Connect with RIPE 60 : Facebook Twitter RSS linkedin

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

Eye the IPv6 session commenced on 4th of May, 2010, as follows... IPv6

The IPv6 Working Group session commenced as follows:

CHAIR: I still see a lot of people coming in but please go and sit down because officially we are already started. So a few more seconds. So, welcome to the IPv6 Working Group. We have a full agenda as usual. To make sure we don't have too much sticker shock, as the last meeting I was going to be the first person here on the podium, we have to introduce two more people today because between the two meetings we managed to appoint two new co?chairs. I want to introduce Marco and I want to introduce Shane who is not here because he had prior commitments and had no opportunity to be here. For people who actually ?? this is the welcome. This brings us to administrative matters. Because we have a full agenda, there will be a little bit, there is a chance that we will run out a little bit past 6:00, that means for people who had interested to go to the new newcomers reception, that they might arrive there a little late. Feel free to leave earlier, but on the other hand there people know it's not a problem if you arrive a little later than 6:00. For newcomers who want to go, don't worry if you run over time, you are still welcome after 6:00. Then, as far as the agenda, we have a talk about, from the RIPE NCC about Java scripts, HTP client measurements, we have a quick overview from Kurtis regarding IETF update, what is happening there. There is a CP overview from Marco himself, despite his function as contributor, and IPv6 global policy update from Philiz, a talk about mobile broadband and IPv6 from Jan. Even more updates from the RIPE NCC, then and very important topics, since we have a new charter we want to talk a little bit about how we are going to activate that and make sure we are going to reach progress on that new charter. And then finally, we have some standard topics that are always important. You know, we always want to give a little bit of a chance for people to stand up and give us some ?? to talk a little bit about events that are happening in the RIPE region that people are interested in or whatever, if there is anything would you like to mention here then during the meeting just come to me and we will make sure we give you a little bit of time at the end of the meeting.

Finally, we will also ?? have a little talk from Martin Leavy regarding some of the tools that they make available. We will figure out where we put it in the agenda but that will be a quick talk. That brings us to this slide, so here are the co?chairs, if you didn't know who they are, Marco and Shane and to make the whole show a bit livy, we decided Marco was going to lead this Working Group session because it's different, he will take a little bit different approach than I do, which is good; I mean, we like change here, so I give the mike to Marco and he will finish up the meeting from here.

MARCO: Welcome to the IPv6 Working Group, we have got two new co?chairs, one of which is me, the other is Shane, who unfortunately couldn't make it here ? well as you all know. Because we are now with three people we decided to put on a new mail dress so any fan mail can be sent to the address which end up in our three e?mail boxes.

We also got a new charter, this was brought up by Shane in the Lisbon meeting, we said down there with a few people to make first draft, it was sent to the mailing list a couple of times. Minor changes in textual things. I guess the main thing it has four focus points outreach, education sharing and fixing operational issues but basically if it has to do with IPv6 and IPv4, coexistence it's now in scope of this Working Group so we like anything to come up, if you have anything just mail us or mail to the Working Group and we will see what we can do.

Well, this is the administrative, the minutes of RIPE 59 has been cycled to the mailing list as well, there was one small correction after that, I didn't see any complaints so I guess we reached consensus. As David said, if there are any other urgent issues that come up, please talk to us during this session and we will try and fit you in. I guess that is about it from me. I will give the stage to Emile Aben from the NCC for his talk on measurements. Calling Emile.

EMILE ABEN: Hello, I am Emile Aben, work at the RIPE NCC and I am going to talk about IPv6 web clients and cache resolvers. So, why we did this. We want more insight into IPv6 deployment, and specifically we saw two IPv6 deployment numbers that, we saw large gap, if you look at the routing table, 6 percent of ASs announce one or more IPv6 prefixes, but if you look at web traffic, as measured by various people like Google and ourselves and our colleagues at APNIC, you see numbers between a quarter to two percent of web clients and we are wondering where is this difference, where do we lose these people in their migration to IPv6, so we thought of a way to measure IPv6 connectivity of end users combined with the infrastructure of the ISPs they are using and this is a measurement that is ?? that might be familiar with people that are, have seen the measurements by Google and recently by APNIC and Sander Steffan has done similar, a whole lot of people have been doing stuff like this but we added a twist, so what we measure is web clients and these clients go to a website that is participating in our experiment, they fetch a little piece of Java script and then they start to measure towards our measurement infrastructure and this is what lots of people already have been doing, which is measuring HTP to a dual stack to, a v6 only, and as a baseline of v4 and see what preference is, but the added twist that we have is we are also measuring the resolvers that these end users are using, so and the key idea here is that that resolver is part of the provider infrastructure and so if a client does a lookup, it asks the resolver and will ask our measurement infrastructure, and with a couple of tricks we can see if the resolver is able to do v4/v6 in preference. So, a couple of caveats: We measure to so there is a skew to these measurements. Not all clients use resolvers that their web streams provide. There is open resolvers and Google's 8888 is also not ? not your typical I think that we measure there and of course, because we are using Java script we can only measure web clients that run Java script. So our results:

First of all, so, low numbers here, which is hardly surprising since everybody knows that there is not that much IPv6 on the Internet, so on a scale from 0 to 100 percent, it's not much. That is the bad part of it. The good part is that you can see it without mag anyifying glasses, basically, so but if I use the mag anyifying glass you see this, so the red line here is the client IPv6 preference, which is given a client has a choice between v4 and v6 it will use v6, which currently for the measurements we are doing is between one and two percent which corresponds nicely with numbers that Geoff Houston has APNIC has recently gathered. And you see between January and now, we see a slight increase, is it flat line or just slightly increasing, it's a little hard to tell. Next thing that we measure is the client IPv6 capability, which is given an object that is only available over v6 is the client capable of fetching that? And you see that number is around 4 percent, and there is actually some nice spikes here, so IPv6 has a heart beat, and actually these spikes are all in the weekend, strangely enough, so if we want to spur IPv6 promotion we should take longer weekends, I think. So, and the blue line is the IPv6 capability of the resolver, so the resolver is using or the client is using a resolver that is capable of using an IPv6 transport. So in that number is at about 5 percent and also slightly increasing. And so if we filter out all the auto tunnel stuff or if we don't count ?? count auto tunnel as not real IPv6, so I call it native but native is overloaded, so this is the guess so no Teredo and no 6 to 4 and you see the glean line that was at 4 percent before, drops to one percent, which is only logical because the difference was caused by the auto tunneling, and in this graph, you can see the difference between the clients and some of the infrastructure they are using so there is quite a gap. 1 percent and 4 to 5 percent. Next thing we did was look at this ?? at country levels, because governments like this and it's also good to see if specific countries are progressing more than other countries, and I just cherry picked a couple here; in the end of the slide pack which I am not going to show because of time issues, I got all the countries listed, but I wanted to pick up France because it is known that they have a free .FR and what we see is they are above average; the green line here is the client average, the blue line is the resolver average over all of the measurements that we had in these three months and you can see France is at 3 percent of clients, 6 percent of resolvers, Czech Republic our hosts doing very well as you can see. I wanted to point out that IPv6 is also happening in the Middle East, Saudi Arabia is nicely on average for client IPv6 capabilities, not a little under average for the resolvers, and our champion here is Slovenia with 8 percent of IPv6 capability and Jan is going to talk later hopefully about how they did this. And also the resolvers are quite high and Estonia, the resolvers are through the roof and almost 40 percent and this is about 2 percent. So and I have to mention that the geolocation here is based on the client IPv4 address so this is an added thing that you might be able to do with measurements like this is mapping IPv4 to IPv6 to geolocation information.

If we drill down a little further into the, to the city level, again, Slovenia shows up at 11 percent so I just wanted to show here that there are pockets of IPv6 in the current Internet already. And also, this last column shows some of the diversity that you can see in specific cities; these are the number of ASs that we measured in the specific city, the first one is v6 ASs and the second one is v4 ASs, so as you can see, for instance, Paris has over ten percent, 14 over ?? under 129, so over 10 percent of the ASes that we see there is v6 capable so it's not only Fee .FR in Paris, apparently. And the other thing we did was look at autonomous level. So if we go back to this number, this 6 percent of ASes that we see in the ?? in RIS, that announce an IPv6 prefix, if you look at the ?? for this, we mapped the our address that is we saw to the AS using data from RIS, and as you can see, in these three months, we saw about ?? over one?third of all ASes that are currently in the routing system, and of the ones that we see, 5 percent have an AS, we measured at least one resolver that has v6 capability, so 5 percent and ?? roughly 4 percent of the web clients where we ?? ASes with web clients where we detected v6 capability of the web client. So this 3.8 is quite a lot higher than the 1 percent that you see if you don't aggregate on this AS level and so that indicates that IPv6 apparently is getting to some of the clients but not all, far from all. So this indicates populations at various stages of IPv6 deployment but I was actually quite happy, surprised by how high these numbers are. So what is next for this? We like to keep this running and of course, next year ?? this is a nice baseline. Next year should be interesting and if it's not interesting then the year after certainly will be interesting. And we are fully aware that we ?? that the measurements are skewed currently because we measure so we measure best case so we would like to measure Joe Average, my mother, so what we like to do is have this piece of Java script installed on lots of other web servers and especially web serves that serve Joe Average Internet user so we can see what these numbers are for the average Internet, of course. So you can participate in this. If you are interested, if you know a site, please ask me. If you have feedback on these measurements, if you don't want to put it publically, ask me, and if you are interested in the technical details I am going to do a presentation at the TT Working Group, soon to be measurement Working Group, on Thursday, sometime Thursday. Morning. That is it. Questions?


MARCO HOGEWONING: Nobody? No. OK. Thank you, pretty clear story there. Looking forward to the measurement. Next up, Kurtis basically showing us what is happening in IETF recently on v6.

KURTIS LINDQVIST: So when I was going to do this presentation, I was asked to speed up so, those of you who know me ?? yes, so my name is Kurtis Lindqvist, I am one of the co?chairs, Fred baker of Cisco. Shane kind of, and Marco kind of asked me to do a very short introduction to what happened at V6ops which is one of the Working Group in IETF, there is no standard documents coming out V6ops but do produce documentation on some of the operational, best practices if you want. Because I only had 15 minutes and then I kind of promise I had David to do this in 10 I decided not to take all the ideas, these are my own personal favourites that is currently being discussed so if your graph is not part of this set, then, you know, it's not because I don't like it; I just pick what I thought was a bit more hot topics right now.

By the way we of course encourage everyone top join V6ops and provide operational experiences and when you see the slides here and some of the topics of interest maybe that will trigger your interest to come and say things as well.

This is a document that is actually ?? the first document is draft v6 in IXP. This is sent to IESG for evaluation and which means there is no more feedback from the community. I thought it still night be nice to mention it, though. It describes IPv6 usage in exchange points and goes through some of the topics there. It describes all the impact on switch fabrics, it describes all the various addressing plans for v6 addresses to be used in exchange points. There is quite a few different models and goes through some of them. It also gives some things to think about regarding multicast at the exchange point and talks about that if do you route servers at the exchange point you should also think about having these enabled for v6 and some of the concerns there.

This is probably one of the more hotter topics and has been for a while, there was some jokes being made earlier maybe I just should skip this slide. This document has been around for a long time in various shapes and forms and names and it tries to ?? it basically comes from the idea that the CPE vendors wanted to have some recommendations of what filtering capabilities, notice the word capabilities, we should have in CPE boxes when they ship to customers. And this document has had various strong wording, various shapes and forms throughout the years by now, and the current status is that it's not a standard; it's a recommendation of the capabilities to go into a CPE. You can make it more fussy than that, but it's ?? that is what it is. And one of the controversies surrounding this document was down to the fact that there is some ?? actually, almost a 50/50 split in the Working Group between whether we should say something that some of these filterings should be turned on by default when boxes are being shipped, and just to give you some background on this: There is two sets of equipment, if you want, about this being shipped: One are the ones that are either managed by the ISPs or shipped by the ISPs and therefore the default configuration is determined by the ISPs and the other ones is CPEs owned by the end users and where the defaults are set by the vendors. There has been some argument that the default should be recommended here so that we might have some sort of similarity between the filtering as it's being shipped by the vendors. And here, so there is two camps here: One say that this is good because end users, today, and operating systems today so are so vulnerable and they have come to expect and trust level of protection from NAT and if we remove this, they will be more vulnerable and the other camp is basically arguing that the entire idea behind why we wanted to restore v6 end to end connectivity and not have middle boxes blocking this end?to?end connectivity. I am going to leave it there. It's a very interesting discussion, I will encourage you all to think about this and don't send too many e?mails because I have to read them.

The next document that might be of interest is the idea of v6 CPE router ?? one of the authors are here somewhere. So, it basically describes basic feature set of v6 capable customer edge router and the functionality and also goes through the various provisions options for one of the VE routers and describes the ?? I would say minimum requirements, if you want minimum features that in order to have v6 CE routers, provisioning for the land addresses how you provision workstations if you want behind this CE routers but also how you do provisions from the provider's side from the VAN port and what you have to support and not support and the various addressing models.

The next document I thought I will talk about and I am soon using up all my minutes here, is is a draft IETF v6ops ISP scenarios. This is a survey done by Brian carpenter at University of Auckland and a gentleman whose name now escapes me and they did a survey among 31 ISPs of various sizing of their deployment plans, this is either planned where there is concrete plans for deployment or already have made deployments. There was presented all the statistics and questions were presented at the last IETF and also in the draft document the questions and some more summaries of the answers. The 31 answers were 65 percent were interesting enough from Europe, they are ?? I think six of the Respondents had on the order of several million customers and 42 ?? and going down to smaller ISPs, and 42 percent of the responses already were offering v6 services. This document doesn't make any conclusions or recommendations; it simply presents the choices that people have made on a statistical basis. So that might be interesting to follow and understand a bit more, if you want to look at it.

Last is also the author here somewhere, there in the back. Draft v6 transition guidelines which if I understood came out a presentation that Fred Baker made in the mobile v6 workshop, whatever it was called, and describes various scenarios of deploying IPv6. I put the word ISP in there, I understand the document is fairly generic, given the audience in here I sort of put it in here, it has some of the recommendations for operators. And gives recommendations on what considerations you should make when you are looking at various of the deployment options or models and also have some examples and common deployment models and what people decided to do and what choices were made and the RFC 429 deployment scenarios and this document is a bit more generic than that. I think one of the things that is interesting in this document is also takes into account some of the things we didn't think about then, for example the translation v4/v6, there was a request to put more of that text in there, but Yari is nodding. So you will have some of that that wasn't available when 1429 was written.

Last draft mobile networks, if you are a mobile operator looking at doing v6 or works for an operator that does mobile networks, this document gives an overview of the various deployment considerations, sorry in 4 networks, doesn't apply standards, highlight some of the various cases and scenarios. Most of the 3G networks deployed with NAT v4 and how do you handle this when you get IPv6 into the mobile networks? Dual stack, etc.. or even v6 only networks. And it also has some of the relationship between various transition, well deployment scenarios and the components of the 3 BGP architecture. That was all my slides. If you have questions on those two particular drafts, I recommend you talk to them instead of me. But I will be happy to answer any questions, too. Any comments, remarks? OK. Thank you.


MARCO HOGEWONING: Thanks, Kurtis. I guess it's my time to introduce myself. I am going to talk about IPv6 CPE again and this was basically based on feedback I got from Lisbon, that people found it interesting. Rest assured a lot of this testing is still done in the network of my employer and employer's time but I am also greatly relying on feedback from the community so this is not at all an ?? presentation. This disclaimer: I basically went all out and created a little survey, it was basically 15 questions which I sent to all contacts I could find on vendors which I know are busy building IPv6 home gateways. Some of them came back, some didn't. Some of them came back with the fact that I explicitly asked them to sign up for publication and, well some gave me a lot of feedback but the actual signature for publication got lost in the legal swamp so unfortunately it's not there. The other big part of this is community feedback, people testing within the axis for all network but also other networks gave me reports on what they see and it's been included here. Here is the final list I will pay attention to. AVM Draytek, Cisco was one of these people who are happy to give me all the information up to where I needed the signature. Juniper signed off techicolour. I will be rushing through this a bit. There is huge amount of information and I actually talk to the NCC and will probably publish more of this on the IPv6, somewhere on the website either right now or right dot labs, to the mailing list when done. Vendor info, basically two models 7270 and 7070 and ADSL only, the latest box is 73, that is current software beta or labora, they expect general deployment somewhere they say summer this year. So expect it third quarter. As Kurtis mentioned, included in the question what do you do with the firewall. IVM brings firewall default to all while the rest is clear, one of the other points is that they really have rich tunnel set up, 6 to 4, 6 axis tunneling, pretty ever flavour of tunneling you can do.

Experiences so far: Well, it's still a bad idea, occasional mishaps that got fixed. The firewall, there is some discussion going on. Vendor claims that all are fixed, in the latest release do I see reports from people who claim the user interface sometimes forget about all the settings. I mean that is stuff that goes on if you test. If you experience those issues, recommendation from the vendor is to use the latest version.

AUDIENCE SPEAKER: Which still has the bug.

MARCO HOGEWONING: Thank you for that feedback. One of the other things, consumers, no manual address configuration and one of the main points there is no routing towards other hosts so you get the/48 from your provider or the 56, it use one/64 for its internals and another/64 for the LAN and the other ones go to waste. That is pretty much. ICMP is let through if it's associated with an existing session. It's pretty good box, especially if you are ?? well, it's suitable for ?? it's nothing fancy but it's nice. Draytek, back in October in Lisbon I showed this slide, they run off with the specs and never heard from him. A couple of weeks box showed up and contained a working modem which is pretty OK. The figure of 2130 series, it doesn't do VDSL, ADSL, only PPoE or bridge, tunnel support, again the firewall defaults to all. Experiences: Like I said no ADSL interface, they built a tiny box about the size of a pack of cigarettes. DSL goes in on one site and ethernet comes out the other, it translate over ATM into so you can hook that up to the 2130 again, also see a lot of people who use this box and hook a Lynnage box or BSD box behind it and make it capable that way. Retail for this, total set?up would be 250, the 120 itself retails for about €50 so that is why it's should popular thing, it does all the routing systems you are familiar with if you run Linux. Cisco, 87880 is pretty complete. Feedback I was pointed out to the fact from after 15 feature parity in the base licence so if you run the latest software you don't require additional licensing. Back in Lisbon ??

AUDIENCE SPEAKER: And the latest hardware.

MARCO HOGEWONING: Thank you. There was some discussion on the wireless, it did work if you route it but it didn't bridge between wire and wireless part. I know in the latest versions it has been fixed. 87 is unknown to me. Packet filters default to off, you have Cisco and the other main thing it requires some knowledge; it's not really suitable for mass deployment because you do have to know what you are doing. Juniper, really honest they were the first ones to point out they are not really building a consumer box but I did include them here because I know a lot of people are using them because they get them from the vendors. It's pretty nice box, obviously the defaults to on, PPoA is on the road map for the Aserix one as far as I know, everything except prefix delegation so that might be an issue there because if your provider relies on that then you are screwed. If they manage to statically route stuff to you, it works and obviously it also speaks a lot of protocol so you can do BGP OSPF and whatever makes you happy. I know one happy user is using one of these. Tech colour: Well experience, it's fairly limited. I have touched one, I have played around with it, then I took it back because it was still better. It still road mapped, I know the work is continuing, I get feedback or questions asked. Unfortunately there is no release date known yet but it should be out there I guess if you ask for it. Apple: This is one that showed up on the list, working together with Comcast, the latest release for the airport extreme and the time capsule includes a router option. I had a customer test it. Well if you said it to PPoE unfortunately it doesn't start to do IPv6 negotiation, so basically, that is where it all failed. We now set up static routing so waiting for the customer to get back, whether if we configure everything statically, it is able to pass packets but on the normal, well the normal the stuff we expect it to do, it's not doing anything.

Linksys doesn't do IPv6 you about they are quite popular, it ?? reports so far it seems to work and basically any environment, the HP 6 also working, it's a nice set?up if you know what you are doing, mass deployment I can't imagine a provider buying 2000 and then flashing them to open ?? it does 6 to 4 I guess you are going to point out by default.

AUDIENCE SPEAKER: Yes. Hi, Remco from EQUINIX, the 610 does 6 to 4 in its undocumented, if you want to know thousand works go look up the presentation, I did last year at NANOG, there is a way to switch it off if you want to run native v6.

MARCO HOGEWONING: Point for discussion. It was pointed out to me by the vendor it says it's ready for advice at that on the box and the 6 to 4 implementation was actually required by Microsoft.


MARCO HOGEWONING: There is another one that, well, not really known, though claim to have IPv6 support for years now but nobody has seen it, they have come back with information that all models leased in this year and onwards should be IPv6 as better, vendor responses basically include everything that was there so they claim to do PPoA, PPoE, bridge, firewall defaults, well Zyxel experience, I have asked for one, haven't got one yet, apparently it's really new and they are running a number of prototypes, they assured me with the next shipment I will get a loner so I can get my hands on it. If anybody else has experience please send feedback to me.

Others here: DLink that was mentioned on the American website as supporting IPv6, I have tried to contact them, I have got no response, not a single bite so. If anybody is DLink is listening please respond to me for anybody else knows what the actual status is of that box.

Iskratel they are building a box that does IPv6, I offered them to fit in the survey, they politely refused because they say it isn't ready yet and another interesting one is I heard some rumours about Sagem being busy, I haven't talked to them, if anybody knows what they are up to I would love to hear about it, which brings me to the end of my presentation, actually it's question time. I have got some feedback that people appreciate this list, it's a lot of work, should I continue doing this? We have discussed this with the chairs to make this a reoccurring event and every right update ?? the other big point is if so, is there a volunteer, especially to help out testing. Most of this testing I have been doing is against the axis for all network, I would like another operator to be able to claim it works against their set?up, we have some specifics Juniper ?? maybe if somebody is running equipment from different vendors I would love to hear. Especially also help together publish information and I guess that part is now being covered by the NCC already and your feedback is more than welcome. We try to make this list build upon the experiences from users and not only from what vendors tell us. We really want you out there to tell us and tell me and tell the others what parts need special attention. So I see questions?

AUDIENCE SPEAKER: More one that you didn't have on the list. Have you had a look at Microtech?

MARCO HOGEWONING: It was mentioned to me ??

AUDIENCE SPEAKER: Has it IPv6 in their since version 3 and now the 5.0 beat at thats out which support PPoE and X works quite well. Our complete network is based on Microtech so we started testing it the PPoE stuff already.

MARCO HOGEWONING: I would love to hear more, send me some info and we will stay in touch. As far as I am toll they don't do ??

AUDIENCE SPEAKER: No you would need a converter there. It's PPoE at the moment.

MARCO HOGEWONING: Why they initially fell off the scale. This is the involvement of a presentation about ADSL and IPv6 and I am slowly moving into parts, but it takes time I guess.

CHAIR: Please identify yourself first at the mike.

AUDIENCE SPEAKER: Private Internet citizen, have a show of hands who use as switch CPE with IPv6?

MARCO HOGEWONING: We could do from the list we have here. AVM? For. Cisco that is the other big bunch. The ?? there is one in the back. Linx is with the 6 to 4. There are pretty much ?? who has native IPv6 at home at the moment. Wow, we are reaching that point. Tunnels? I guess that should be the other half of the ?? still a lot of people who don't do anything, but that is cool. You have your answer there Wolfgang.

Daniel: RIPE NCC. I think you already discussed that with the colleagues but we are certainly interested in publishing this and not just and updating it and not just at every RIPE meeting but continuously.

MARCO HOGEWONING: That was my point exactly to keep it out there at every RIPE meeting, just list the highlights of what has been changed and out there.

Daniel: Stay tuned. I am quite sure we will do this. The other suggestion I have and that is maybe not for Marco to do here but maybe for someone else, is to do a similar survey for these prepackaged firewall deals that are out there, because OK, it's not really mainstream consumer equipment but I know a few people who use them who are not really computer literate, have a box they use as a firewall, and some of them found out that there is no v6 in there and I said yes it's in there if you can do IPFW or whatever, but that is actually the next little survey that I would like to see. I am not volunteering but if someone else was an expert that would be useful to have.

MARCO HOGEWONING: I guess it puts some requirements on your modem to be able to route subnets through fewer firewall.

DANIEL: I have a /48 at home and I can't use it now and I am happy with static routes for the time being.

MARCO HOGEWONING: Thank you. IfF there is any ?? if there are any volunteers who want to help out testing or get more feedback, contact me or ?? yes, or raise your voice and we will see what we can do. We are doing this to help you, I mean if nobody else is stepping up, then we will keep doing this but some more feedback would be appreciated and not only rely on what we get thrown at vendors. So if there is any more questions, then I would like to hand over to Filiz to show us what has been happening in the rest of the world regarding IPv6 policy


CHAIR: Normally we don't do and we still don't do policy in the IPv6 Working Group that, all happens in the Address Policy Working Group but occasionally we want to make sure that people who might not always go to Address Policy Working Groups are well up?to?date on what is going on in other areas so don't see this as an attempt to do policy in the IPv6 Working Group. This is just an opportunity to reach a larger audience regarding IPv6 policy issues.

Of course you can ask questions, even a policy question. We might shout you down at some point.

Filiz: RIPE NCC policy development manager. I just want to adopt a few things on what David has said. The reason I am here now today is because we ?? the chairs thought it may be an idea to show examples of what other communities are busy with within the IPv6 policy framework because IP version 6 is getting more and more operational nowadays. IP version 6 policy was as a document over a decade ago and that was before your operational experience and now people are deploying it more and more, they manage to find those little holes and maybe not so sufficient parts of the policy so things are getting acceleration in the policy framework, as well. And the idea was maybe those news, what are happening in the other regions, can be given as an update here and maybe we would trigger some more brain cells here within RIPE region as well. So that is the motivation.

So, yes, I am going to it be reviewing few things from ARIN, APNIC, RIPE and LACNIC because they are the regions that are talking IPv6 policy or having discussions at the moment. The only reason AfriNIC is not listed there is because they just don't have a particular policy at the moment.

So, what is ARIN doing? ARIN is busy ?? is quite a busy region in terms of IPv6 at the moment. They are looking at different criteria for IPv6. One of them is the PI, and they have several proposals about this, some of them are purely trying to clean up and tidy up their IPv6 policy document, again another activity due to the fact that IPv6 policy document is quite old and is catching up now with the operational experience, but also, they are changing actual policies in regards to that. One of them is the criteria for PI and they are to change it, there was a proposal and it was saying that either you have to have an IPv4 PI so ? of your network or your IPv6 within the IPv4 or you need to be seriously trying to be IPv6 multihomed or you have a plan of the deployment and you are able to justify this document, this ?? for the next year, second year and the fifth year and you have to justify why you want IPv6 PI and that justification is important. They are asking for why you wouldn't be able to get IPv6 space from an upstream provider and if you would qualify with this criteria, the next step would be that would you receive up to /48 for each site. Now, this is where the discussions got a bit exponentially heavier because they started talking about what a site is. I am not so sure if you people here in this room are also wondering what a site here. The IPv6 policy document talks about, and I believe people have something in their minds about this IPv6 site but I am wondering really if it really matches in those different brains and are we still talking about the same thing? Because one of the concerns was is it a location? Is it an organisation? And especially when you are talking about within the realm of the end user networks, then you are talking about PI, is it part of an end user's network, so what are we really talking about here when we say "site" and I am hearing this from our community as well, so that might be a point of discussion, maybe. What ARIN community decided about this particular proposal was that, yes, it is interesting and we need to be looking at this more so there will be more discussions about this. They are not ?? they just not made a decision about this yet.

Then, they also talked about criteria for allocations, so for the LAR space. Similar criteria for the first two, and then the third one is planning to make 50 assignments to others in five years. Some of you may have a déjà vu about this, number of customers because we also had this situation and it was changed a while ago, we used to have 200 customers criteria to plan making, planning to make 200 customer assignments from your location only if you make that plan and demonstrate that plan you would receive an IPv6 allocation. Now it is changed, for a while now actually. In RIPE we don't have this criteria any more, being LAR and having a plan is enough. However in ARIN this is still the case. So they basically, the proposal was to change this number, arbitrary number to 50. Then again, obviously, the conversation was around is 50 a good enough number, is it a good number anyway? And there were suggestions maybe they should get rid of it altogether but they decided 50 is a good start. So this is a last call now and this will be the criteria in ARIN land to receive IPv6 allocation if there are no objections after the last call. ARIN again, they had a proposal called simplified IPv6 policy. However that didn't prove to be that simple. What was interesting about this proposal was that, you know, at the moment, the justification is provided by the LAR for a request of IPv6, and the size of ?? the minimum size of the allocation is already /32, this is clear enough in the policy. However, policy doesn't give too much ?? or detailed information or guideline to the RAR about the size of bigger allocations, so if you can demonstrate more, you will get more, but how big you will get, you don't know. And in ARIN, there was this proposal which was trying to craft off those ?? those created or pre defined sizes for certain size of organisation, so they were called small to XX large and according to the justification you will provide, basically the Hostcount and the number of /48s you are going to make assignments from the allocation, you would be fit in one of these, say, sizes. That triggered discussions about classful, this was found, are we going back 50 years and going to be discussing this within the IPv6 world as well? Is this really good for the routing tables? And such. And yes, it was abandoned. So it didn't receive much support.

LACNIC: LACNIC is the RIR serving in the South America, and they have already an IPv6 policy for the infrastructure, critical infrastructure, and when I said that I already get goose bumps because that is very special world, we all know that, and when it's usually you see people having those rolling eyes and running to the mike in our region. In LACNIC they consider certain type of organisations or services under the definition of "critical infrastructure" and that is IXPs and NAPs, National Access Points, and they refined the criteria for them, they just need to provide LACNIC about their capacity, information about their capacity, which means having at least three members and justifying the need and they will receive minimum /48 or maximum /32. Now, more on that, the proposal went further and decided to recognise the need of criticalness on end user networks, so that part is maybe more interesting for you, and it is basically trying to define, well, some people may think that their network is extra critical for themselves and this part will not be routed differently over the Internet and this need should be recognised and they can be receiving minimum /48 or minimum /40. This is currently under discussion and they will talk about this in two weeks time when they have their meeting ?? annual policy meeting.

Then, LACNIC and APNIC: They are also talking about removing routing requirements, which we have done here, so, routing ?? in the previous IPv6 policy, announcing the single block, that the LAR will receive; was a requirement. In ours it is removed now, we don't have this requirement and it was decided that this issue which is also important for routing, obviously, can be tackled outside the policy requirements but on an operational basis in Routing Working Group, so now at the moment there is a proposal in the Routing Working Group which is trying to make some recommendations about the best common practices for the announcement but the policy is not referring to this requirement any more. LACNIC is trying to have the same. APNIC also reached consensus for the same thing so there is a moment going on there.

As a result, like I mentioned, yes, policy is trying to catch up with the operational experience and actually it is really nice to see because this is real evidence that something is going on in IPv6, people are finding those problematic areas and trying to solve them within the policy as well. Two main subjects I see is the site definition getting into more, you know, more details now, and trying to clear up those fuzziness around the terms that we all use, we all refer, and the other one is the routing practice, which still is needing work, and ?? but the main thing seems to be simplified criteria. People are trying to make sure that the incentives are increased so that the deployment is going to be accelerated even more. And that is the end.

But before I leave the mic, I just want to say, if you ?? if I haven't ?? if you haven't been caught up by me, and if I haven't attacked you already about this one, there are some ?? this is an article about IPv4 and IPv6 PI policy differences, and there is more at the tables in the lobby. There will be an agenda point ?? there is an agenda point in the Address Policy Working Group tomorrow, this will be discussed so it will be great if you can have a look. Thank you.


MARCO HOGEWONING: Before you actually leave, are there any questions for Filiz? No. Nobody is standing up. Thank you. See you back in Address Policy, I guess.
Next one up is ?? well, it's actually nice one, I came across this two weeks ago and talked him into giving a presentation and I really respect the way that they ?? Jan Zorz from Slovenia who is going to talk about what they are doing with IPv6 and it actually works.

Jan Zorz: My name is Jan. And I come from the country that Emile said we are the champ in IPv6. I think we won a RIPEness contest or something. Actually, yeah, Marco saw the working mobile phone where I showed him how to access the site from the wood, over the mobile phone, so and Emile asked me how we are doing it so well in IPv6, so I will present both of them.

First, I come from Go6, this is the initiative and not for profit organisation. We are IPv6 pushers in Slovenia. So, few words about Go6: We have the time issue so I will run forward as fast as possible, I promise.

Then I will show to you how Tusmobil and Mobitel deployed IPv6 in their cellular network, you can do traffic and all kind of stuff over v6 native. So, about Go6, IPv6 initiative in Slovenia, we are not for profit. We formed a strategic partner with Arnes and LTFE laboratory for telecommunications on faculty of electro tech nicks. We are an open platform. We are based on membership, have all major ISPs, content providers and integrators and industry that are our members. So we talk to everybody.

We are financially supported by members of Go6 platform, all of these guys, and Go6 is steered by Go6 expert council, and we have representatives of government, he is not here today; then telecom post regulator is ?? he is sitting right there, academic research network is Mattiesh is sitting right there, we have faculty laboratory, down there; and we have Slovenian IPv6 Working Group chair as a member of expert council. This is steering Go6 and we are deciding what to do with IPv6 in Slovenia.

So, our activities: We formed the IPv6 Working Group. It's not open; it's invite only. We might change that over time. We organise Slovenia IPv6 sum its, the third is coming next week and Daniel is going to keynote it. Previous summit was keynoted by Maarten J Levy, and I thank you for that, it was brilliant. And also, in November, we invited Patrik Falstrom that kindly accepted the invitation.

What we do, consulting to members of the platform form, 6 deploy workshops that will be held in Wednesday, I think. We operate and host 6 deploy lab. We are doing some stuff with RFC draft, we formed the Go6 academy, aggregation of IPv6 education from the industry and Go6 lab and we do lots of testing.

So, what is the mission: We make people talk to each other about v6. We are small country, we have 2 million people, we know each other, literally speaking. We push IPv6 deployment and you can see by RIPEness graphs that we are doing our job very well. We aggregate knowledge and make it available. This is the important one: We connect government regulators, ISPs and industry between each other, and this is the most important part, 10 percent of IPv6 is technical, 90 percent is politics. We make competition companies talk to each other. As you will see, the two mobile networks, they are competitors but they got IPv6 deployed together. So we became an IPv6 pivot point in Slovenia and apparently it works. You can see in these graphs here we are pushing up the limits. From this point on I am a proxy presenter, I will briefly show you these two mobile networks that deployed v6. We helped them, we tested it, we acted as a catalyst for both operators, to get together and start working. And this is one of our success stories. So, I am not able to show it to you here because the Czech mobile operators are not able to do v6, apparently, and so I will just show you the nice pictures. These are not pictures from the lab; I did those pictures at home or in the lab, so this is the real stuff. Tusmobil, they have this, IPv6 allocation, and when you ?? with this phone, when you go to, you receive this picture, IPv6 address. This is from Arnes and it is something for measuring stuff and you can also see that you reached it over IPv6. What is very, very interesting is that you see, this are two different IP addresses and I have no clue why, because these are two different bookmarks in the browser and apparently, for the different book mark, it does different PDP context but we need to find that out. As you see, from the same phone, if you do plug in, you set up the e?mail on the phone and try to connect and fetch your e?mail, that is the IP of my server and as you see, for iMAP there is different IP than for SMTP for sending e?mail. We still don't know why but we will find out.

OK. This is Tusmobil network, this is their picture. As Nokia currently is currently not supporting dual stacks APNs they did the trick, they do PPP APNs redirect to second back to the different APNs, but it's working, and it's working very well. We are in time rush, I think you can come to me later and we will discuss the pictures.

So, versions of the software used is this list. So, this is being done, we are not just talking the talk but walking the walk, apparently.

And this is from Mobitel, this is national mobile provider. This is different phone. They decided to do different techniques, so if we access site from the Mobitel, but these guys user I can son for access network and interesting is that different IP addresses here, so I suspect Nokia phone to do the bad stuff. This is their picture, it's kind of complicated stuff and they do IPv6 only PDP context, don't do dual stack or hacks but it's working. Each terminal gets /64. Some problems were with IPv6 PPP and DNS, you have manually to configure it and there is very limited terminal support. So these are versions of software used. No magic. What is still missing? Both companies reported the same issues. No I support ?? no wide support currently only Nokia and I found out that Android in the last version supports on a wireless. I don't know about the ? port yet. There is no content based charging, it's not existing so this is a big draw back for the mobile operators. And there is limited or no support for dual stack in terminals. If they have IPv6, they can't do the dual stack. They do v6 PDP and v4 PDP so this is kind of small show stopper.

So questions regarding ?? questions regarding IPv6 in Tusmobil or Mobitel: here are the e?mails so they will be happy to respond to your questions and the details. And for everything else, it's me. Thank you.


MARCO HOGEWONING: Thank you. Well, actually, I have a question: You mentioned it's not dual stack, it's basically dual session single stack but does that work or do you rely on the net 64 to get you ?? capable of IPv4 and IPv6 session or do you have to choose?

SPEAKER: Yes, but it closes down the PDP and makes another one. At the same time, on the same phone, you can't have a look into two different websites on v4 and v6.

MARCO HOGEWONING: Thank you. Any other questions for Jan about this set?up? I think it's wonderful stuff.


SPEAKER: I forget to mention roaming. Tusmobil, very small and they don't have the coverage in all Slovenia, so if you run into the hills, you are ?? you are just end up connecting to Mobitel in roaming mode and IPv6 is still working, so we have fully operational national IPv6 roaming going on in Slovenia. This was missing. Thank you, Maarten.

MARCO HOGEWONING: Thanks again Jan. Next up is a small change to the agenda, and Martin Levy would like to present something on what he has been doing with hurricane electric and we decided to add him here and have Chris head up ?? so Maarten.

Martin Levy: Hey. I should be able to break everything by the time I am started. I am going to ?? sorry, Martin Levy, Hurricane Electric, occasionally do v6. So, yes, I am going to do something live. So, 10678 you know about this because we have told a few people but if you don't know about it, that is the reason why I have got three minutes here. We put up a site about maybe six months ago now under the URL that does pretty much daily processing of the RIRs data from RIPE and the Oregon route views data. This is not coming off of our routing tables; it's coming off from the collections that are done and we are very thankful for that. It gives us the ability to do searches on a bunch of different network elements, ASes, IPs, etc., etc. And I just want to run through a couple here because we added some features on Friday, I didn't know if it was going to be here, and we don't have very much space on the screen here, but this will work; so, we added ?? there is V ?? everything here is v4 and v6 related, and we get information on a particular network so we looked at 333, which is used by RIPE labs and RIPE. I decided to stop that. And we get the ability to show how routes are seen throughout the rest of the network. V4 at the top, v6 at the bottom. But because I only have a few minutes I want to show you what we added on Friday, which is really the core part. That will work. So what we added was we added graphing to this and it now suddenly has that classic statement of being able to visually look the at things. I am going to show you v6 routes only but in this case you have the ability to wander through and look at different networks and how they are routed, including the ability just to zap on a particular route and actually go look at how the routing looks on the global Internet. Now, again, I am just going to go through this somewhat quickly. I can't actually hit the one I wanted to hit. So we will hit that one. So all of these have the ability to give you a view of how a particular route, I am trying to find an easy one here, just something simple, so this gives you an idea of how a route from the right?hand side of the screen is seen back from its origin on the left?hand side. As I said, the data is refreshed daily out of the RARs and route views data and it has the ability to sit there and click on a particular AS and actually look at what that graph looks like. It gets more interesting as you get to the edge. It gets more interesting as you look at your customer routes and how they are coming and being seen and, also, you have got other tools here. I have got to keep this down to three minutes and I am going to show you v6, here, for example, are announced routes, here is a list of peers or customers, so you can pick. You can pick. I am trying to pick only European players here. You can pick European players, any AS number, type in IP addresses. And as I said, the graphing is what got added recently and that is what I wanted to show. So if there are questions, grab me, I am wearing a v6 T?shirt. I have been known to do that. And more importantly, go play with it. Feedback back into either myself or the contact us link here. And if you find data anomalies, obviously let us know but I will tell you right now this is not off of our routing it's out of the RAS data so you may be surprised at what you are seeing because the truth is is a little bit more interesting than what you think it is. And that was it, that is all I wanted to get to in three minutes. Thank you.


MARCO HOGEWONING: Thank you. And you broke the spell on live demonstrations. Next up, Chris Buckridge from the RIPE NCC, showing us some out reach activities and other stuff to promote IPv6.

Chris Buckridge: I will assume this is the right one. Hi, I work in the communications department at the RIPE NCC and I want to give you a very quick update on some of the activities that RIPE NCC staff have been engaged in over the last six months relating to IPv6.

A lot of this is coming from an internal focus group that we have formed, which brings in people from the science departments, science group information services as well as communications and training so it's really across the entire company.

I will start with something which will look familiar, Emile's presentation today was part of this, but RIPE labs, which was launched late last year has been a venue where we have been able to do a fair bit of analysis and statistics relating to IPv6, so these are some of the ?? I am not sure why that is happening ?? these are some of the statistics and analyses that people from our science group and information services have done. I won't go into any great detail about most of them, except for IPv6 RIPEness, but I would encourage you to go to labs dot to have a look and there is a compilation of statistics from right around the web so that is not only stuff that NCC staff have done but also things that everyone else is doing in an easy to get to location.

I mentioned IPv6 RIPEness. Some of you may have seen the handouts that we gave out at the beginning of this session which had some of those statistics. The back page of that is actually devoted to this IPv6 RIPEness measurement that has been developed by the training team, with Emile in science group and it's basically a way of measuring how the status of the Internet community in a given country in terms of IPv6, so we take all of the LARs in a country, for countries that have more than five, we give them one staff ?? if they have an IPv6 allocation from the NCC, another star if they have a prefix visible in RIS, another for a Route?6 object, another for reverse DNS information and so, so the result is is a graph like this and so you can see, it's probably easier to look at on the hand out, but you can see these are all the countries in the RIPE NCC service region and then the bars show how well advanced IPv6 is in those countries. As has been mentioned, Slovenia here is coming out head and shoulders above other people, but yes, you can have a look more closely at that in your own time.

We are also really keen to get your feedback on this. It's on RIPE labs and there has been a bit of discussion on the IPv6 Working Group, we have four stars at this point, we are looking to expand it, you know possibly with a fifth star that could be connectivity, but there are a couple of different methods by which we could measure that and so we want to get people's feedback on what you think would be best, what would be most accurate, so yes, I would encourage to you join in that conversation.

Moving on to IPv6 act now. Like the shirt says. This is a website that we launched late last year or early last year. And it's basically targeted at government and the community generally, but it's sort of show casing information about IPv6, it's a one?stop?shop showing statistics, information and the up dates we have made since the last RIPE meeting include how to act now which is something that came out of partly the Cooperation Working Group, we had some government representatives say, what we would really like to be able to show people is targeted information, say if you are an enterprise business, this is what you need to know, if you are a small business this is it, if you are a government, this is what you need to know. We have tried to address that with the how to act now section and we have gotten some feedback from governments and have been mentioned in discussions that governments are having about IPv6.

It is an ongoing process, though, with this site so we are constantly looking to improve it, trying to add things that people might find useful, so I would encourage to you have a look. If you want to join in the conversations that are takes place on the forum, that would be useful because we have some people contributing there who aren't necessarily on something like the IPv6 Working Group mailing list. We have a list of consultants, you can contact us and we can add your name to that. Let us know what you are doing with IPv6, I have been talking with Marco for ways to integrate the work he has been doing, we are always on the lookout for ways to expand the information that we have there. The RIPE NCC IPv6 training course. So this was launched last year. In 2010 there have been 13 courses held so far, there was one held on Monday, so yesterday, at this meeting, and there are 20 more already being planned so this is something that we are really getting out there into the community. It's also, it's a really good chance for us, our training services department is one of the very public placing aspects of the RIPE NCC so it's really good opportunity for them to learn more about community attitudes to IPv6, I have mentioned role playing here, I know one of the activities they do is trying to get, we have a lot of technical people in the training but getting them to role play the managerial section of their business and trying to develop what would be the business case argument to make to management to encourage IPv6 adoption. So that has been useful. We do follow?up surveys to try and gauge how useful these training events have been. And finally, outreach:

We have been doing lot of work here, Jan mentioned IPv6's 10 percent technical and 90 percent political, it certainly feels that way, but Internet governance forum is something has been talked about already and we are strongly focused on IPv6 N 2009 the IGP event was held in Egypt and we worked with that list of partners there to put on ?? adopting IPv6, what you need to know workshop. There is another IGF, the fifth this year, in Vilnius, in September, and that still being finalised, the programme there but we have submitted a proposal to host a workshop called IPv6 around the world, so current and future deployment of IPv6, we are still finalising this and the panel of experts to talk about this, so if you are going to be at the IGF, if you would like to be involved in this we would like to hear from you because we want to get the best panel that we can to showcase what we are ?? to governments and stakeholders at the. The ITU IPv6 ?? is is a new development, you will hear about that in the Cooperation Working Group tomorrow but this is something that we have also been heavily involved in. This group had its first meeting in March 2010, we will have another meeting in September, I think before the very big meeting that the ITU has but this ITU ?? IPv6 is in some ways the playing field on which a lot of the political aspects are really being played at the moment so it's taking up a lot of our time. We have also involved in the European Commission IPv6 workshops which Per mentioned this morning, and obviously, engaging in technical forums like FRNOG and PL NOG. Also on the outreach side is Middle East roadshow and I am going to hall over to Paul Rendek ?? no, I am going to keep going. This is also something that we are working on. The Middle East, we have obviously been focusing on with the development of the Middle East network operators' group, we have worked closely with them and these ?? Middle East IPv6 roadshow is something where taking IPv6 training to the region and really trying to do dedicated trainings which not only targeted at technical staff but also government, just trying to spread the word there about IPv6 and we are working with MENOG, with AfriNIC, we might be hoping to expand it into Africa as well and with ISOC. There have been a couple ?? affiliation with the MENOG meetings already and three currently in the planning stages, certainly MENOG later this year in Jordan, I believe.

AUDIENCE SPEAKER: I'm Paul Rendek from RIPE NCC why we have gotten involved in this. This is a MENOG effort and we don't quite have the commitments of ISOC but we are hoping we are going to jump on board with them and with AfriNIC. What has happened here is that obviously, you know, here in Europe we have spent a lot of time with governments talking about IPv6 even actually going to all their meetings and workshops and training them, but the Middle East is also part of our region and what has happened is that we realise once ?? as we have gone to these IGFs and ITU meetings we have seen that we have got some problems in the Middle East with uptake of v6 and things that are political there so what we decided to do is to go out and do some hands?on workshops inside of MENOG we have done that with a lot of support from this community where we have had five days hands?on workshops at the meeting in Bahrain, Lebanon and also just very recently in Saudi Arabia, this was supported by the Saudi Arabia government which was fabulous. What has happened because of that have is I have been contacted by the UAE government and he actually, the person who is the chair of the ITU IPv6 group and he has asked us to come and do the same thing in the UAE and I was contacted by the Jordan government, they want to us do the same thing there and we are going to do Syria, so all of a sudden this thing has exploded for us there and it is a region that we have to pay attention to. It is part of this RIPE region. I think from now until January there are four potential workshops planned because one will be in Istanbul with MENOG meeting in the fall so I think we have got some great momentum going. The documentation ?? Chris penned this, which was really great so we will be approaching AfriNIC and ISOC and hopefully we can take this to the AfriNIC region.

Chris Buckridge: That brings us just to questions, if anyone has anything they would like to ask? No. OK.



I am pretty sure for all the presenters, finding it quite hard to keep them short because we want to have some time left for this one, should I get the ?? so, brainstorm, yes. Well this is actually where you come into play. We did this whole thing with a new charter and this is basically what we ?? what ?? want you guys to get involved in. This is the new charter. We have got basically those four bullet?points there on outreach, education and sharing but basically everything is in scope, it has to do with IPv6 deployment, IPv4 or IPv6 coexistence. We saw some wonderful examples today, operational stuff, what is happening in the world, a bit of policy, what we can do, the outreach programme, the education, the stuff NCC is doing on behalf of us and the community. The basic question I want to ask is where do we go from here? Is there any specific points we should focus on as a Working Group, things that need attention? What do we need, extra outreach? Which audiences should we approach, which new groups? Platforms that need basically ?? need our attention or we should take part in? Education ? do we need any, and if so, what do we need on the education part? Cooperation, should we have a platform for technical issues, talk with other organisations. We had Kurtis up here, I have been approached by some people that maybe the community itself should get a better voice in the IETF or maybe the community should self?create a standpoint regarding some IETF issues. So, basically this is an open mic session and it's actually up to you to tell us where do you want to go, what do you need? I see Kurtis walking up there.

KURTIS LINDQVIST: I will make a suggestion. I actually think we are doing quite well. I think the presentation like we had today about the CPE devices and maybe also have, I know that is ?? Olaf came to NetNod customer meeting and gave an excellent presentation what happened in standardisation, I think some of these more hands?on presentation and actually showing some of the deployment scenarios would be interesting, maybe here, and have value to the community and maybe that will trigger more interaction with for example, v6ops or what other interests groups there are. I think we have overlooked a lot of the implications in the mobile networks and I do think there is more mobile networks in here than we think, and we might want to have some more agenda topics on those as well because we are seeing a lot more integration between mobility and the fixed networks. So more of that type of ?? because I think that is both outreach and education at the same time and I think presentations have been great for the last few times.

MARCO HOGEWONING: Thank you. As Kurtis said, if you know this kind of deployments like I happen to run into Slovenia into some guy named Jan showing me a working mobile, those things and point them out if you happen to come across people doing this stuff, point out to them there is this IPv6 Working Group and there is that place to show what you are doing and get some credit for it.

Jan: As you pointed out, what I would like to see in this Working Group is stop talking the talk and start walking the walk. As we need to find people that deployed stuff and are in position to present stuff, that is working, not intentions to do something that is working but that is actually working. We need to go more operational, I think. Thank you.

MARCO HOGEWONING: Thanks. Any other people? David, you might want to add something?

CHAIR: Let's get some input from the audience as much as possible. Anybody else anything to say about this?

MARCO HOGEWONING: Geoff might want to comment a bit more on ?? Geoff Huston; you said something about measurements yesterday. I saw you sneaking in. Anything you would like, you might want to add to this.

Geoff Houston: What we find at the moment is actually very few folk who run large scale services willing to test v6 and have it measured, so we find ourselves and you saw in the presentation earlier on this session, the statistics gathered from RIPE, the NCC and we all willingly admit that all we are doing is looking at the geeks and it's a biased sample that you are looking at the folk who are your favourite audience and you are not actually seeing a wider picture. It's difficult for us as the measurement folk to engage in the broader industry, and I really would like to understand how, because you folk all work in it, how we can actually take some of these measurement tools and get them into websites that enjoy 100,000, 500,000 visitors a day, because right now we are seeing numbers that I think we are largely kidding ourselves about in terms of their accuracy. So, how do we use some of this obvious interest in v6 at this point, to actually engage in some useful measurement that will hopefully inform the industry itself, inform vendors, and inform policy. You might have noticed that some of the statistics you saw pointed to 5 percent of the end host being v6 capable. To be perfectly frank, if that is a real number, then you are almost two?thirds there because five percent of a market that is this big is commercially viable. You really don't have to push any more. So the question is, is that really 5 percent? What are we seeing.

MARCO HOGEWONING: I understand you and I actually putting on my axis role hat, what I did was to talk to a couple of our key customers, one of them hags a food delivery service and we are working out to get Pete at that over IPv6 but this suggestion, I made towards the Dutch, this also came up in the Dutch IPv6 task force and one of the suggestion I made was to have those measurements be running on the tax office website because apparently everybody visits them regularly and that might be your Joe Average there indeed you are right, I think in fact, yes, the figures that RIPE shows are nice but they are a bit biased because it's measured on dot dot dot

Geoff Houston: So two things, one, 1.6 percent failure rate that I saw this that whole path MTU, not good enough. Anyone who is trying dual stack up there in user land, 1.6 percent of people not getting is not enough, it has to be 0. There is ways of doing this that I think we should concentrate on telling folk. Yes, you know, go for a smaller MTU on the server, try and get folk to say "I can do v6 dual stack and it will work and my loss rate is 0." The other part of this is to try and get the measurements happening without the browser saying loading dot dot dot; in other words, what we really need is someone to actually do me a Java script that will work on every known browser on the planet, that only gets activated once the page is loaded and shows nothing to the user that work is going on behind their back and if I can have that piece of Java script I am a happy camp every.

MARCO HOGEWONING: Any volunteers.

Geoff Houston: It's got to work in explorer which is the most weirdest on this. Your hand right, let it be known..

MARCO HOGEWONING: We will chase Sander, thank you Geoff. I have two people, Daniel, first.

Daniel: Just to re?emphasise what came up in the discussion this morning where we had the person come out here and, on the stage and say name and shame works and on a country level please, I think please participate in these kind of measurements, even if the Java script is not perfect yet. Get the Dutch tax office or any office, but that is probably a bit high hanging fruit, because we can get a bit more low hanging fruit, I would like at this week, Emile has had commitments from 5 to 10 websites all around the RIPE area to actually run these scripts because we need the data and we need, whether like it or not, to do a little bit of naming and shaming in order to accelerate this stuff. So, there is nothing good unless someone does it, so please consider the little pain of asking the people responsible for websites that actually are getting widely used to run this little Java script. And yes, we might want to improve it but do it now rather than put it off to the next year or so. We need these time series data, we need the name and shame, we need the hard data that we can present to people who are making business decisions and to policy makers and so on and so forth. Do it.

MARCO HOGEWONING: Also, don't only focus on your own website, but look at your customers and go there and tell them and explain to them because that is probably where Joe Average goes. Kurtis you were referring to Daniel.

AUDIENCE SPEAKER: Where do we go from here? I look into the future and think, IPv6 is all about global connectivity and continuing connectivity but if I think, if we have v6 connectivity established, what happens with the v4 clients that cannot be upgraded, not by software, so it's like having a car running on gas but we run out of gas, so in a transition period, if you don't have an electric car in the future you are stuck, so just listing devices that can be upgraded but bring you not into a global connectivity scenario again?

MARCO HOGEWONING: I guess that would be really big list, but yes. Thanks for the suggestion.

AUDIENCE SPEAKER: I understand it's a big task but there are some better devices and some real important things running, nuclear power plants and stuff like that that could be critical.

MARCO HOGEWONING: There is this thing that you might run out of space. Thank you.

AUDIENCE SPEAKER: Global connect activities.

Kurtis: Just to follow Daniel's point, I also think what is happening and to Geoff's point, what will be interesting was the more have a standardised way of collecting data rather than having Google, RIPE, etc., all do this Java scripts by themselves and if RIPE is willing to make a collection. I think we all have in our countries, there are big websites around using v6, one in Sweden called Piratebay, they have statistics and visitors and they publish statistics and one of the big newspapers and they publish statistics too. I am sure these guys would be happy to collaborate and get some of these data, I am show someone in each country knows somebody we can talk to.

MARCO HOGEWONING: One of the things at that came up in discussing this in the Netherlands and the Dutch thing was there is this something about company secrets and nobody is willing to show how much visitors they have, I hope they can show what a percentage of usage is without showing numbers. Two other people.

AUDIENCE SPEAKER: Jamie Stallwood, the information that you showed about RIPEness from the NCC labs earlier on was very interesting. I think it would be very helpful if that information distinguished between sort of end ASes that only have one or two peerings, and those big transit ASes and I think that would help us to maybe look at more geographical targeting of some of this activity, because I know there are certain areas, especially Britain, where the big, big transit provider such as NTL just don't at the moment.

MARCO HOGEWONING: I know Geoff also mentioned something in his presentation about the transits becoming active and the difference between transit and edge networks on that particular case.

Emile: We can certainly look at stuff like this. If you have suggestions about the RIPEness, just come talk to us and we will look into interesting stuff that people come up with.

Alex: I am a trainer at the RIPE NCC and by now I have delivered, well, IPv6 training, well quite a number of times, I have literally talked to hundreds of people who are interested in IPv6 but don't really know where to go. You know, most people they can sort of wrap their heads around the fact they have to use something with letters and numbers and something with colons instead of dots. That is something that they can find on the WIKI page. But at the end of the day, they have gotten as far as convincing their management that they should invest in IPv6, they have done their audit and request an allocation and get their /32 and go, OK, now what do I do? And there doesn't seem to be a single place, a one?stop?shop, really, where you can find all of that practical information. I mean, most of the LARs in our region, they fall into the small category, they are all fairly small businesses with relatively ?? the oversight that you have of the network is reasonably manageable but most of them still have a ?? problems, what the first step is they should take and what kind of practical experiences people have, and once they have ?? once they have fallen flat on their face a couple of times and figured out what their best current practice is and would like to share that knowledge with the rest of the world, they would also want a place to share that information with others, so they could benefit from it. So what I would suggest is that you really create a really solid platform where people can share information and exchange experiences. There are a couple of initiatives out there, but there seemingly isn't a single place where this discussion is taking place.

David: The interesting thing is, in some way, the kind of information, so not so much the ?? what is actually long time ago available in like the 6 BON data and project but obviously there is no 6 BON any more. I nearly felt like is this still a problem, but apparently it is, so

AUDIENCE SPEAKER: I can give you a really practical example ??

David: You could look up providers in your country that did v6, you can look up information ?? places where you could ping to just to see if things are working and stuff like that. And yes, that kind of information is probably lost now.

AUDIENCE SPEAKER: Marco did a presentation on what kind of network set?up they did at Axis role, for the /56 here and the /84s and the POPs over there and really just a very simple addressing plan explaining like this is how we approached it, having like simple tricks like using the port number of the service that you are providing, as part of the address, and people ?? you see ?? people in the ?? their eyes light up and such practical information and I never thought of that myself, I am so happy that somebody else figured this out for me and why isn't this kind of information written anywhere on the web site.

David: You are more looking for internal information actually, so much for the connectivity towards the Internet but how to do it in your company, how to deal with that.

AUDIENCE SPEAKER: Yes, all of it, essentially.

MARCO HOGEWONING: It's true, I do get a lot of feedback on the slides and I already offered you guys to help out and produce some more. I was also yesterday, having a chat with Miriam and I am not sure if she is in the room right now, but Miriam suggested that we should create or maybe possibly create an IPv6 section on ripe.labs and put this kind of information there and I am perfectly happy to take this up as an action point because we are already looking for how to put that CP list and other information but I guess that is where the community comes in. If you have those deployment scenarios and are willing to share your address plan, please ping me or people within the NCC and share that data and show us what you are doing by popping a link to the mailing list, as we were mentioning measurements, I know there has been a lot going on in Norway, we saw some great results on IPv6 readiness from House, that kind of stuff, we rely on you as a community to give us the information, we can't invent it all all by themselves and I guess it's the same problem for the NCC, we need input from you and that is why this Working Group exists and that is where you come into the whole question. Anything else? Any other comments people with feedback? I see Aran walking to microphone.

AUDIENCE SPEAKER: Aran Hughes, on this topic I have made an attempt in the ARIN and NANOG region to create a site that sits somewhere between ARIN and NANOG that manages living documents, primarily for the purpose of this, something with an interface for editors and authors and a way to vote up and down changes to them, to keep them alive, as well as an interface for people to read and search them. We all agree that it's a painful process to go back through PowerPoints over the years and as soon as a presentation is done it's no longer current. I haven't introduce that had in this region at all because there were some concerns that it might be different practice in this region than it would be in the north American region but there is an active list called BCP ? discuss at bind dotcom where we are talking about how to do this, where it lives, what it's called and there is a BoF at NANOG 49 for discussing exactly how to get this accomplished as quickly as possible.

MARCO HOGEWONING: Cool stuff. Maybe Aran, post that URL and some other information to the IPv6 Working Group list if you can do ?? people can take note. There is one other minor point I want to raise myself and I have been working on and that is, well Alex mention it had during the IPv6 tutorial Monday morning, there is some stuff going on with the IPv6 policy document about registration needs and well, I am working on a draft so I will probably push that out after this meeting, unfortunately I don't think I can make it to get it into AP just yet but keep an eye out, it will ?? I will post references to the list. Anything else?

David: So coming back to all kinds of topics that we can give more attention to. There is another thing we can do and in the area of do we have to set up our meetings in a different way? And, I mean, like duetorials can be done, we can also not have a Working Group session but go more to a plenary format or we could go to like an IPv6?only day where we only give attention to IPv6 topics. These are things that are also coming up during our sessions with all the Working Group chairs because we regularly reviewing the format of the RIPE meeting and input for that will definitely be very welcome because it might be that people have comments on that area as well and we can bring that to the table at Working Group chairs meetings, so if there is anything people would say, change a format to give attention to this problem, that might be helpful, as well.

MARCO HOGEWONING: Feedback welcome. At IPv6 Working Group chair at, the address is added to the website right now. If that is it, actually ?? again, mailing list, all ideas, information, discussions, you know where to find us, subscriptions, minutes of the mail list soft and URLs right here. I have got one last comment to make or Maarten Botterman asked me to point out: There is going to be IPv6 survey, it's running in June. I guess the announcement will follow to the IPv6 Working Group list. Keep an eye out on your mailbox and should you have any questions on the contents or the way it goes, ask Maarten, he is sitting right there in front. Thank you. And thanks it. Thank you for attending. See you all back in Rome. Hopefully we have got some work and looking forward to suggestions, presentations and any other feedback you might have. Thanks for coming.