space between RIRs mean that the RIR with address space can offer it to the RIR that is running out faster? Certain regions are growing faster than others, if the presentation I saw yesterday is right. Are the RIRs worried that they are losing resources that they can't get back?"

RANDY BUSH: OK. The question I'm going to ask is - how many people support inter-RIR transfer? Transfer between a member of APNIC and a member of some other region on the condition that both sides are following the rules of their region. How many support that proposition?

David, are you about to say that my question wasn't clear? I'm asking how many people support inter-region transfer under the condition that the rules of both regions are followed?

OK, how many people are against it?

Got you, got you. And again, the consensus was for that proposal. I believe we have reached the end of transfer.

How I tend to proceed is that I think we're going to lose coffee break or tea break. If you want tea just go out and go out. Because oops, David.

DAVID WOODGATE: Sorry, Randy. I'm not sure whether it got lost in my e-mail or something, but I thought we were discussing timing questions there and had already discussed the inter-RIR region question?

RANDY BUSH: So I've gotten lost?

DAVID WOODGATE: I wouldn't have gone so far, but I may want to reconsider something.

RANDY BUSH: OK, let's go back and clean up timeline. Right, thank you David!

NURANI NIMPUNO: Just to clarify, I think I can see that this is at times a confusing discussion, we did ask for consensus twice. Luckily we reached the same conclusion, twice, so maybe we can leave those questions where we've actually reached consensus and move on and then I think that the timeline - the last thing we discussed most people said that all morning, we discussed - assuming that this would be implemented immediately and therefore this point is moot. But we're a little bit confused here.

IZUMI OKUTANI: I have a slightly different view. We did agree that we wouldn't ask for evaluation as a condition of the transfer in the morning, but this is the timeline of implementation for the transfer itself, so I think it is slightly different and maybe it sounds like a little bit repetitive, but then I would like to confirm consensus, just in case.

RANDY BUSH: Does anybody... who supports the proposition that as soon as APNIC figures out how to do this, that transfer will be allowed? In other words, no delay past APNIC figuring it out. High, please. If I'm confused as to what it is, I'm going to get confused about hands.

And how many wish delay until we're on the last /8?

Thank you. I'll go slower, how many wish delay? Hands were kind of going up there. OK, no delay. Sam, what am I missing? Anything? Done! OK.

OK, that's transfer. And what's going to happen now is poor Sam is going to come up with some writing that is like this which says what we've... stuff that! Oh-oh!

Prop-063, reducing the timeframe of IPv4 allocations. Is somebody going to speak to this? Philip or Jonny?

JONNY MARTIN: OK, so this is the same policy proposal that we had at the last meeting with a minor enhancement. So I just want to reduce the allocation open time from 12 months to six months. The main reason for this is as we're running out of v4, it means any LIR that applies for v4 space at the moment is going to receive a whole year's worth. Obviously as the pool diminishes, it is likely to cause unfair distribution of what does remain in the pool. Obviously, this hasn't been submitted to other RIRs, they're currently based v4 allocations based on the allocation window. So just to propose that we drop this to six months means that any LIR applying for space is only going to receive space for the six months. Once the six-month allocation has run out, the LIR will have to return to us for more.

Now, the enhancement to this or the change to the proposal from the last minute is that for LIRs applying for the minimum allocation, they continue to receive an allocation based on the upcoming 12 months. So that retains the current ratio of address space to time, otherwise we're effectively doubling the requirement for the minimum allocations.

So one advantage of this is that it ensures that there is even distribution of the remaining address pool, or more even, and ensures that everyone has a greater opportunity to participate in the remaining pool as it runs out.

Disadvantages, organisations applying for v4.

JONNY MARTIN: The last point, obviously, there is no impact to LIRs seeking the minimum allocation. A bit of additional data. The same numbers as last time. So the past few years, it shows the LIRs returning for more resources. We can see most LIRs are returning for more resources in kind of a 6- to 12-month window but equally in a 0- to 6-month window. So based on that, the impact on both organisations and APNIC is not as bad as we originally thought. So, questions? Comments? Other issues?

OWEN DELONG: Andy Linton says, 'I think it is unnecessarily complicated and it should be the same as other LIRs. If I was planning it, I would want more than six months.'

RANDY BUSH: We have a comment from Colombo? If you were trying to speak, raise a hand, otherwise I am going to slide past you. Izumi?

IZUMI OKUTANI: It is me, regarding the point mentioned from the chat, I have the same question. I think the strong concern was expressed at the last meeting was might create inconsistency with other regions so might create disadvantage for LIRs in the APNIC region they have to come back to this more frequently. So how do you think about this point?

JONNY MARTIN: Given how long we have got left of the v4 pool, is it such an issue. There is no point asking for an allocation window that can't be fulfilled. It will happen to the last few large allocations. All we are trying to do is make it a bit smoother for all the smaller LIRs basically as things are run out.


DAVID WOODGATE: Given that assessment of the requests, I noticed there was a comment from the Secretariat in the mailing list this morning, I believe the comment was along the lines that it was an estimated increase of 0.8 a year to be able to support that particular proposal. Can I just, can somebody from the Secretariat just confirm that is the case? I don't know whether that would be Sam or Geoff?

GUANGLIANG PAN: The host master, based on all the requests received in 2008, count the requests and then come up with a sure number. The host master spends most of the time preserving ongoing v4 requests but consider some of the requests has come earlier, like Jonny mentioned. Also the request size would be reduced. So the figure is about 30% of the host master workload will be increased. At this stage, APNIC have four host masters. Based about 7.8 host master workload will be added to the host master team if that policy is adapted.


IZUMI OKUTANI: While a lot of the ISPs in Japan also understand the intention and the concept of the proposal, people feel that it probably might not have it the effect that it tends to have, because I think that six months would be maybe too short to meet people's needs, so what people will start doing is that instead of actually honestly saying what they did, they might just put what their need is after the six-month need, so I think it is better to keep the policy as it is.

RANDY BUSH: In the interests of time, let me try the following. Is there anybody else who would like to speak for this proposal? In other words, I have a suspicion that there is not consensus for this proposal and we are running short of time so I'm going to be a little rude.

MASATO YAMANISHI: To reduce the number of address requests, so if there is, some new miracle and concrete study about the effects of the proposal, I think it is very helpful for better understanding. And I'm afraid that the proposal just doubles the number of allocations being used with the address space.

OWEN DELONG: From Jabber, Andy says, "Given that we have so little time why we introduce this and multiple people in the room responded essentially so that not one big ISP can come and get the available address space just before we run out".

PHILIP SMITH: Philip Smith, Cisco, as co-author with Jonny. I was one of the multiple people. The idea is not about trying to reduce the number of host master requests, the idea is to try to share the spaces as evenly as possible.

JONNY MARTIN: Fragment larger.

PHILIP SMITH: As you get less and less and less and you're trying to share it out, you subdivide, so in other words, short-term actual allocation. That was all.

JONNY MARTIN: And I guess my personal opinion on this one, this is not talking to the co-chairs, but if we don't get consensus now, it is probably not worth trying to get this policy through again. The time frame is really going to be too late.

RANDY BUSH: Anybody else like to speak for this proposal? OK, how many people are for this proposal to reduce the request timeframe from 12 months to six? And how many people are against it. Oops, hold on a second. Again how many people are against it. Colombo, I'll get you in just a second. Raise your hands a little higher please. Colombo, you're just wanting to make sure that you have two hands against, and what do you have for? Colombo? You're either frozen or... oh, OK. I don't think that we have consensus. Thank you, Jonny.

Prop-069 is next, question mark? I'll do it next. Oh, one is going to be fun.

AXEL PAWLIK: OK, I'm the managing director of RIPE NCC and I happen to be one of the authors of this global policy proposal. Just to run through this with the whole list of others. This came up basically is that we are usually not involved actively in the policy development process, however, we do sit together occasionally and make plans to think and have ideas and in the process of doing this, we thought that this is actually a policy that we could use later on and I'll just run through that. The problem statement is that the current global policy on IPv4 allocations from the IANA to the RIRs does not cover reclaimed address space and the RIRs among them have no mechanism to transfer that into any space basically among themselves.

So in terms of introduction, this is a global policy proposal that will, if accepted, govern allocations from the IANA to the RIRs again. And particularly, it speaks about reclaiming the address space. Now, it is about the principle or the details on how on the day-to-day basis, work with IANA. It will be defined separately and we've done the same thing with the global policy that we found in the same place. Some definitions - speaking about recovered address space. What do we mean about that? Basically, this is reclaimed address space and we have some experience in reclaiming its return data space should that happen. And it does not include freshly reclaimed space from expired contracts. It is a little bit too...

If we speak about the IPv4 address holdings at the RIR, that is the allocated address space that is reserved for a later allocation.

We look at two phases in terms of implementation here. One is phase 1 we call it, that's the phase prior to depletion of the IANA IPv4 pool. Basically what happens here is that IANA prepares for this and sets up a recovered IPv4 pool. Then, as it happens, the RIRs reclaim or recover in the sense of the IPv4 address space and we would then return what we have recovered to IANA every three months in blocks of /24s or larger. Now, that's all pre-depletion. If at some point in time, IANA announce the depletion of the unallocated IPv4 address pool and, of course, after ratification of the policy, the global policy, the currently active IPv4 allocation, global policy, ASO-001-02 would be resigned and allocation by IANA would then process using this global policy.

Now, how would this work when this is actually active. We have again some definitions and we talk about the allocation period that's basically arbitrarily six months following the 1st of March and 1st of September every year. And then we speak about the IPv4 allocation unit that is flexible thing. It is defined as a tenth of the current IANA IPv4 address pool as a whole, rounded down to the next CIDR boundary. And allocations to the RIRs will then be done with one allocation per RIR, per allocation period, in that case the RIR to receive the space will receive it. We'll talk about that in a little while, and what that means. And the allocation is one allocation unit as defined a little bit earlier.

What does it mean? Under which circumstances would one RIR be eligible to receive the address space in the policy? As we said, there is just one such allocation per'allocation period, so the RIR must not have received an allocation in that allocation period, and the RIR holdings, the IPv4 holdings should be less than 50% of the current allocation unit of that period.

In the case, we would see the RIRs come up, one allocation unit to the RIR independent of any other requirements on RIR recognition, or if no space currently is available at IANA, then as soon as possible.

And then we have with the reporting, IANA will keep logs publicly of returned IPv4 address space, also of the allocated IPv4 address space and keep a public registry of all of the IPv4 address space and IANA make public announcements of transactions under the policy.

So other thing - timeline. The global policies take some time after they go through all the regions that will be given to the ASO, Address Council. The process will be verified that it actually happened properly and then the policy will be given to the ICANN board and ratified hopefully.

Advantages and disadvantages, this is basically a policy which defines mechanism for the redistribution of the reclaimed address space. Disadvantages - we don't see any, and in terms of changing the allocation relationships as it says in the policy between APNIC and its members or APNIC and the NIRs, that doesn't change anything.

Status in the other RIRs, this is the first time that we prevent the policy proposal and it has been submitted in, I think, the ARIN region and also recently in the RIPE region and we'll see what becomes of that. And basically there is a quick run through.

Now, I think this in general, with housekeeping being prepared, we heard quite a lot about this earlier in the transfer discussion. This also is a similar policy. We done know whether it will be used. It says the RIRs may reclaim address space, it doesn't say that they actually will do that, it doesn't say that the RIRs will receive address space back from the members. In case that happens relatively late in the process of the next year or so or two years, it might happen where we can reclaim lots of other space from the UK Ministry of Defence, I would be sitting possibly on a large block there while other regions have run out is that a good situation or not, I don't know, but in this case because there's policy in place, you would actually be able to redistribute the address space among the other RIRs, I think that's in general a good idea. But if you have questions, we can discuss.

DAVID WOODGATE: I support the proposal on principle. However, I do also note that I suspect that it will never actually be used much. And I can also see that it has the potential to disadvantage the most efficient region in terms of the recovery of the addresses. But in spite of all that, I actually support the proposal.

IZUMI OKUTANI: Yes, just like David, I think it is OK to set up the framework like this and clearly define how historical or - well, it doesn't have to be historical, but I assume that it would be. But to set up the framework and we have minor suggestions and questions about the implications of this policy.

One, I think it doesn't really set the limit on the size that IANA distributes to RIRs. And maybe, there's not much point you know for example keeping going and going with the smaller /27 or something and I think it would be good to set the limit on the minimum size of allocation from IANA to RIR, that's one.

And another thing, I think this is just restricted within our region. I would assume that for this last /8, the final /8 block, that would be an exception to the policy is our assumption. Is that right?

AXEL PAWLIK: Yes, it is regular address space which is given out to the RIRs. This is only for returned address space to the RIRs and then back down again.

IZUMI OKUTANI: OK. And another point was that - sorry for being into real details. I know that it is really small point, but the definition of how IANA defines the minimum allocation size to the RIR, it seems very complicated. Like one tenth of the spaces, so it might be good to put an example or something so that it is very clear that what would be the size that IANA would distribute to the RIR. But generally, I think it is a good idea to define this kind of framework.

BARBARA ROSEMAN: I would like to make it clear that this is not IANA proposal for this. This is part of the proposal as it was written by the people putting it forward.

RANDY BUSH: Barbara, you, as IANA, do you care to comment on your feelings about the implementability at layer 7 through 9 of the proposal?

BARBARA 'ROSEMAN: It's feasible we could do this. I think that there are actually some other complications that have to do with how we would maintain the database moving forward with the IPv4 registry and what kind of complexities we would get into with distributing one tenth of the available space moving forward. It is not the kind of proposal that IANA would have put forward in terms of - it is a little more complex than I think we would prefer for having to deal with this space.

However, I do think that returning space to IANA if you're interested in doing that, then having a mechanism like this is probably appropriate.

OWEN DELONG: Two comments from Jabber. Andy Linton, "I support this, it fits well. I also agree with David's view that it is unlikely to get a lot of use". Mike Jager from New Zealand says, "I also support the policy in theory but I also agree that it seems unlikely that this policy would ever be used. RIR members seem unlikely ever to hand back space. They will'. 'They'll sell it', is Andy's response and Mike says, "Bingo".

AXEL PAWLIK: That goes well with the idea of the transfer and the market in general, the people who say that yes, people will never give up the address space, they'll keep it. We done know how much liquidity there is in the market, possibly would be. So we do transit policies anyway to be prepared and I think this is basically about being prepared for the situation that might come up. I think it is good housekeeping and that's expecting from the RIR and their communities.

RANDY BUSH: OK, questions, answers? Opinions? I am not going to rephrase this policy. So I would simply ask, how many people here support it?

And how many people here anti-support it.

I believe there is consensus, thank

I've been told by - oh.

IZUMI OKUTANI: Can I ask a question? So are you going to define the minimum size for this or how do you address that point that I raised earlier?

AXEL PAWLIK: I will take that back to the author's group. We can certainly put in some examples to make it easier to understand and something like that.

IZUMI OKUTANI: At the minimum allocation size as well?

AXEL PAWLIK: We can, yeah.


JASON SCHILLER: I have the same concerns that Izumi does have about possibly giving small blocks to the RIRs. If IANA has a single /24 in the pool and you cut it up ten ways, that's a /28. If we made it at the smallest of /24, that would be better.

AXEL PAWLIK: At some point in time, you should move to IPv6!

RANDY BUSH: There were also three or four supports from the Jabber chatroom by the way.

Next, shall we try prop-070. Where's the pretty little thing. Mr Spenceley. Would you care to sing. Oops, I'm in trouble. I'm in big trouble. Senior management says we are going to take a break.

All in favour of a break, please raise your...! Is there anybody who strongly objects...!

(end of session)

Thursday 1600-1800

(Start of session)

Next we have prop-070 by Jonny and James and let's rock and roll.

JAMES SPENCELEY: OK, I'll start off. The photos on the top are...

RANDY BUSH: Pardon me. Those people on the mailing list, you can read the text Sam has accumulated what she believes we agreed on so far on transfer.

She has put it into one coherent policy statement and sent it to the list. You can read it now. What's going to happen is, before the close of today's sessions, we're going to put it all up and say, this is what we think we agreed on. Is there consensus on it?

JAMES SPENCELEY: So at the start, I was looking for an analogy for the photos of the empty dry pools, hence no analogy needed.

The aim of the policy is to ensure that more even distribution of the final IPv4 resources but no one entity can come along and take the remaining last /8 that has been allocated to APNIC. The other aim of it is fairness. That small members have address space and that a larger member or an NIR doesn't - I suppose request the last remaining space ahead of maybe 20 or 30 small members.

So, last people requesting space, at the moment you get six or 2007 months, well, now it is 2007 months with prop-032 not reaching consensus. It means that the last person requesting space gets 12 months of IP resources. Whereas the person coming the next day gets no resources. That doesn't same fair to us.

A large LIR with the justification can request, you know a huge portion of the last remaining space so that might be Government assistance or based on a large deployment. So the question is, how can we more evenly distribute the address space. Prop-063 had a similar intent in shortening the window down from six months to 12 months. Prop-070 - also attempts to address this problem. Potentially a simple maximum allocation resolves this as well. Either way, I think this is an issue that I think needs resolving and relatively quickly. So the details of the proposal is an additional maximum of /8. People have commented that maybe a /8 is too large, so upon enactment of the policy, that would become the maximum that a member could request.

And as the IANA free pool decreases, you've seen in the table, the maximum that the LIR can request also decreases, so as we get to the final five /8s, so that's before the absolute final five kick in, a /18 is the maximum. Those are just some starting points we came up with.

Disadvantages - if this is only implemented by APNIC, it disadvantaged the region and with the rest of the world continuing to allocate large blocks of address space. And based on the current allocation sizes, it creates significantly more work for APNIC.

So we have some considerations - potentially the option to increase or introduce the wait state, so that as you request address space, you have only one request per'month so that members can't come back and just ask for a /16 after a /16 almost daily. So that's been proposed off the list. We've also introduced the different approach, so we're proposing a number of different things here and just interested in industry feed back. And the second approach is the version of this and I'll hand it to Jonny to explain this one.

JONNY MARTIN: Given the current proposal we have here, we don't actually expect it to get past here, so talking to a number of you and gauging the mood of people, I can't see people agreeing. So we were having a bit of a think about other ways that we can perhaps more even by distribute the remaining space, so the simple approach is LIRs can only go back to APNIC once every three months or six months or whatever the time happens to be. And at any time, the queue of requests for allocations we service the QoS first, so there is a series of /10s and /2s, you service them first, no matter where they are in the queue at that time. So this is going to allocate or distribute the remaining addresses a lot more evenly, and presumably over time, the switch bottom allocations will reduce, because obviously it is going to be an advantage to LIRs to request less space, you know because they've got a higher chance of that actually being allocated sooner. So I think we've got a few examples here.

Now with the request for /10. At that point, we go through the queue of allocation, of applications that are currently there. And we basically service anything that's a minimum size allocation. At that point, we go over to the next slide allocation and if it was a /10 which is an extreme example, we then go and allocate a /10 worth of the existing allocations before we service this one. So what you're basically doing is that we've got the largest application and the queue is going to be waiting a lot longer because we have to service a whole bunch of others so that's kind of what we're looking at. I guess we're after some feed back on whether there's any interest in refining a model like this? Obviously time is running against us so it would really be something that we would have to get to for the Beijing meeting and look at consensus there.

RANDY BUSH: Discussion?

DAVID WOODGATE: David Woodgate Telstra. As has been said on the mailing list, I certainly believe we should be looking at these various approaches in trying to identify what opportunities there are. I suppose I do think that any input that we have in implementing prop-062 and we have preserved a /8 for the principles for the fairness of distribution, I think that I'm probably concerned about the proficiency of distribution along the way. I absolutely think that the proposal as it stands probably does need to be sent back to the mailing list given the development of the discussion. I really don't think that it is in a position to be taken to consensus. But that's of course my opinion, not the rooms.

I'm concerned that at the end of the day, a need for addresses is a need for addresses. As long as addresses are being allocated efficiently, and where there is clearly the demonstrated need to supply the users for whom so ever the users are, and maybe there's the constraints of use within the timeframes, I think that's a key thing that needs to be addressed and ensured at the end of the day. The question about should large blocks, you know, should you be trying to protect against large blocks too much? Or rather, if you did cut down in repeated blocks, should you limit to one a Monday or something like that and I think that needs to be considered quite carefully. Thank you.

JONNY MARTIN: I gets the first point there is that the final /8 is aimed primarily at new entrants at that point, and also catering to existing LIRs who get to the end and realise - oh dear, I actually need some v4 for transition technologies and that's the prime motivation I read into the use of the final /8.

The second point is an allocation algorithm such as this actually increases efficiency of new allocations from this point onwards because it is insenting LIRs to have a smaller amount of space. So sure, you might be able to justify a /8, but you might have a chance and at least a short delay if you go for a'/9. If you go for a '/10, there could be a lower wait. So it certainly going to achieve that.

JAMES SPENCELEY: Also it should keep the address space around a lot longer so you may get the chance of a'/12 and not necessarily a /8, and in that respect, that's a good thing.

OWEN DELONG: From the Jabber room, Andy Linton says "We've agreed on the rationing once the last /8 is allocated. Let's get on with getting v4 out the door and make IPv6 more attractive'

IZUMI OKUTANI: A concern that we have about this proposal is that it would put a restraint, constraint on the maximum size that the LIRs can receive, so that might effect it. You know, if they need a certain amount of space for their business, APNIC will be saying no. So that might be saying, you know, how much more that APNIC should say. If that's what the customers know, they can't provide it because APNIC is saying no, so that's the concern about the proposal.

JONNY MARTIN: Not wanting to tread on the chair's toes here, but does anyone support the original prop-070, the hard maximum limit on allocation size? That's probably the first question? If there's no-one kind of prepared to support that, then we may as well abandon that and spend a few minutes discussing potential other ways of achieving the same thing.

OLE JACOBSEN: Can I make a procedural request, that you increase the minimum size of the font so we can read it back here if is a PowerPoint. If it is a PowerPoint.


OLE JACOBSEN: It's really hard to see it.

JONNY MARTIN: So I'll take that as a lack of support for any hard maximum limit on allocation for the purposes of this discussion?

JOHN SCHNIZLEIN. Have you considered the risk that this form of rationing might push ISPs into radical NAT and walled gardens in a way that following through on the exhaustion of the limited pool which under the current rules does not?

JAMES SPENCELEY: Well, I suspect that running out of IPv4 is going to push them into a similar sort of area. Half the talk this week is relate today carrier grade NAT. I think yes it possibly may. Maybe that's just preparing people.

JONNY MARTIN: So the concern that because we're prolonging the run-out or the allocation of the last bits of v4, that people may get more used to the idea of running NATS.

JOHN SCHNIZLEIN: It seems to me that the rationing proposal says that instead of using our current policies for allocations until it is gone, we are going to give up smaller amounts of space to the requesters.

JONNY MARTIN: Only if they ask.

JOHN SCHNIZLEIN: So for a longer period of time, you're encouraging operators to make do with a smaller number of addresses. That seems to me to be a different message to operators than... there are no more. Find another protocol. And find another protocol message would come sooner if you didn't lengthen the process with - try to make it work with half as many addresses, because for more people than I would like it to be true, make do with half as many addresses is way too encouraging of sharing an address.

OWEN DELONG: So again, we have from Jabber "I would be concerned if it was only in the APNIC region and not other RIRs. APNIC will be saying no because we haven't got any shortly."

PHILIP SMITH: Philip Smith Cisco. So, in prop-063 which was reducing the allocation time from twelve months to six months, we were talking about increased workload for people to apply for address space time and again and quickly and we were also talking about, I think it was Andy Linton mentioned that the service providers were providing big networks to get what they need or something along those lines. I struggle to see how this is any different.

This seems to introduce a huge amount of extra work and ISPs are just going to go back, back, back, back until they actually get what they need to deploy the network and that leaves me with the question we were talking about in Christchurch, why are we even going here when we should be looking at what to deploy here.

JAMES SPENCELEY: I agree completely. The first slide of this is very similar to prop-063. One is dealing with looking forward and how far and one is saying here's your maximum. And one member couldn't take a large amount of the pool and surprise us all. That's a good insurance policy on its own. It's through lots of input it got more complicated. That may be a bad thing. It sounds like it is.

IZUMI OKUTANI: So, if we have to choose between prop-063 and this one, although we didn't support prop-063, I think prop-063 is more preferable, because restraint is added on everyone, not just large ISPs. Whereas with this one, it is only the large ISPs where the restraints are added. But of course, we shouldn't give the disadvantage to small ISPs. But at the same time, we shouldn't do the same thing to large ISPs as much as we shouldn't do small ones. And since we didn't support the earlier one, we also don't support this one as well.

RANDY BUSH: further comments? OK, how do you want to poll this? Because you did change it and segment it?

JAMES SPENCELEY: I don't think that we're after consensus today. It doesn't seem that there's been a lot of support but I guess we encourage people to talk on the mailing list and come up with a solution that I think addresses a fair allocation of the resources, thus everyone, I certainly think an insurance policy of a large member taking a lot of area is a worry. But certainly to encourage people to discuss.

RANDY BUSH: Jonny suggests that we ask are people happy with the current status 'quo of the allocation. I of course am abused because of the current status quo. Ask as you wish, Sir.

JONNY MARTIN: Is it worth anyone working out any more policy proposals to change the way that addresses are currently allocated?

RANDY BUSH: I think that that would be interesting. If you remember, Philip said prop-063 which uses the other time dimension mount to have tail has workload increase etc. There can be a number of proposals in the tapering down philosophy.

Does the group think that some work should be done in that space, that tapering down kind of proposals, whether it be the amount or the time or the flavour or the colour be worked on in this SIG?

Those who think so, please raise your hands? Whoops. So much for that. Thank you.

JONNY MARTIN: Easy. Thank you. OK.

RANDY BUSH: There's no slides. We didn't get to talk this time. This time we're going to talk about it. I think we can find it on the summary of proposals. 4 pm and 9 am, that would be... and that would be proposal summary and that would be and now I have to figure out how to operate the computer.

OK, change the criteria for the recognition of the NIRs in the APNIC region. It is hard to create a new NIR because it requires both the community and the Government in the relevant economy. And it allows the Government to have a heavy hand. So, it proposes to allow NIRs to be approved only with community approval. NIRs would be created through a vote of the APNIC members in the economy. Is that a fair statement, Sam?

OK, and not only that, it would limit Government positions on the boards of NIRs. Would somebody familiar with this proposal, does it limit the existing NIRs or just new ones created under this proposal?

OK, there was much discussion on the mailing lists, all of nine posts. My memory is that... and there was a message describing all of this thanks to the diligent work of the Secretariat polling or going through the old list, but essentially it was that people expressed a fear that 12 people could crawl into a room and declare an NIR which was not representative of the entire economy and that's why Governments, etc are involved. Those proponents of the proposal were frustrated at not being able to start new NIRs. Would anybody care to discuss this proposal?

NARESH AJWANI: NARESH AJWANI: The author has put his views in the proposal through five fundamental processes or logics. The initial logic is that there is a change in growing economies which happened in the year 2008 and we all know about the change in the scenario since 2008. It is much, much more than what had happened in 2008 and the role of sovereign in the current economic meltdown is something unlike what had come in this initial proposal.

Numbers 2, 3 and 4 are also paradox.

The policy is on endorsement and I'm sure we all understand the difference between endorsement and control.

Number 3 is again a paradox. It talks about policy which is indicative versus restriction.

Fourth paradox: from the country where this proposal has come, the Government and the Regulator are two different bodies.

Lastly, it has been camouflaged with the security perception from the Government's side.

Looking at the dictionary and the newspapers of the last year, it would be sufficient for us to correct all above because both would indicate the contrary's situation. Thank you, Chair.

RANDY BUSH: Does anybody wish to speak for this proposal? May I ask those people who support this proposal to raise their hands?

May I ask those people who do not support this proposal to raise their hands? So there is anti-consensus. And therefore is dropped.

OK, that is the only one that I had. Sorry, I have a poor memory

NAO FUKUSHIMA: I report for IPv4 at the end of last year. We introduce these items and the objective and the result of this trial. As a case, we don't use two services. The trial is as we have reported many times. It is time-limited address leasing. And one is extra in Phase one to 2001 to 2005 and executed from January 2006 to December 2008. And this trial ended at the end of 2008. The participants are there.

This is the final allocation list of this trial. We have used it for 2, 2, 4/11. We have an example of the address space now.

In this, we have reported - we have received some results from the trial. The main part of the address space is as follows. The address space and it is tied into the individual management of equipment by fixed IP address is lack of demand. The access network doesn't support ISP can not support too. Our other problem is the access network doesn't support IPv6 in Japan. Deployment is difficult only in the last ISP.

The service the service name is IP business phone and in this case, the phone was set up on the base of 280 or more. In this case, the provider was able to do only by seeing the other designs, large, middle and small. Because there was an Internet Protocol address.

The process of the other design decreased, therefore the provider was able to reduce the cost.

The other can be set automatically. The provider can set the labour costs and it costs more than the IP service.

And the convents delivery network and the CDS is the contents to service for IPv4 and IPv6 dual support. And Dura Site-aDNS. And corresponding to IPv4 and IPv6 that the user is using.

Result of this prom pro-gram, we think that the APNIC support made this program happen and acceptable. Thank you very much. From this trial, many lessons from the IPv6 transfer have been obtain and shared with this community. IPv6 PC will continue to support in the AP region. That's all.

RANDY BUSH: Thank you. So under the policy as agreed, I believe the IPv4 /8, which was allocated under policy to be used in this trial was agreed to be returned on February 1, 2008? January 1 2008.

RANDY BUSH: This is a /8, we're running out of them. Are we going to get it now.


PAUL WILSON: The address space was taken for 43 /8 but it wasn't used for the trial, so the amount of address space which is coming back to APNIC is a /11 in fact, not a /8 and that is in progress at the moment.

RANDY BUSH: What policy is the remaining space?

PAUL WILSON: 43 /8 is an historical allocation so it was taken by the IANA as an assignment to WIDE. It is now under administration.

RANDY BUSH: As legacy space.

PAUL WILSON: But it is only a /11 within that block which has come back into the RIR system and it has come back to APNIC/JPNIC. The rest is all historically allocated or assigned as it was to WIDE.

RANDY BUSH: There are some that have been returned and the rest is actually a legacy allocation by the IANA. Nope, I'm still wrong.

IZUMI OKUTANI: I don't think the rest is returned yet. Ms intercultural liaison.

IZUMI OKUTANI: My understanding of this, yes, I think it is the '/11 that should be returned to APNIC. I agree with Paul about that part, and about the remaining part that should be returned, I think IPv6 promotion castle is still working on it, so the space has not been returned to APNIC yet - is my understanding.

RANDY BUSH: Paul, come to the front. Good try, let's get the cards on the table.

PAUL WILSON: The address space took a subject to the trial. The subject of the trial was the total of the /11 so the reminder of the /8 is


PAUL WILSON: WIDE space. Because the administration at the IANA has changed though, APNIC is now the administrator of that space, although that has not altered the disposition of the majority of the space, everything except the /8 that is. So that is basically where it currently is, and the /11 is in the process of being tidied up after the trial. OK, we agree.

RANDY BUSH: Thank you.

KENNY HUANG: I remember this trial has been continuous in the presentation of the Policy SIG and especially, as far as I remember and the trial from the /8. So just trying to clarify whether the /8 in the experiment is correct. Because we have a specifically defined time frame for the trial, so no matter - IPv6 promotion keeps going or not, the trial is basically complete from the APNIC point of view. Thank you.

RANDY BUSH: So, whoops, Philip!

PHILIP SMITH: Philip Smith Cisco. 43.232 and upwards is still appearing in the routing table, but I thought that I would let you know.

RANDY BUSH: I don't know what we can do about that. Thank you. Moving back to transfer, I'm sure that this is exciting for everybody. But look at it this way. If we just do these three, we can go! So, this is what we believe was agreed. Yes, thank you.

For the multiple points during the transfer discussion. APNIC will process IPv4 address transfer requests involving the current account holders following adoption of this proposed policy. Subject to the following conditions. The minimum transfer size accepted will be a /24. Until such a time in the prevailing APNIC IPv4 allocation practices, practice uses the final /8, policy 1, that's the current policy, the recipient of a transfer is to justify use of the transferred space, using the allocation and assignment policies enforced at the time of the transfer. After that time, no justification is needed. Organisations using this to check the policy are not available for IPv4 assignment or allocations for two years. APNIC is to maintain a public log of all transfers.

Address transfers should be permitted between APNIC account hold and NIR members if and when individual NIRs implement the transfer policy.

NIRs are tree to adopt the transfer policy for their members in their own time and to use locally determined policies for their NIR membership.

I'm sorry, we did not agree on locally determined policies. It was locally determined win. Address transfers are perm mitted between APNIC account hold and other RIR account holders following the policies of the respective RIRs. This proposal is to take effect as soon as the APNIC Secretariat can implement mechanisms of the policy.

Does anybody disagree with... that these are what we discussed and agreed. So looking for consensus on this statement as a whole.

OK, could those people who support this proposed policy as a whole please raise their hands.

Those people who object to this policy as a whole, please raise their hands.

Those people who would be extremely unhappy without this policy, please raise their hands. We did not have a decisive vote.

GEOFF HUSTON: With or without?

RANDY BUSH: Who would be extremely unhappy without this policy. In other words, who just have to have this policy. Please raise their hands.

Those people who would be extremely unhappy with this policy, please raise their hands.

I think we do not have consensus.

Thank you, all. Does somebody have any idea of how to go forward. We look at it piece by piece, we read it piece by piece and when it comes to voting on it, the answer is no.

NURANI NIMPUNO: I'm sorry, I don't have any answers. But I do... we've been discussing this all day. We've been going through each and every point of this policy. We fought it out and we reached consensus on every point.

When we finally move on and then write it up, we don't reach consensus. I find this very strange and I think I'm a little bit concerned, I don't think we'll ever reach consensus on anything if we continue this way.

RANDY BUSH: The reason you don't understand is that you don't have NIRs in your region. I was trying to be polite! Essentially, it was block voting by NIRs.

DAVID WOODGATE: So, I was just wondering if anybody who just proposed the proposal as a whole just then would be able to come to the mic and explain their reasons, given the long discussions that had been held today and there was no general consensus. That would be very helpful to understand what the reasons were for some of the objections.

RANDY BUSH: You're asking for those who opposed it as a whole? Or supported it?

DAVID WOODGATE: Opposed it. Who opposed it as a whole, please speak. Anyone from CNNIC, JPNIC care to explain?

HYUN JOON KWON: Because we discussed the items, which items we have in consensus, but when you combine the issues together, they become like a new policy. And if someone disagrees with some parts and you combine everything and then as a whole, maybe some part where you have a disagreement with that part, because of that, you agree with the whole policy, so that's the problem I guess.

RANDY BUSH: OK, then let me ask this. Of those people who objected to this, how many would object to only the first point - APNIC should have a transfer policy?

OK, then let me further ask, who would strongly object if the proposal was only db APNIC will process IPv4 address requests following the adoption of this proposed policy subject to the following conditions. Maintain a public log allowing transfers between APNIC members and NIR members when NIR members implement the policy and transfers between other regions.

Let me review, only the base - 4.4. The first paragraph in 4.5. And 4.6. Would anybody strongly object to that proposal?

Who would support that proposal?

Just to make it clear, I'm going to do a quick edit. So we're agreeing on what we think we're agreeing on. Is it OK to add a /24 in? Anybody strongly object to the /24?

No, OK, we'll throw that in. Let's throw the evaluation and justification out because that's going to be a contentious one. I have to figure out how to do "page down" in Power Point.

KENNY HUANG: I want to update to update the proposal.

RANDY BUSH: Modifying prop-050 or 67 so that it is an existing proposal. Right. OK. So, here's what we are left with. APNIC will process IPv4 address transfer requests involving current account holders Following the adoption of this policy subject to... the minimum transfer size accepted will be a /24.

APNIC has to maintain a public log of all transfers.

Address transfers should be permitted between APNIC account holders And NIR members, if and when individual NIRs implement the transfer policy. Address transfers between APNIC account holders And other RIR account holders Following the policies of the respective RIRs - I'm going to insert the word "all" here, just to be very clear here. All the RIRs. The proposal will take effect as soon as the APNIC Secretariat can implement the mechanisms of the policy.

Does anybody have strong objections to that proposal? James? What's your problem?

JAMES SPENCELEY: Did you take out "subject to meeting the current criteria?" Is that what you took out?

RANDY BUSH: Yes, I did. OK, James wants me to put in the "current criteria" again. Does anybody strongly object if I put back in that you have to justify receiving the address space? OK, we have a problem. So for the moment, I'm going to leave it out and we can have another policy proposal later trying to do that. Philip?

PHILIP SMITH: Philip Smith, Cisco. I would like to hear from the strong objectors To the package why they object it. I think that would be useful?

RANDY BUSH: I asked that question and the only one who spoke up was Kenny who said - individually, it sounded fine and when you combined it, it tasted horrible.

PHILIP SMITH: I don't see the difference. Somebody should explain exactly what the difference is, not just make up things that wastes more time in this space.

RANDY BUSH: Just a second. You might use the microphone.

JIAN ZHANG: I think probably, just my guess, probably because some people are not comfortable because we reached consensus on each element individually, but definitely not everybody agrees on every element, right. So maybe because some people are comfortable with some certain elements and some are not, so as a whole package, that's why we couldn't reach consensus. Does that answer your question?

PHILIP SMITH: Not at all. If people are taking part in a debate and you're not aware of what you're talking about, you should not participate. Right. We're discussing each individual piece as part of a greater thing. We're not discussing each individual piece as an item.

RANDY BUSH: But, to be fair.

JIAN ZHANG: I don't agree. I think you're forcing people to...

PHILIP SMITH: No, I am not.

RANDY BUSH: Philip, to be fair, when we were discussing green, we did not yet know how yellow would be decided. So those people who supported green might have been happy with green but no yellow and... I can see it. I mean, I'd like it but... Izumi-San?

IZUMI OKUTANI: Just talking as an individual, not as anybody. You know talking to some people during the break, I got an impression that not everybody fully expressed their concerns that they had about the proposals, so I think would those people who raised their hands against it and hadn't expressed their concerns for the proposal, I think now is the time to express it. That's my suggestion.

MASATO YAMANISHI: I completely agree with the objection comment is without concrete comment, which part of the combination is not preferable and why it is not preferable. We can not move forward.

RANDY BUSH: I tried to take from the subset.

MASATO YAMANISHI: I don't object to the chair, I'm saying my comment to someone who has a strong objection to this

RANDY BUSH: I believe Geoff was next.

GEOFF HUSTON: Geoff Huston APNIC. But not as an employee, as a member of this community. So, here we are so close and yet somehow, just one step away. And I look at this and I go - I really feel from my heart that this is a deep problem that we have to solve. So the first kind of thing I ask myself - those folk who said - I have a really strong objection here. I just can't live with it if we did it here. I sort of have this issue of, can you not live with this particular flavour of outcome, or are you simply saying - I don't care about what is going to happen, we will not do transfers no matter what?

RANDY BUSH: I think you know the answer to that, Geoff.

GEOFF HUSTON: No, because if it is kind of a little bit, it would be really good to understand what needs to change to accommodate it. But if it is no matter what, then I will give up, we'll all give up. Because if there's no way of achieving this, life is too short to keep on battering up against it.

RANDY BUSH: If we went through the proposals that we thought no-one would object to and I believe that question was answered.

GEOFF HUSTON: You believe that we're close.

RANDY BUSH: I believe that no-one, only one person strongly objected because he wanted something put back in that I could not get consensus on. I believe only one person strongly objected to - there will be a process, a minimum allocation size, public log, NIRs when they are ready, inter-RIR when following the policies of all the respective RIRs. I believe no-one strongly objected to that.

Does anybody strongly object to that?

JAMES SPENCELEY: We spent a great deal of time debating and reaching consensus of this, I don't think in the last five seconds of the meeting we can start randomly pulling things out. I'm finding it difficult to find what we're pulling out and what we're not. I think we need to potentially go back to the original proposal as we all agreed in the meeting told and discuss if there are specific issues rather than just randomly pulling things out. Then let me ask this question, those people who strongly oppose the total proposal, are you willing to negotiate further? Those people who do not think no matter, what they can live with the total proposal, please raise their hands.

WEI ZHAO: Since I'm the person who is not very confident with this proposal right now because I sit very quiet in front of the room. I didn't realise how many people with me, but I just want to clarify that why against this proposal right now. But I would like to clarify about the proposal we are talking about and think about the coverage to do nothing. Still moving on, why people are so happy with the recovery proposal that we reach consensus, and why people are not very happy with the transfer that we reach it. I believe that people have the same considering of if they will bring the biggest disadvantage later on.

Think about last time in we reached consensus on the last /8. I'm not judging here if that's OK or not, but after the last /8 proposal has passed and was reached consensus, there was a lot of wondering in the mailing list. I don't want to see that this proposal has been rushed into agreement so that people are still wondering about doing It right or wrong, so that's why I'm not very in favour of this proposal.

RANDY BUSH: We've been discussing this proposal for two years.

WEI 'ZHAO: Also, I was strongly against the timeline right now because we want to implement it straight away, but I just didn't get the idea of why we have to have it straight away, right now, even though we still have lots of, you know the IPv6 addresses in the free pool. Maybe some part of the organisation - they say OK, we have the addresses which we have with companies and other parties. It is what it is for the past ten years. Why can't we is wait a little bit more time to make everything clear. We reached consensus and it is not extremely necessary right now, from today.

DAVID WOODGATE: After everything that we've seen in the past few minutes, the only options Randy and I can see going forward are, you know either continuing with the process we were going through or frankly, looking like you have to break it up or intercept the proposals and every one in separate proposals and take it back to the list. I can't see any other way around it.

JAMES SPENCELEY: If the people objected, would there be interest in supporting the consensus that will be discussed at the next meeting. How else are we going to reach this unless we set ourselves a date?

RANDY BUSH: To make it clear, no-one strongly objected to a reduced proposal. except for you. In other words, I think if we stop right now and we can ask again and we can pass with this reduced proposal set.

JAMES SPENCELEY: I'm surprised by that. I spent a lot of time today and a lot of people wanting justification for the resources, but if that is going to get the proposal passed.

RANDY BUSH: We also suggest that we can keep talking and sink it. Do you withdraw your strong objection?


RANDY BUSH: He did, a while ago.

KENNY HUANG: I fully agree with that. There are very similar anticipation and some slightly different. So I would suggest a single proposal in different pieces, probably can get 90% consensus, thank you.

RANDY BUSH: Kenny, you were before Izumi-San. Is that it?

HYUN JOON KWON: Also, I'm agreeing with Kenny's comment and actually when we came here with the transfer policies so with the original one, that it does not include the NIR transfers, so at the time, we had a problem with that so I came here for the policy, but suddenly the policy came up and it was including NIR members but I wanted to you know ask the members to think about it. Actually, I'm really against this timetable, it was hard for me.

RANDY BUSH: The timetable does not bind NIRs.

HYUN JOON 'KWON: But if we split like the original three proposals, I believe we can save it. Some of it.

RANDY BUSH: That's what I'm trying to do here. Izumi-San.

IZUMI OKUTANI: I see two concerns expressed by people. One is about the process that you know we have three different proposals and then all of a sudden we come up with one, and the second is that people have to have remaining concerns about the details of the proposals, but then if we keep going, it's not just the community you know which direction should we move? So I would like to first at least fine out whether people agree with the concept of the proposals and we should, you know, work it out. Because I don't think we've reached that point yet, we're not really sure if we agree with the idea of the transfer.


PAUL WILSON: Just a small point. Paul from APNIC. I understand there is a regional concern here that we have started off the meeting with two separate proposals which have two separate titles and proposal numbers and that we haven't formally dropped either of those and we have another one which is now not identified separately and I think it might be - what I've heard suggested is that we actually formally drop one of the proposals, I done know which one. And take the other proposal as the one that we are discussing and specifically make the modifications to that. So that's the procedural point.

RANDY BUSH: Philip, are you happy dropping that in proposal. This is proposal prop-050 as modified. What's more, does anybody strongly object to prop-050 as modified? Resulting in the following, APNIC will process IPv4 address transfer requests involving current account holders following the adoption of this.

4.4, public log of all transfers. APNIC account holders can conduct transfers with NIR members and when individual NIRs implement the transfer policy. And address transfers are permitted between APNIC account holders And other account holders following the policies of the respective RIRs, and this proposal takes effect when the Secretariat can implement it. Does anybody strongly object? Does anybody have a problem with declaring victory and going to dinner?

DAVID WOODGATE: I think you need to specifically ask who supports?

RANDY BUSH: OK, does anybody support this proposal as I just read it?

Thank you.

Does anybody object to declaring victory and going to dinner?

I hereby declare victory and let's go to dinner. Thank you all very much for your patience.


For your patience during the day. It's very hard to deal with serious and substantive issues and compromise.

KU WEI WU: The other proposal that was dropped?

RANDY BUSH: We agreed to drop it, Phillip and I agreed to drop it.

JAMES SPENCELEY: Just a comment, I don't necessarily agree with this process. This is basically a serious decision. I am not against it but the process of getting consensus here.

RANDY BUSH: Come beat me up about the process, please.


And hello to our friends in Colombo and thank you very much for your patience and for participating, it is appreciated.

SRINIVAS CHENDI: So, do I go for the last minute transfer proposal to cancel the dinner tonight! Tonight's dinner is at Bella 'Vet and the buses will leave at the same parking lot last night. Before we go for dinner, just a reminder that the member meeting will commence tomorrow at 9:00. The registration desk will open at 8:00 and you can collect your ballot papers as well at the registration desk as well as APNIC proposal USB wrist band, 1gb. We have orange and blue colours at the desk. The key learning demo is outside of the member services launch. If anybody is interested in the APNIC E-learning, you can ask anybody and their staff to help you.

Tomorrow we have IPv6 server prizes as well as member server prizes. And if we have receives any proxies, please ensure that you receive confirmation from the APNIC Secretariat before you collect your ballot posts tomorrow. If you don't have that confirmation, you have to check if you have voting rights to vote in person here tomorrow in the morning.

(End of session)

The first bus leaves at 6:00 from the same place as last night. And the last one I believe is at 6:20.

And we would like to acknowledge Owen for volunteering to be on the Jabber monitor.