These are unedited transcripts and may contain errors.
Notice: Use of undefined constant steno - assumed 'steno' in /var/www/html/ripe-60/steno-transcripts.php on line 24
The Address Policy session commenced on 5th of May, 2010:
SANDER STEFFANN: Good morning everybody. Welcome to the Address Policy Working Group. We have three sessions this week, the first is now; the second one is after the coffee break; and the third one is tomorrow, so...
I will first show you the agenda for this week. First, we are going to do the administrative matters, the notes from the last RIPE meeting and things like that. Then, Filiz is going to give a presentation on current policy topics and afterwards we will discuss some new proposals we had since the last RIPE meeting.
Then, after the coffee break, Filiz is going to give a presentation about the cosmetic surgery project, the cleaning up of the RIPE documents, and then there is going to be a presentation also from Filiz about the author RIPE of RIPE policy documents. The next topic M, is going to be moved to the Cooperation Working Group, so that will not be discussed here but I just wanted to mention it. And then we are going to talk about the differences between IPv4 PI and IPv6 PI space. And then at the end of today, just before lunch today, Rob is going to give a short presentation on the need for registration policy.
Tomorrow we are going to discuss all the open policy proposals, and as usual, with the open policy hour and any other business.
So, this is the agenda for today, but more detailed. I will just skip that one. These are the new proposals we are going to discuss today, the proposal from Nick Hilliard about temporary assignments of resources. That is going to be discussed first. Then we are going to discuss the allocations from the last IPv4 /8, which is a new proposal, I will talk about that later. And then at the end we are going to discuss the global policy state in the RIPE PDP and Dave Wilson is going to give a talk about that.
So first, administrative matters, I would like to thank the scribe for making all the notes and I know it's a lot of work. And then agenda bashing. Does anybody have any suggestions or changes for the agenda as presented? OK. Then we will keep it as is.
Then we need to approve the minutes from the last meeting. They have been circulated on the mailing list. Does anybody have comments or maybe found an error in them? Something missing? OK. Then I want to declare them approved. And then Filiz is going to give a presentation on the current topics. I am sorry, Filiz just reminded me, I am doing the Address Policy Working Group by my sieve this time because Gert is ?? isn't ?? isn't at the location right now so I just wanted to let him explain that himself.
GERT DOERING: Good morning Address Policy Working Group, and it's nice to see so many of you there, but there is still some seats empty, so we need to get more people in the room. It's quite an unhappy event that I cannot be with you and run the show because well, it's important work and it's fun, but I am sure that Sander will do it absolutely perfectly, even if he is a bit nervous at the moment right now. Yes, well, the reason why I am not in Prague is a happy event on my side, we are going to have a children tomorrow or Friday, so this is really a conflict of schedules and there is a few things that is more important than a RIPE meeting and I think, well becoming a parent is more important, so it's possibly that my phone will ring in five minutes and then I am off for good, but until then I will stay in Skype and try and follow what you are discussing, I will ?? I will insert a comment here and there. OK, back to Sander or back to Filiz.
(Via Skype)
SANDER STEFFANN: If you hear him speaking, he is participating remotely.
FILIZ YILMAZ: Hello. Filiz Yilmaz, before I start, I just want to give a few topics for feedback to Gert that it's very clear here and we can hear you very well, together with the video, so please feel free to drop in whenever you feel necessary.
So the ?? I am sorry, this is not my presentation. I think I have done this yesterday already:
GERT DORING: The presentation is on the web page so I could open it and stream it to you if you want.
FILIZ YILMAZ: We got it here too. So this is the current policy topics presentation and you weren't dreaming and you weren't left, you know, on yesterday, so everything is cool, you are awake. The point of this presentation is to give a RIPE update first and then go on with what is happening on the other side of the world. We are doing this because you guys previously asked us to follow up, spy on, what is going on in other regions, and report back in these sessions. And at the end I will have some promotional slides, as usual. Quickly starting with the RIPE update:
We had two proposals accepted recently. The first one is run out fairly, 2009 ? 03 and this has already been accepted in the last month of 2009, and it has been already implemented, and we are actually very close to get into one of the phases that is pointing out so this is maybe quite important for you. This one is basically gradually reducing the allocation and assignment periods. What does that mean for you? When you come for further allocation, at the moment we are considering 12 months of your needs, your needs spread around in the next 12 months. This is what we consider. With this policy being implemented by July, we will start looking at nine months of needs, and then that will go down to six and then eventually three months. So it may be an idea to actually follow this and make sure that you are up?to?date with this because our registration services department will point you out ?? starts pointing this out to you vigorously soon. So this is coming. Internet assignment numbers authority policy for allocation of ASN blocks to regional Internet registries. This is basically the global policy describing how the ASN blocks from IANA to RIRs should be delegated, and why we had this recently, this was because of the rather slow deployment of ASN 32s. The reason for that, this slow uptake had an impact on the global policy because once the pools ASN 16?bit and 32?bit blocks are ceased to are differentiated by the RIRs and IANA, then due to the very low usage in one pool, merging them would cause, still, a huge gap in the combined pool, and IANA wouldn't be able to qualify us for a new block, so while there would be still some 16?bit ASN in IANA pool, the RAR, the RIPE NCC wouldn't be able to get those from the IANA, thus you wouldn't be receiving any 16?bit AS numbers, so the solution was to push that date, the affidavit the ceasing of the distinction between pools, to one more ?? with one more year. So this will only happen now in December 2010. We have a little more time here. However, just not to have any déjà views, by the end of the summer 2010 regarding this, I want to make a little point of reference or mention something in parent sees; we realise that, still, the one?third of the ASN 32?bits, so the 32 AS numbers that are assigned, are coming back, with the reason that they can't be used because of hardware or because of peering problems, and so the peering party or the other party is not, yes, they are not able to connect through that AS number. So, this is still ?? this may be still, you know, this should be made in your awareness still and something to look at. My colleague Anand has been using these ASN 32?bits on an operational level in his DNS services and he claims that he didn't have much problems, so if any of you are interested in talking to him, he is here during the week and actually he will present his experiences, operational experience, on Friday, here as well, in the plenary, but please catch him. If you don't know him, come ask me and I will point him to you. Maybe an idea to talk to him.
Anyway, coming back to the global policy. It was accepted in September 2009 and at the moment, it is still not implemented officially because there is ?? we are still waiting AfriNIC board to rectify the policy over there, so then it can go to the address support organisation which will then guide ICANN board to implement this.
Then, we have two withdrawn proposals and I am not going to get into the details of this, but these are ?? these were about the usage of the last /8 that the RIPE NCC will receive. Now, a little bit of history here since I have time this time, it's nice; the ?? all the LIRs ?? when the pool of IANA free IPv4 pool reaches to a level of, you know, the last one digit numbers, the last five will be distributed among the old LIRs so every will receive one /8 at the end of the pool and this is about that last /8. All the LIR regions and communities are now looking at how they would like to use that /8 ?? last /8, not because it is something maybe should be very different but it is just a consideration that should we do something special, should it be, you know, the allocations should be distributed fairly maybe, evenly, because it is basically the last one. And what is happening in our region was that there are two proposals relating to this same resource so these proposals were actually, you know, competing at some point and what happened there, the proposers decided to join forces and come up with a new proposal, which is now 2010?O2. Now, you may realise now I will start talking about new proposals and the ones under discussion, and I am not following order as I didn't stop with 2010?01, that is for a reason that things rerelate to each other somehow to my order relates to that logic, so you are not falling asleep here. So we are at 2010?O2: What is this about? This is again, you will see the details, Elaine will present the details of this today it, it's in the agenda but shortly it is talking about giving ?? sharing one /22 per LIR from that last /8 and the ?? still needs to be justified. The other condition is that LIR should have an IPv6 allocation either directly from the RIPE NCC or in form of sub allocation from another provider. 11th of May was the end of ?? is the end of discussion phase, and the latest comments about that was, basically, which last /8 are we really talking about? The very last /8 as a whole block that the RIPE NCC will receive or the very last /8 that will be sitting in the reserves of RIPE NCC at a given time, because those are two different things. Often, we receive a /8 while we have a bit more than a /8, and the letter is basically RIPE NCC raising a flag "OK guys, now we reached" in what form or shape it doesn't matter but we reached a total amount of last /8 in our reserves so now we are going to start doing this. So the implementation time, there is a tweet there and I believe Alan will be talking about this more in detail. The impact analysis for this will come only after we hear that the proposer and the Working Group chairs to move to to a new phase. 2010/01 ?? we actually do have temporary assignment policy, scattered around in our policy framework, this proposal is an attempt to bring all them together in a stand?alone policy and also refine criteria. The refinement in the criteria is basically focusing on economic research experimental purposes conferences and events and it is also introducing some time limits, quite strict time limits which is not in the current policy. It's talking about the assignment to be for a maximum of one month in the case of events and six months in case of long?term projects, with a clause that renewal will be possible based on the justification that can be produced to RIPE NCC. Now the latest comments were mostly in textual form how things were worded and as far as I understand, the proposer Nick Hilliard take them into consideration but he will tell us more in how they see them and 15th of February was end of discuss phase. Seeing there was interest from the chairs and Nick that they would like to see more progress of this proposal within the PDP we have already done some impact analysis. This is not published yet on the website, we will have ?? we will publish extensively, properly written way as the way you are used to seeing it but I just wanted to give you those ?? the little findings we already have because this may facilitate conversation and dialogue and discussion you will probably have today, hopefully.
What the policy proposal is talking about some ? for these temporary assignments and that is left to the RIPE NCC's discretion, so we looked at some recommendations, what would be logical from our point of view, what sizes and numbers could have been used for this and for IPv4/14 seems to be reasonable from our side and that is a little liberal over what we have actually as a number because that is/15 but on the safe side we propose/14. Which can be reviewed at a later time obviously, this is just a recommendation since the proposals pointing to such recommendation from our side. When it comes to 16?bit, 16 of them sounds a fair number, there is no magic there. It showed up like that. And 32?bit ASNs and IPv6 blogs we do not see any earlier advanced reservation because we believe there is plenty of them but if you would like to still reserve for those two we will do based on your recommendation. We also looked at some administrative stuff within this assessment we have done internally as of the impact analysis, and the proposal is talking about signing contracts for these end users who are coming for temporary assignments, and obviously, will needs to be a new contract template to consider this kind of resource. And there is a little thing there, we also notice, and this may be resolved by just changing the wording and we need to see what Nick is thinking about that, but the proposal itself, when you look at the wording, it very much focuses on the end user and we are thinking, what about an LIR? We understand probably you mean LIR can have end user role when it comes to temporary assignments but legally due to contracts we may want to still get that wording much more clearly in the proposal itself, not to run into problems later on.
It also makes a reference to conflict arbitration procedure. That will need an update. The reason for that, again the wording is very much around end users within the current wording of the policy proposal, while the procedure conflict arbitration procedure is very much talking about contributors, and within today's reality end users are not contributors, so this document will need an update, and I here there is already some kind of review going on there, this may happen, but we just need to point out because this is a GM?related action, so that you know.
Now, legal analysis. There is something there. We have found ourselves to consider and point out to you the proposal is talking about also issuing temporary resources for commercial tests. It bears that in itself. And when we consider depletion in IPv4 and considering the extreme case of IPv4, we thought maybe there might be a case quite a big organisation coming up justifying the need very well, all very to the book, quite a big chunk of temporary assignment for a commercial test, and then imagine at that point of time they take all the possible resources for that test, and then that may create some ?? the commercial advantage on their side and this advantage on the others because the others will never get a chance to test that thing commercially, because simply there won't be no resources to do that. So this might be something to think about.
Then, the ?? I am now moving on with more proposals. We have some from previous years, 2009?01. This is another global policy. This brought all to us a different level of thinking. This was an interesting proposal. This is, again, related very much to the depletion problem and what could be done in regards to that, not to, you know, say more resource to make it more fair generally and globally, and what happened there, it was proposing, you know, RIRs from time to time, we see returned space so unused space is coming back to RIR pools and this proposal when this happens, this returned space would go back to IANA and in IANA there would be a new pool generated based on that return space and later on in the second when IANA totally returns out they will start using that new pool to issue space back to all RIRs. What happened there, AFRNIC, APNIC, LACNIC and RIPE, we focused on one ?? the original text and we went on in our PDPs to consider that text but that didn't fit with ARIN community and they found themselves needing to make a change over the proposal so the text became different. And that kind of created some kind of global confusion together ?? procedural effects, so how that start happening and how do we deal with that situation, apart from the content of it the proposal which I will touch upon in a minute but the latest thing in RIPE discussion, RIPE fora, was that we were actually waiting for to see ARIN's latest version, and in the meanwhile, they published it, reached consensus over it and moved ahead with the proposal, so now, just before the meeting Nigel sent out a mail to the ?? to the mailing list asking the community what shall we do? Because at the moment, RIPE community looked like we didn't have an opinion in regards to the content of the proposal, apart from everything, procedural stuff is one thing but the other stuff, you need to make an opinion about this, and put this forward in the global policy?making process. And the response to that mail, as I understand from the chairs and Nigel and what I see from the mailing list, community generally agreed go ahead, we will discuss and conclude this proposal, one way or another, so this is, again, in the books of Address Policy Working Group agenda. It will be discussed here. But hearing that, we again performed impact analysis and these are the results we could come up with. What we see aggregation and fragmentation part, overall it is very hard to tell. We won't even go there to dream about something and make predictions. I think we will, yes, that is just ?? it will have an effect but, yes, it's not easy to tell. But we can talk about reclaimed numbers because we have some numbers in regards to that. Just as a point of reference, one million IPs per year is coming back to our pool so if this goes through this will be the number that we will be looking at, possibly. And we don't see any immediate legal impact to report at this point, but we are, you know, legal things do change so this is what we see at the moment. Please consider that remark within that disclaimer.
Now, then we have the RIPE policy update discussion for the other proposal 2010?03. This is the global policy proposal and this came into ?? under light in a way because of the global policy that I just mentioned. It was the first time there was a very substantial difference between two versions of a global policy, and communities found themselves especially RIPE in here, how can we deal with this, so how do we adopt ourselves to a change that happens in one community? And just a remark, yes, I am now going on, but global policies, they all need to be more or less the same in all the regions to be global enough to be applied in all regions, sorry, I realise I went a bit without that explanation. So having a change in the version obviously came up like, oh what do we do with this, so how do they adopt? And Dave Wilson has a proposal to have a new state which could be used for to make it much faster to have a response to that change, and latest comments are mostly questions about clarification and there will be a session here, so we will see how it goes.
Then, we have three other proposals that are sitting, still, in our agenda. The first one is certification proposal, Nigel will update us on this today. Ensuring efficient use of historical use of IPv4 resources. This is about asking for utilisation rate and making sure that they are utilised efficiently in legacy space domain. So ?? this is for legacy holders. And PI assignment size is about making the minimum assignment size to a specific /24. At the moment, we don't have that. We don't have a minimum or a maximum, so this is a new thing.
Now, worldwide view, what is happening in the other parts of the world? I will quickly go through IPv4 and transfers. ARIN adopted three?month allocation period for those allocations from the last /8. What happened there, they started a bit like discussions which are parallel to a run out fairly and due to their refinement, during their discussions they eventually agreed they will go for a six?month ?? three months fixed period for those allocations from the last /8. They are also discussing putting a /24 minimum assignment size for the end users assignment unit. I kept their terminology because if you go back, check their policy proposal archives, I want to you make sure that you find the RIPE proposal about that. But this is the PI in our world. And this is interesting because ARIN actually has already a minimum assignment size for PI, and that is a /22, and now they want to make it a /24. So, we are talking about just segment one so you can see different way of different regions. Waiting list, this is an interesting one because it's relating to some region, it's talking about there will be that time, an LAR will go to ARIN and they will ask for say/18 and at that point of time, ARIN due to the depletion, they want to have one block which is in size of /18. So what do we do? So to preserve the spirit of aggregation, these people will be given two choices: You refine your requests and go for a smaller chunk or if you really, really want to have a bigger aggregated block then you wait in a waiting list and when something comes free then you will be the first one to get it. So they are looking at it. This is under discussion at the moment. They found it interesting enough to continue the discussion about that.
LACNIC is discussing transfers and it is not very different than we have ?? what we have today. They are basically looking at a /24 minimum size for transferrable blocks, they are also having justified need considered but they are ?? they also have within the proposal, they have an item about having a publically accessible list of approved applicants so those people who are interested in receiving some space who can't it from LACNIC directly due to there is not enough any more, they will be published somewhere so those people who could transfer some of their address space, they know who to reach and how to reach so they are working on ?? this is an operational matter but they found that related enough to put it in their proposal. They will discuss this in the third week of May. So in a couple of weeks.
Now, IPv6, yes, that was the remark, that was the first presentation that was on the window that you see before this, which I claimed not to be my presentation. I gave another talk in IPv6 Working Group yesterday and this ?? the slides are there, so if you want to have more detailed information about in regards to IPv6, please go there and have a look, but the main thing, one of the very common things is the removal of routing requirement. The operational community is saying basically routing decision is an operational decision, nothing to do with addressing policy and it shouldn't be a requirement to receive address space. ARIN, APNIC, RIPE, reach consensus on that, we are applying already, and sorry APNIC hasn't implemented, as far as I am aware, and LACNIC is discussing.
Now, all these, more details are on these links. I just put it there so if there is anybody there still wanting to know more about this, go there click the link and you will find the archives. I want to finish off with some PR as usual. I always check on the subscriptions and how many people are on Address Policy Working Group mailing list and it's increasing, it's really nice to see. There is a steady degree so people are getting subscribed to. I don't know if they are reading the mails, but they are on our books. That is good. More PR in, regards to if they are really reading the mails, more hint, a bit more hint because you never know, I also checked how many posts are sent to the mailing list, and that is also increasing, so the monthly average is ?? has been going up since 2006 steadily, so this is nice, you were talking and please keep up.
I have a new colleague, Emilio. He is here, too. And it is his first RIPE meeting as a policy development officer. Lovely Ingrid has decided to go back to her first home in RS so there has been a little change in our staffing. He looks like he has those angel wings at the back in this picture. He is an angel. And easy on him. But if you have any questions, you can ask him any of us, as well as the Address Policy Working Group chairs here, we will be all around this week.
Thank you.
(Applause)
SANDER STEFFANN: No questions? One tiny remark. Filiz said that the certification proposal would be discussed later today, but that is tomorrow. So just for...
We are going to discuss the new policy proposals now, so some guidelines for the discussion; first of all, for the newcomers, we don't make decisions here; we just discuss the policy proposals. All the official stuff is going on in the mailing list and that is where we try to get consensus, so while we do have some serious discussion here, we will always finish it on the mailing list.
Yes, please remember to speak into microphones and when you do, state your name and affiliation, especially since Gert is now working from a remote location, he cannot hear you if you don't speak into microphone and all our other remote participants of course. So please remember that.
It is webcast, remote participation is by Jabber, so for all those who are not physically here, if you want to participate, just use the Jabber channels and we will make sure that your voice is heard.
Now, Nick Hilliard, he is going to give a presentation on the temporary assignments. Filiz already gave a good introduction. So I would like to give microphone to Nick
NICK HILLIARD: Good morning everybody. My name is Nick Hilliard. And I am CTO of INEX. And I want to talk to you about this temporary direct assignment proposal that was first talked about at the last meeting.
Of course, we have had experimental assignments for years. There is a very limited context in which they can be applied for which is to say academic research only. And where the results of the academic research must be published. And consequently, the usage of these has been rather low but the idea is codified within RIPE policy and I think we are all fairly used to it at this stage. In the future, there is going to be an all of lot of constraint on address supply, but there will always be a requirement for IP addresses for short periods for people who need to use them for conferences or projects. So it would seem to me to be a sensible thing to do to codify this into RIPE policy as well. This also applies to ASN 16s because they are a very limited resource and we are nearly out of them.
So the proposal then is to make a generic, a temporary assignment category. The policy proposal itself doesn't really draw any distinction between the type of number that the proposal applies to; it is it includes ASN 32s and 16s and both IPv4 and IPv6. As Filiz mentioned, it's not particularly applicable in any meaningful sense to 32?bit ASNs or IPv6 because there is such an abundance of supply, so really, those were put in there in order to make the policy more consistent.
The categories under which you could apply for Internet addresses on a temporary basis are listed, a couple of samples are listed but I think the important thing is that we don't know what end users are going to use these addresses for so there is a good deal of latitude open to the IPRAs, they can use their judgement here. Some people may not think this is a good thing; I think this is a good thing. Yes, the utilisation rates are different. Two years is is a very long time for a permanent or at least for a temporary assignment and I believe that in the context of a very constrained address supply, six months is a much more relevant time period. So we have changed the utilisation rates. There is the possibility of an extension and the current proposal mentions an appeal to the RIPE arbitration board, which deal with other aspects of RIPE or judgement of RIPE policies and how they would apply to end users so I think probably it's applicable to under users over there as well if they feel they need an extension for whatever reason. There is no B/S clause. There is an awful lot of people who apply for address space and one suspects that there are not, that they don't always tell the truth in their assignments. Because these addresses are going to be available on a permanent basis in a constrained market, I don't think it's unreasonable to expect that people would be required to give a little bit more justification for using them. And in every other way, they are the same as regular direct assignments.
So, from the RIPE perspective, the RIPE NCC are authorised to reserve a bunch of pools, which is good. There are a couple of recommendations made to the database format, in particular there is going to be a start and an end of assignment. And this is mostly to deal with abuse and black?listing and that sort of thing, so that when, for example, an address makes it into a black list of some form, that the black list hoster will be aware that at a particular time, it will then be appropriate to remove that address from the black list, because it's been de?assigned. And there is going to be automatic deregistration. It's my impression and I haven't really talked to anybody in RIPE about this, but it's my impression that deregistration is is a slightly awkward event when RIPE talks to an end user and says, well, you know, it's time to have those addresses back; probably it's easier just to say, look, you've requested it until a particular date and the software that we have just de?assigns it on that date and that is the end of it. I think this is having having no human intervention in this situation is probably a good thing.
There are a couple of open issues that I would like to get feedback from the floor on. The first thing is the time limits. There are two generic categories: There are events and there are longer term projects. The events are conferences like this, say cc C or the various hacker conferences or whatever, trade shows, that sort of thing. Seven days longer than the planned event is probably more than enough, a few days before and 24 hours afterwards. That is generally the time limit that these conferences have to work within, anyway, so that is probably not unreasonable. Projects no longer than six months. Research projects do go on longer than this and it may well be that if there is really genuinely no IP address space available that a research or somebody else who needs IP address space for longer than six months can make a successful appeal to the board of arbitration but the results ?? excuse me, the justification would want to be quite good.
The size of pool to be reserved by the RIPE NCC, as well: Filiz mentioned in her presentation that the current policies indicated that a /14 would be appropriate. I wonder about this, and my understanding is that the justification for a /14 was that the maximum concurrent historical assignments under the experimental category at any one time were a couple of /16s and one or two other address blocks all of which would have made a /15 plus a little bit more, which would have fitted in a /14. Now, we have a different situation up ahead of us because the new policy is much more broad than the current policy and, also, the ?? there is going to be huge constraint of address supply for regular assignments which is going to drive end users towards the temporary assignment category, if that is what they need, so probably a /14 is not going to be enough, but I would ask for people's opinions on this.
The other issue of course is that if a /14 is allocated by RIPE for an address pool, it's very difficult ?? later to try and expand this, because the address space simply may not be there. And finally, the maximum assignment size for end users, this wasn't put into the original proposal but it probably does need to be put in; this was flagged by the RIPE legal department, as Filiz mentioned. And as she said, it's simply to stop single users from hogging the entire pools. These are the issues. We have a few minutes to discuss so if we have any opinions from the floor, we would be glad to take them.
AUDIENCE SPEAKER: Hi, Rick ? Hi Nick. I have two questions: One is based on the practical experience we have as an ISP and LIR which is I don't see anything about keeping track and something like a log of the temporary assignment different times. What I am concerned about is that usually the police comes to us and wants to know, within 24 hours, who is the person that used that specific IP address at that time and nothing more, no more information to find it. If I have ?? and they usually base their request sometimes when they can do so, on the RIPE database, they look at the address and they see who is managing at the moment that IP address. So, I cannot answer to the police "I don't know who is the person using the IP address two years ago" and I have to keep track for I think 36 months of the usage, up to 36 months. So I need some kind of system to directly point to who was using that pool at a specific time.
The second thing is, we found some very unpleasant experience about people requesting, for example, AS 32 numbers and asking for change that to AS 16 number ?? AS numbers, and not seen using the S number or problem reclaiming unused IP addresses from our customers when we asked them to give them back. I am not so sure that is so easy to stop without the cooperation of the up streams. People advertising the temporary allocation, and at least in our experience, it's not so easy, but at least one /20 we had to black?hole the allocation to stop them advertising the space which really had to be used by us.
NICK HILLIARD: All right. Two issues there: First of all, law enforcement access to historical temporary assignments. It's my understanding that all assignments are logged at RIPE permanently so there is a permanent record, is that correct? Alex is just nodding correct. It's not in the public RIPE database and I suspect without any evidence to say this, that it would probably require a huge amount of software change on the RIPE Whois database in order to support multiple concurrent allocations of the same things so I don't know what RIPE NCC will do in terms of making this information available.
The second issue is what to do when a customer doesn't stop or an end user doesn't stop announcing the address space. And essentially, there is two answers to that, neither of which are particularly good. The RIPE NCC are not the police and all they can do is guarantee uniqueness within their registry, so if an end user doesn't stop announcing that address space, then it's no different to hijacking an address hijacking event of any other form. However, having said that, there are recommendations to use RPSL route and Route?6 objects so that you can perform meaningful Internet filtering from your down streams, and if you have a BCP 39 you can simply stop accepting traffic from them and you can do filtering. It's not I deal but, the RIPE NCC are not the police.
RANDY BUSH: Randy Bush. IIJ. Carston, were the the last at this mic? Just a little point on that. ARIN has available who was, in addition to Whois. I am actually a user of experimental address space and you will see a report of an experiment using it at 2 o'clock, except it's in conflict with this meeting, I believe. Six months ain't enough. One solution but I can see why you wish to be conservative, something about easy renewal and I use it in ?? I live in Tokyo so I use ARIN experimental address space, don't ask me why, and I am now in year two of this one experiment, using a /16. And in another experiment in year three and it's actually only a /24, it's the one just north of 192168 to watch for worms crawling north at a private address space. So either easy renewal or an application that states the length of time that one estimates is going to be needed and it's not only the length of the experiment itself, because it's ops stuff, sometimes set up, to set up, experiment on the live Internet
NICK HILLIARD: OK. That is a very good point and thank you very much it's very good to have live experience for this. You are correct to say that six months is a relatively short time period and we are aware that there is some situations where six months is simply not going to be enough. However, I think it's probably quite likely in the future that people like you are going to be in the minority.
RANDY BUSH: I am now
NICK HILLIARD: Not physically.
RANDY BUSH: Not mentally. The use of experimental space is very small.
NICK HILLIARD: The policy proposal as far as I remember, reckons if you want to apply for an extension you send in the justification again and I would guess that just means taking the original justification and saying we expect an X year or an X month time frame for this. If it goes to the arbitration panel, you know they are a pretty reasonable bunch.
RANDY BUSH: As I said, easy renewal is perfectly fine.
SANDER STEFFANN: Just for housekeeping, I want to close the microphone after the current speakers because we have two other proposals coming up.
AUDIENCE SPEAKER: Dave Myer. I want to reinforce what Randy said because I have a piece of experimental space from ARIN region and as Randy said setting up experiments, if you have to redo that every six months, that would be difficult, so the easy renewal aspect of it is very important to us, I think.
NICK HILLIARD: There is one alternative which isn't mentioned in the proposal and that is to keep some sort of nod towards the original academic intentions of experimental assignment and say for academic results where there is an obligation to publish the results, then maybe two years. I think we should consider this off line because this is a very relevant point.
AUDIENCE SPEAKER: This time wearing my arbiter's hat. (Wilfried) I think we need to do a little bit more thinking on two aspects: First of all, why the arbiters are going to be the first entity to talk to regarding sort of renewal of the assignment. I would rather have the IPRAs sort of do that and only in the situations where there is a conflict or maybe only in the situations where there is sort of conflicting applications and there is not enough address space available, then to do something which is reasonable. But I would rather see a mechanism written down which takes care of the majority of the cases on a procedural basis rather than sort of going to the review process and putting the arbiters sort of into an awkward position to, in the end, decide whether something gets extended and another party doesn't get sort of a chance to play, sort of that is more structural end, I would be happy to take that up with you off line, I think it needs more thinking. The other thing is more procedural regarding aspect of automatic deregistration. I think the time lines which are discussed right now are not really compatible with the time lines written downright now in the arbiters' procedures, so I think there is a good chance that we run into the situation that either an applicant has to apply for an extension at the very beginning of the tests already, just to be sure that there is an arbiter's decision within the validity time frame or we run into deregistration and have to roll back or undo, but that is clerical stuff.
NICK HILLIARD: OK. These are good points. Thank you very much.
SANDER STEFFANN: OK. I think we have seen two major things to talk about, one is the timing and the other is what Filiz raised, how to make sure that no single organisation claims the majority of the available space. So we will continue the discussion on that on the mailing list.
Then, now, we are having a new proposal about the final /8. There has been some going back and forth on this subject. Phillip Smith has sent in a proposal and another, they were a bit stuck so I asked Remco to shake things up a bit and it has come down to that and Philip have combined their efforts and I would like to ask Alan to give the presentation on what is now the single last /8 proposal.
ALAIN BIDRON: Thank you. I am from France Telecom. Good morning. So, a single proposal coming from Phillip Smith and myself, Phillip Smith was unable to be here this morning so I will make a presentation with visuals and the ? together. The objectives is to think for the future in the context of IPv4 depletion and that contacts we must ensure that no organisation during the coming transition to IPv6 lax IPv4 routable addresses, small part of IPv4 addresses that will be really needed in these specific time frame period.
So the solution try to find is applicable for new and existing LIRs. New LIRs are clearly one important ?? will benefit first of this proposal but it is not restricted to new entrants. It's also applicable to existing local registries that could qualify for IPv4 blocks.
So the context, we will skip this part because you presented it. We try to take into account the common objective and feedback from the community, doing the round of discussion of the two first proposals and trying to keep it simple, and something clear and simple as merged proposal and not to add some complexity merging these two.
So the description of the proposal: Implementation. That is an important point. The proposal is to implement this policy when the RIPE NCC has reduced the pool is reduced to the equivalent of a /8 or less of IPv4 address space. It is not when the RIPE NCC will receive the last /8 according to the global policy but when the overall pool of the RIPE NCC will be reduced to that size.
Criteria for receiving resources: To receive allocation proposal, you must satisfy the criteria in effect in the RIPE region at the time of application. So it's based on needs. The LIRs must already have received an IPv6 allocation from the RIPE NCC or an upstream LIR. And it's only applicable to LIRs and for IPv4 PA resources. PI are not included.
The size of the block that will be allocated is a uniform size of /22 and we look at registry only be able to receive one allocation at a time.
Where additional part in the proposal, it's unforeseen circumstances part. /16 is reserved for future uses. If this /16 remains unused at the time of the remaining /8 has been distributed it returns to the pool to be distributed.
The arguments for the proposal and main points that have to be highlighted:
First, each to allocate based on need and not to distribute all resources to any user in any registry independent of these needs. Will have to qualify according to current policy in place and the request will be reviewed by the RIPE NCC. A uniform size of /22, we think it is a size which is sufficient for the purpose of IPv6 transition. And we think it is consistent with the current number of LIRs in the RIPE region, /22 gives 16,000 possible allocations. Currently we have more than 6,000 in the RIPE region so we think it gives a possibility to serve existing LIRs and new LIRs for relatively long period of time. So many years. That is why we, think for the future in fact in the context of IPv4 depletion.
Questions and issues? We receive two main questions for explanation. The first one is why does this apply when the volume of IP addresses held by the RIPE NCC will be reduced to a /8 and not when the RIPE NCC will receive the last /8 from IANA? According to the global policy. We recognise that the second option is very easy to implement as a trigger for implementation is very simple, but why should we use this policy before it is really needed? Is the RIPE NCC pool is sufficient to serve existing LIRs or existing requests, current policy, why do to use this policy and the option we suggest is to only to implement it when really needed, it means that when the overall pool of the RIPE NCC is reduced to a /8.
PI are not mentioned. The second question, is it intentional? Yes. Because the objective is to use resources to help ISPs to transition to and ?? including PI means potentially introducing requests in the range of 10,000, 20,000 and reducing drastically the size of prefixes or reducing the time we could apply these proposals in a very strong way. And we think that including PI in that context could have some side effects in the context of IPv4 depletion, stimulate PI requests and maybe move from PA to PI to have addresses. So we suggest not to include PI, that is our proposal. That is it. I guess you have some questions?
AUDIENCE SPEAKER: Gert sent in a question about the PI space, do you mean that is once this goes to effect we don't do PI at all or just do PI like we do now with any other address space or how do you see that?
ALAIN BIDRON: That is a good question. I think the pool being depleted, PI could be served from other addresses but as the pool of RIPE NCC is reduced to /8 and that /8 is used for that purpose, my idea is that PI would not be able to assign PI any more in that context.
SANDER STEFFANN: OK. Thank you.
AUDIENCE SPEAKER: Thanks for that explanation, that was not completely clear to me and I think if it's not clear to me it might not be clear to the audience as well. So if we are at the last /8 (Gert) from that block there will be no PI, and I think we should write it a bit more explicitly into the proposal text. Just to make sure that there is no misunderstanding, otherwise I am fine with that. Thank you.
RANDY BUSH: IIJ. The problem I have with the first point of, that it's the level of the amount of address space, is that that can change back in the other direction; in other words, somebody can return a /8 after you have already started on this does that stop this or does it keep going, etc.? You laugh, but if the global policy for reallocation does say that things going to the IANA pool and the US military actually does what they say they are going to do is dump more /8s back and ARIN recovers more /8s you might be down to less than your last /8 and more /8s become available. That is the problem, I think, with this technique, instead of when triggered when you get your last theoretic /8.
ALAIN BIDRON: That is a good point. If we are keen enough to receive additional resources from IANA as a global policy is approved or in the reclamation process as originally, yes, we should adapt policies and find a solution to exactly ?? to use the solution all resources. We can add it to the resources used for this policy or we can put in place a process by which we allocate or assign addresses as today. But, I think
RANDY BUSH: So there will be both policies active at the same time? Or one for this pool and one for that pool? In other words, I just get a little confused. But I get easily confused so don't worry too much about it.
ALAIN BIDRON: Always in solving the APNIC region ?? how is it solved in the APNIC region
RANDY BUSH: By the way it was originally phrased which is when the last /8 is received from the IANA, it switches to this policy and never switches back, even if we get ten more /8s. That is what I think it is, I am not an authority. I haven't seen Sam here. Speak as a policy officer Filiz, you can do a substitute Sam.
FILIZ YILMAZ: I can never be in her shoes but this is my understanding as well from what APNIC reach consensus.
RANDY BUSH: I have to confess I am one of the co?chairs of the APNIC policy ??
FILIZ YILMAZ: He knows what he is talking about.
RANDY BUSH: This is why I don't know what I am talking about, I moderate the meetings and don't actually think.
SANDER STEFFANN: So no more questions? Then, I guess we continue this discussion on the mailing list.
ALAIN BIDRON: Thank you.
(Applause)
SANDER STEFFANN: And then I will switch to last topic of today, Dave Wilson is going to talk about his proposal to change to PDP in the RIPE region and I think he has some interesting news.
Dave Wilson: Hi Gert. I am Dave Wilson, I work for the Irish national research network, another hat of mine is member of the ASO address council and I presume I push this button. Talked about this as the Address Policy session last time around. This was as Filiz said, if you like, inspired by the bind one of global policies found itself in, although I won't fix it, I want to be clear about that, it isn't solving that problem. The idea that I had in mind here, the idea is this: We have a perfectly good policy development process where a policy proposal moves from one stage to another before finally being accepted. For global policies, the global policy in particular, this is one which effects the IANA and therefore must have consensus in all regions, such a policy does not take effect or cannot take effect until all five regions have agreed. If there is a change in the policy text in some region before ?? after it is adopted here, what happens? It gets kind of weird. So my suggestion was we add a state specific to global policies in the PDP, being accepted pending adoption in other regions, the intent being it is specific to global policies but allows it to go back to a previous state if a change is initiated in another region. The concept was, we can implement the policy if there are no further changes, we can speedily adopt it without having to worry what if there is a change later. There are a number of other approaches we could take this to, and something that has become clear to me, and this slide I used last time, I say that it's possible that wording changes might require the proposal to be resubmitted if we choose not to go this direction and approve this change then I think the cleanest way out at this moment, under the very state we have right now, if there is a change in policy is to resubmit the policy proposal. That is my opinion, by the way; I am not an authority on this in any way. But it seems to me to be a clean way out. The more I think about it and the more I have spoken to people, the cleaner that looks and the dirtier this other idea starts to look. When we are talking about five, six months ago, the idea was that it was a very small change that was very clean and had very predictable outcomes. It is, however, making the PDP more complex; it's making an exception and it's making a way back that doesn't exist for some policies and does for others. I, in retrospect, think it's getting harder and harder to see what the consequences might be of that. I don't think we can know them until we try it in anger, so the point it does it do something important enough to take that risk. It may be relatively small risk but is the extra speed worth it? Compared to the current state, and this is taken directly from the proposal text, there is a section of course for counter arguments. There is, if we decline to make any change, we assert then or one remaining way to make changes to policy text and process, is to resubmit the proposal as a new proposal. It's become clear to me, thinking about it, the global policy development process, if I can call it that, the system whereby each of the five regions go through their own PDP to approve a policy which gains consensus across all regions is very, very good at doing one specific thing, and that is where a policy has consensus in all five regions, it's very good at asserting that. This is trying to hack things to make it easier also to reach consensus on a policy that doesn't yet have consensus and in retrospect I am not sure this is the best way around that any more. There are other ways that probably deserve a lot more scrutiny than the single proposal. For example, a fast track system which can be applied a lot more generically and can get the right amount of scrutiny. So, right now, those words at the bottom there, any changes will get the full rigour of the policy development process if we don't accept this change to the PDP, is now my favourite, unless someone on the floor wants to convince me out of this my current thinking is let's leave things the way they are unless we can find a clean and more generic way to go through with something like this. And that is my current thinking. Thank you very much.
(Applause)
SANDER STEFFANN: Niall.
Niall O'Reilly: I am not sure whether this will convince you or not but it seems to me that with a small tweak here you can get the best of both worlds and that is I am not sure what you are going to call this state because I haven't read the document in detail but let's call it cache pending or such and if a document ?? if a proposal is to come out of that state, it seems to me that we could arrange the process so that it can come out in either of two ways: Either in the obviously natural going forward direction or if we impose some kind of checkpoint there to decide, no, enough has changed, this really needs to be back to square one, we could do that either and it seems to me it would be useful too far decision point there to assess the situation. At the moment when the circumstances have changed so that the proposal should come out of this suspended state and see if, in the meantime, it has somehow become sufficiently tainted that it needs to go back and get the full rigour, because then we still have the advantage that if it doesn't, we have a fast trackway forward. I don't know whether that is convincing but those are my top of the head reactions to what you are saying.
Dave Wilson. It certainly seems reasonable to me. My inI think ittive reaction now with any proposal like this is to be cautious. So, on the face of it, it looks good to me. I would like to give a close look because it took six months to realise for this very one but I appreciate that, thank you.
AUDIENCE SPEAKER: Wilifred, also a fellow address council guy. Initially, I was very much on your side with this proposal. Looking back in history in particular at this global policy that triggered all that stuff, my feeling right now also is that we shouldn't mess around too much with our regional policy, and we should make use of what is in place and what is agreed globally. I would have ?? well, at that point in time, I think that changing the procedures or amending the procedures would actually not have helped too much with the basic problem and the basic problem actually is that the regions do not agree on a common text. Whether we do that upfront or post facto whether we deal with that, I think that is a matter of the situation. What I agree with Niall here is that we might want to think about some sort of fast track procedure, particularly in the light of expecting, assuming, fearing that this particular global policy will not become a global policy in time. We might see the benefit of, throughout all the five regions, do something pretty quickly to provide IANA with something we want IANA to have in place at a particular point in time in the future.
SANDER STEFFANN: I see no more people standing at the microphones. So I think we are done for this session and and we will continue the discussion on the mailing list and I think we need to discuss how to proceed with this proposal. So, thank you all for coming this early this morning and we continue at 11:00 in this room. Thank you.
(Coffee break)