These are unedited transcripts and may contain errors.

EIX Working Group session on 6 May, 2011, at 9 a.m.:

ANDY DAVIDSON: Morning everyone. Let's propose to make a start then with EIX this morning. Thank you all for coming. It looks like there wasn't too much casulties, certainly all of the present he is appear to have arrived so well on it. There is a very, very slight change to the order, Nurani is going to present the first update but then it will proceed as originally planned. On update on the policy proposal that we discussed in the Wednesday session, there was certainly support in the room for EIX, the Working Group, to ask the NCC to reserve some address space for future Internet Exchange points so that Internet backbone projects in the future can start way after the deployment of the last /8 or start of allocations for the last /8 and that was fairly well received by the people in the room on Address Policy so I am now going to work with Sander and Gert to write up a policy to actually cause that to happen, and then I will be asking awful you who showed support in the room to continue to show support on the Address Policy Working Group, which is where the actual decision will be made. So that work is progressing and hopefully that will be at a very advanced stage, if not complete, by Vienna, as long as we don't meet too much opposition in Address Policy. Nurani, are you ready to begin? Do we have NCC people doing the scribing? We will give our scribes and Jabber monitors a moment to set up. There will be ?? the plan is that we are going to get to any other business sort of five or ten minutes before the end, so if exchanges in the room want a moment at the microphone, then they will be able to come and grab me in order to do that just before the end. Additionally, it's worth plugging the mailing list. If you are in the room right now and you have an interest at all in Internet Exchanges operations or this session, join the EIX Working Group mailing group which is where the agenda is assembled and where discussions can take place in between the meetings. So, do also join the EIX Working Group mailing list if you are in the room. That seems to make sense.

NURANI NIMPUNO: Good morning, everyone. Aren't I lucky, standing here this morning? I have a bottle of whiskey from the whiskey BoF last night, does anyone care to ?? you had enough yesterday. I am from Netnod Sweden. A lot of slides I am not going to talk to but I put them in there so people can refer to them if they need to. How many of you know Netnod? Most of you, OK. So I don't need to go through it. We are a nonprofit organisation and run exchanges points and the route servers and do Anycasting of TLDs. I won't show any picture of bunkers. We have got 73 peers, approximately, at the moment. And for those who you not familiar with us, in Stockholm the situation is a little bit special because, because we are located in these underground bunkers, we actually include the price of the fibre when connecting in the connection price so if you don't understand why the fee structure looks the way it does, then come to speak to one of us.

Kurtis is not here now but he is probably here later and my colleague Emily is at the back of the room so feel free harass her later on. I am not going to go through these because ?? yes, you can see them on the website. We have got ?? it's two piece of news that might be interesting. One is that we are establishing, well, we have established a route server at Netnod, and so we started with Stockholm at least and we will see how we go, and because we have got two separate redoesn't switches in two separate bunkers in Stockholm, we have two physical machines, one at each switch, and we are using BIRD which seems to be a nice piece of software. We do both v4 and v6 is it's available on all VLANs. And it's a separate autonomous system number, 52005. One of those big ones, we support MD5. And there is a lot of tech information here about, you know, how to use them, but since we don't too have time I won't talk about those. The other thing that is interesting and new we are looking at one of these remote solutions to connect to Netnod. We don't have a fancy product name yet so let us know if you have any suggestions, it's pseudowire carrier connects with a 10 gig port to us and then they can re?sell that. I think they are rather ?? there are other exchange points in the room who have similar solutions so the concept is not entirely new but it's something a lot of people have been asking us about.

So we currently in a testing phase, actually. We have been ?? just this week, I was reading the tickets back home and that seems to be going well. We don't really have a lunch date yet; we will probably be looking at autumn, we just need to sort out contracts and finish the testing and all those things. So let us know if you are interested in that, either as ?? well, as pseudowire carrier or if someone is interested in purchasing it through the Netnod remote thing.

That was quick. All right. Well, that was it.

(Applause).

ANDY DAVIDSON: Any questions? Great. That must have made a lot of sense then. Now Elisa who is presenting a brand new exchange, Arizona IX.


ELISA JASINSKA: I have coffee instead of whiskey, I think in the morning that is better. Some of you might remember me from my previous role at AMS?IX and I have joined Limelight networks in November last year and Limelight dot so they could make me feel more at home by opening a new Internet Exchange out in the US so I can do things I have been doing before already.

So we opened up Arizona IX as an initiative too far local exchange in Arizona. It is operated by Limelight networks, we build it up and put in the hardware and the data centre providing the space in corporation with them, it is still run and operate bid Limelight so if you have any questions, concerns, you can come up to anyone at Limelight and ask about it.

So to put a little bit in perspective, I don't know how many you know where Arizona is? Somewhere here in case you ever get there with your network. The idea was that a lot of the traffic that we have seen at Limelight was going through POPs in LA or somewhere around the US, not really anything that has been delivered locally in Arizona, so this was general idea why we would need a local Internet Exchange there.

We offered the usual contents, every other Internet Exchange does too, a peering LAN ports, various connection type, etc., you can look all of that up on our website. We opened up the exchange in January this year, had our second member connected in February, opened it up with Limelight as the founding member, obviously, so second member, and then in March we established a first peering session over the exchange, we are busy connecting the third member right now and I thought those members are very abstract so we have to put them a little bit in perspective.

So, January 1 ?? I will read it out, it's a little small, so January one member, February two members, May, we are now connecting up a third connection to the exchange, extrapolating from that we should have six members by November this year, we are looking forward to that. And a fun fact: The first peering session that is now up and running on the exchange was a v6 and is right now a v6 exchange only.

(Applause)

And it even generates little bit of traffic.

Thank you.

(Applause)

ANDY DAVIDSON: Any questions for Elisa? Can I get a 10 gig port?

ELISA JASINSKA: Of course, there is an order form on the website.

AUDIENCE SPEAKER: How many sessions were up in January? I know you said there was one member.

ELISA JASINSKA: In January, zero. But we were aiming for v6 only exchange and it took a while to dual stack everything so the first session was in March.

AUDIENCE SPEAKER: And I have got another question: Who is ultimately responsible Dave Tempkin ?? for engineering and operating it.

ELISA JASINSKA: That would be me.

ANDY DAVIDSON: Thank you for the questions. Andreas, are you ready? The next update is from DE?CIX in Germany.

ANDREAS STURM: Good morning everyone, I am Andreas and I will give you a brief update on what we did in the past and more especially what we do in the future. So that is new claim, it's DE?CIX is where networks meet, currently 400?plus networks and we are just promoting Frankfurt as digital had you been because it's just ideally located in the centre of Europe, ISP, POP and 10 facilities and that means it's very competitive market and it means best value for you and location means of litency to many ?? this is fairly new. We just opened up our focus on peering but more and more on interconnect solutions for ISPs and carriers. DE?CIX should be the one?stop?shop for metro co?location and bandwidth and campus fibre so it's peering and interconnect solutions for you guys.

What is metro bandwidth? It's just protected and affordable 10 gig connections between all major data centres. I guess you all know Frankfurt and how we are distributed over Frankfurt, we have got an east side campus and west side, they weren't interconnected so we changed this into interconnection between these sites, so it's very easy to reach your customers from Ancotel to Interxion and so on. And the next one is the campus fibre, it's the old PI private interconnect but now only used to be on interaction campus, now it's everywhere or nearly everywhere, we are present in the different data centres like Ancotel and so on, so you can make a fully end?to?end interconnection between east and west of Frankfurt.

Other news: We open up a new peering site in Dusseldorf, what I would like to stress is the save the date in October, the first two days are business events, real business for carriers and ISPs and on Wednesday, the DE?CIX customer submit so I would like to invite you all to join this nice event as a party at the end, it's more technical event and then on Thursday and Thursday, Friday, I guess, it's the German network operator group meeting in Frankfurt, so you only have to travel once to get all these nice events together. Thank you very much for your attention and sure, join DE?CIX.

(Applause)

ANDY DAVIDSON: Thank you. No questions. Are you ready, Peter? Peter will present on behalf of ECIX, also from Germany.

PETER LIEVEN: I am from ECIX and those of you not familiar with us, I want to bore something with numbers. We have 76 connected ASNs at the moment, 65 gigs peak traffic and 64% of our members are IPv6 enabled and we have high number of 2710 gig. Ports really right now, still heavily increasing.

We are operating in the northern part of German name three locations in Dusseldorf five locations in, Hamburg with four locations with our core is IPHH, some of you may know this location, and in Berlin we are operating also in four locations and core is speedbone. You can look up at our map if you are interested where exactly we are.

So, this are the rough facts. Some fiction: There are some press releases and announcements recently that may suggest Dusseldorf looks like this. To be honest, I want to gave and friend of mine called me and said there is no Internet Exchange in this and I said really, that is all we want to say about this. As you can see, Dusseldorf is quite a big location and very nice city, it's worth travel. So no other, from fiction we go to the future. We are also starting a re?seller programme will officially will start at the 1st of June. It will be available in all ECIX locations, and.the idea is the same partners transport to peers in VLANs like from AMS?IX or presented like it will be at Netnod, and we do this with 10 gig e?ports only. So this is the first news, and the second news ?? sorry, the first partner for this programme will be Atrato and there are more to come so if you have your two layer connection provider, ask them and they will be in Dusseldorf and will be pleased to transport you.

The second big news is that at one of the first exchanges we look across the border and we will add value services by adding VLANs on the member ports, each v6 member as of today is able to connect to AMS?IX with our I think in the third quarter of this year we will remote connection ability to LU?CIX which makes us a little bit better connected. And the good news with this is, we had the first AMS?IX set?up on Monday, our customer decided to join AMS?IX at 10:00 and he was ready at 6:00 in the evening to peer there. So this is quite fast, I think.

So we have also done a lot of upgrades, just some rough numbers. Dusseldorf interaction is upgrading a new co?location room which is independently feeded by power, different power supplies, different cooling aggregates and we will also put up a switch there for those peers who want to have a redundant connection, we upgraded Berlin with new switching infrastructure, upgraded our Hamburg set?up to have more core at IPHH and in all locations we deployed equipment to support the MPLS and remote peering services and we finally, decided to do all our object particulars with the flat optics, quite a good decision.

So events you might want to join: At the 26th of May there is the opening ceremony of the new data centre rooms in Dusseldorf and I think every interaction customer is invited to join this event and have a look at the facilities and also at the Dusseldorf. In June, we will be planning beering and grilling barbecue in Hamburg, different news for that on the website and mailing lists and we also invite you to join DENOG 3 in Frankfurt in October when we will be present.

So that is all from me. Thank you, if you have questions, ask me, visit our website or write an e?mail. Thank you.

ANDY DAVIDSON: Thank you. Any questions for Peter? No. Thank you, again.

(Applause)

Now the update from France IX and after that InterLAN and links.

FRANCK SIMON: Good morning everybody. I am the CEO, France IX. Just an update after 10 months of preparations. We are now, we do have now 130 members connected to France IX. We have 6 POPs live and the seventh one In Marseilles should be live by the end of the month. It should have been live before but we were needed equipment from Marseilles to Paris because we had to migrate the biggest and it has merged with France IX and we were needing, we were in Paris to do that.

We have, today, about 50 gig of traffic but it's increasing a lot and we do expect to have much more than 100 gig of traffic within one month, one month?and?a?half, because we have a lot of orders pending and we will need to upgrade the backbone and need to make sure that we can provide such connectivity to the members.

Our route servers are working quite good, so today we have more than 10,000 IPv4 prefixes and some few IPv6 prefixes also.

So, I will not develop that slide, it is just reminder about the services. The main point I want to override here is Gateway we have with user exchange so we have Gateway with Sphinx, one ongoing with LU?CIX and Lyon IX, across to Paris. And the main point I wanted to show you today is that we have a lot of 10 gig E connections pending so we will need to upgrade from two times to directly 120 gig and this is not just doing a provisions by doing this, that is minimum we have to put to make sure we have no connection with the backbone so we have done a design up to 40 times 10 gig in the core of that, we need to increase the capacity we will be able to do it and we have implemented VPLS in the core backbone to make sure we can increase the capacity and use all the links as active links.

So, that is it. I wanted to make it short, so if you have any questions, just drop me a mail or directly at and mail at info [at] franceix [dot] net and we will be able to answer.

ANDY DAVIDSON: Any questions? Thank you, again.

(Applause)

Now the InterLAN update. Thank you.

LUCIAN OBROCEA: Hi, this is not really an update, it's introduction we have come to RIPE for some years but this first time we are doing a presentation. I first tell you some information about our location because during my talks with some people around here I found out that not many know things about Romania. In eastern Europe, our neighbouring countries are Bulgaria, Yugoslavia and Hungary, Ukraine. Our main connection points in Bucharest and one of the largest in the area both of traffic number IP routes and participants.

IP traffic in Romania, is the fourth in the world by general IP speed and first in Europe by state of the Internet report. Usually connections in Romania are 20 MB PS with 1 gig coming up fast. And local peering traffic and is done with IPs, we got ?? 100 megabit and international traffic is somehow limited to a few megabits.

InterLAN was established in 2002, it's owned by nonprofit association. We have 60?second partners, 10,000 IP routes available by route server. Everybody peers with everybody. There is no mandatory point?to?point BGP, anyone can apply own peering policy, if you don't like someone you can skip that. We have 100 megabit, one gigabit and 10 gigabit connections available and we are present in many locations and the biggest one is the ?? largest neutral co?location.

We have a few projects rolling out. The first one is national POP project. We have the problem is we in Romania there are hundreds of small ISPs from a few 100 customers to a few thousand and they can cannot reach us because all over the country. The fix is we have discussed with some large partners in Europe and opening some virtual POP, over 30 in the largest cities in the country, and all of those ISPs can get to us very quickly. This will be done with a very low cost four times the usual transit. This national traffic will be available, only for Internet Exchange traffic and not for IP transit or other business.

Other projects is to get some big content providers to our Internet Exchange. The problem is that all those 100 of small ISPs don't have direct access to large content providers and are fix to this is identify those large content providers, invite them and offer them some knowledge and support for deploying in Romania.

The last problem that we have now is that all these partners are already connected don't do anything else but peering traffic and we try to encourage other type of traffic like selling services by increasing ?? decreasing the price for VLAN and lower high speed port pricing.

That is about it.

ANDY DAVIDSON: Thank you. Any questions? We are really racing through the presentations. Thank you very much.

(Applause)

Now John from LINX represent the last EIX update for today.

JOHN SOUTER: Hopefully a traffic graph?free presentation today. I am going to talk about a couple of topics of things going on at the moment. We have quite a few new staff. Three of the four I have highlighted there are in the room so, Derek, if you could put your hand up, and Mark and Ben. Jen is not new but has put her hand up anyway. More significantly, we are finally decided to recruit member relations team, one of the tasks I asked Ben to do when he joined us in January. We have, in fact, found some people; we haven't actually hired them yet but we are going through the process. We are probably a bit late in coming to do this but when you reach a certain size you have to acknowledge that a lot more effort has got to be put into care for the existing membership so that is what we are doing. We are still recruiting quite heavily, we are finding it difficult to find a database developer, we have been trying to do that for quite some time and we are going to start recruiting some more engineers shortly as well.

Why am I showing you a picture of the /TKAO*UPL now? Any Italians in the audience? You will know it has been under construction for at least 500 years, no sign of it being furnished any time soon and what does that remind you of? Monitoring systems. However much you think they are done, they never are. We have had a brocade of outages in recent times. And the good part about that is that it has forced us to improve. And we have improved in a number of ways, obviously in a short presentation like this it's not scope to go into it in great detail. I think the monitoring system particularly has been forced to improve. We have come up with all sorts of creative ways of monitoring for the bizarre things that we have been experiencing in recent times. And that has been good. It's somewhat in parallel to that and for a different motivation, we also decided to rewrite from scratch the system we had to make it more generic and more extensible. We had an alarm?suppression system because that is what the key issue is and in a exchange you can generate as many alarms as you can conceivably think of but the question is which one of those are you actually going to want to propagate through so attention is given to them and hence referring to it and we are currently in the last stages of a complete rewrite of that.

I think when we finish that project we may well share the results of that with the community because even though our code is not suitable for other people because some of the concepts and idea might be of more general interest. We are much more protective set?up now. There is not really scope today to go into the details of that but it was borne out of necessity.

We are just about to install new optical platform. The motivation here was to make more efficient use of our existing dot fibre infrastructure. We had made some discussions which locked us into certain types of optics and then accordingly, using the ?? only using 4 port line cards which is of course a very inefficient way of using switches so it's going to get outside out of some of that lock?in. We have chosen MRV after a fairly extensive evaluation and this is going to be an infrastructure that fits in between member connections and switch gear, it seems to be a very logical development that others are doing. We are still in the progress of migrating our architecture. We made the decision in 2010 to abandon our long held ring?based architecture and move to VPLS. Recent events have caused some delays but we are still committed to doing it and we expect to make the switch sometime during the summer. The internal debate of course is what is the summer, according to the engineers it appears to last until November. As CEO I think it ends in July. And that is it, unless you have any questions?

ANDY DAVIDSON: Any questions for John? Thank you for the update.

(Applause)

I thank everybody for the update. Now, this is ?? this presentation is rather than just an IX update, it's also message about some of the research and development projects that happen.

ALEKSI SUHONEN: My name is Aleksi Suhonen, I am from Finland, and maybe I should first start with some history. So, Tampere which is the second largest city in Finland and you might be better aware of the other exchanges point operator, physics, operate in the cap pal area and one in northern Finland. There has been private peering and some IP public peering in Tampere since about 1997 and TREX has started up back then but we actually started or created the company in 2002, so right now, we have eight members and there are 10 peering partners on the exchange and this corresponds to 11 ports and 13 peering partners so there is a couple of autonomous systems there that you can peer with that aren't really members.

We use an Arista switch because when we were upgrading the switch last time, it was the only available model that suited our needs, and we are organised as a company instead of a nonprofit organisation and the reason for this is that physics, the other exchange point, is a nonprofit organisation, so in the turn of the century, we, some operators in Finland perceived that physics had trouble moving forward because they had ?? they were an ATM based exchange and they wanted to move to a gigabit ethernet exchange format but they had trouble getting consensus so we wanted to create a counter?balance to their way of decision?making by creating a company which had a small group of neutral owners and a small chair ?? I mean board that does their decisions. And if physics had been a company, then we would have probably created a nonprofit organisation, so we wanted to create some counter balance. So that both exchanges could develop their services better.

So we are not really in competition with physics; more like in cooperation.

OK. And when we were creating the idea for the exchange point, another trend on the international arena of Internet Exchanges was that traffic was frying away from exchanges. Big operators TL1 and near were creating private peering sessions everywhere and some were even withdrawing from exchange points and it was feared that the remaining medium and small operators wouldn't have enough traffic on exchanges to justify paying for the exchange membership or cost ?? connection costs or whatever. And actually, this trend is still continuing. So, a couple of IXPs back then had the idea of starting to create other services to make it lucrative for the big operators and, well, any operators, to stay at exchanges and we also realised this and we made it part of our core ?? the core of our operation so in addition to basic peering, VLAN were operators connect their router and set up BGP and so on, we wanted to find out where that the Internet Exchange point could make itself more useful, so an Internet Exchange point is at the very minimum, an ethernet switch where operators already connect, so we started thinking about ways we could reuse that switch for other things as well, and so, we at the ?? the research bit means that we are basically promoting new technologies and helping working in the research community. Development is that we develop and help develop new services at Internet Exchanges that are useful, and this is, of course, always tricky; there are some services that you can do at an IXP that would actually be competing services to the very members, the customers, of the exchange, so you can't provide just anything. So it's a fine line and we are all the time watching how our members react to whatever we do. And then as a marketplace, so basically, operators, there are usually very small operators, medium?sized operators and very large operators at every exchange and, well, while the very small and very large operators usually don't come to peering agreement for obvious reasons, basically the IXP can still facilitate business between them or, yeah, well, various kinds of user groups can be created where the operators can offer whatever services they wish to each other.

OK. Now, if you build it, they will come is the old saying, and in our case, it's sort of didn't happen ?? that is laser, so we started in 2003, and, until about January 2010, we had just 10 to 20 megabits because initially we had three members and the number of members went between three and five, back and forth. We sort of got caught in a chicken and egg problem where we had a dozen potential members who said if you had half a dozen more members we might join as well. But finally, in 2010, we were able to get over that problem and we have had nice growth since and the last graph shows typical daily traffic graph, sorry, graphs; it's about one gigabit which sounds like a small number but consider Finland is a country of population of five million so the total traffic is about 33, 35 gigabits daily, so we currently have about 3% of the national public exchange traffic.

The way that we were able to get past this problem of chicken and egg was that we started to hold workshops which were basically centred around certain small or large topics and the first one we held in 2009 was around multi?exit peering. This is something that everyone here is very familiar with, but in Finland, up until that point, most operators only peered in the capital region, so physics had just started an exchange point in northern Finland and we had the one and we noticed notice weren't peering in the northern exchange point either so we made a workshop of it so that we got a few speakers to tell how it should be done and what the pitfalls are and how to do it. And that is got us over the first hump, and then the next ?? the next big hump in the traffic graph came from the next workshop where we concentrated at visualisation at IPv6 adoption and we are planning the next workshop for this year for securing infrastructure, things like securing BGP and DNSSEC and things like that. Each of these workshops is also associated with our Tampere peering forum where first operators can or whatever members can short presentation of why to peer with them and so on and migrates to a social event which is usually located at a sauna location. There is food there and so on. Not in the sauna itself.

And now to the actual topic. So research and development, what sort of things have we been doing. The first bigger project that we were involved in was called IPLU, IP reliability, and the main focus was to try to create gauges to measure how secure and how reliable is the whole Internet in Finland, for example. So, basically, while the average person might think that the Internet, if it breaks, nobody cares, or maybe a few kids won't get something but the fact is that if the Internet did break, logistics and banking would stop there and then, which means that, if you go to the local shop, there won't be anything in the shelves in a couple of days and even if there was something to buy, you can pay for it. And they say that every country is two meals away from an arcy so it is national security issue. So we took part in the project in 2005 and 2006 and we have been indirectly involved in it since, as well. The actual research was conducted by the national Finnish national research centre and it was mostly fund by a couple of ministries. Basically, the idea was that you don't want to break the Internet to find out how reliable it is; you want to be able to model it and they took some inspiration from nuclear reactor reliability because you don't want to break the nuclear reactor either to find out how reliable it is. There is a URL if you want to read some of the white papers or find out more about it.

The.second bigger project is future Internet Finland which is part of a cluster of international projects so there is EU framework and in the US and Japan and so on. And the Finnish, it's called ICT SHOK with its full name and there is Nokia Ericsson, the national research and so on and so on. A lot of universities, including the local university, local technical university in Tampere and we take part as a best bit provider. There is a URL for public deliverables and stuff.

And this is a huge project ranging from layer 0 issues to layer 9 issues, so I will come back to this at some point.

Basically, when we were thinking of making the exchange useful for potential members in other things, in addition to just peering, Anycast services was the first thing that popped up and these day, pretty much everyone has at least some of these, so IPv6 adoption services and Teredo and 6 to 4 routers and the autonomous system 112 serves are RFC 1918 reverse mappings, community DNS is one of the Anycast providers for the Finnish top level domain and they don't have any nodes in Finland before one of ours. We always wanted to F?Root and maybe some other services again. There is a URL if you want to find out more.

DNS recursors, last year the root zone was signed and we found that the operators in Finland didn't have recursors that could take part in the validation of ?? DNSSEC validation of any records, and also we couldn't find any recursors from operators that were dual stacked and, at that time, actually, there were a couple of operators where we heard a lot of complaints from their customers that the recursors weren't really working that well so we decided to make our only internal recursors public and made sure that they actually supported all these features. We run unbound and BIND 9 on them. There is some more information on them there.

And this is an example of a service that does compete with members, sort of, so we are, all the time, wondering ?? so far we haven't got any complaints, but, well, remains to be seen.

An expansion of that. The DNS 64 recursors are something that maybe will be widespread, at least I am very enthusiastic about the whole DNS 64 technology when it comes to, you know, modular ?? you know these USB sticks, 3G USB sticks, because each one of those it's one IP address and now we have run out of IP addresses, the operators won't be able to sell any more of those, soon, because they require the unique address but if they migrate to IPv6 and they have access to NAT 64 then you can still do pretty much anything with those ?? that connectivity, except maybe play some games or something. Now, this is actually part of the future change project I talked a couple of slides back. Right now we are testing Ericsson's implementation. It's not a stand?alone implementation; it's part of one of the bigger products and, I understand that the product has not been announced yet, so I am not going to go into too much detail. Nokia, Siemens has an implementation is part of our research group and we might test out that, as well, and in due time we will probably also test open source implementitations or Juniper, Cisco, whatever other implementations. Of course Ericsson has tested it internally and everyone has, but if you try to do load testing on it, it might behave in one way, then you might test with protocols and it might behave another way, but in ?? on the real Internet, when you have all sorts of customers using it at the same time, like games and e?mail web browsing and everything, that is when the regular bugs really come out to so we wanted to try to create a testing environment where the real bugs would come out and we found some bugs and they have been fixed. So to make a public trial, we had to use a public prefix instead of the private prefix and right now, the first prefix is the one we are using. There is some information again in the URL.

Route server, everyone is doing this and pretty much everyone's implementation right now looks the same as Andy and Wills from LONAP and we are pretty much indebted to them as well. Right now we are running BIRD on two different Linux machines, but since we are actually running an Arista switch and has neat feature that you can run virtual machines inside the switch itself, they support you hacking around in it, we are planning to actually run one of these route servers in a virtual machine, open BDS D in the switch itself so right now we have one switch so faith sharing doesn't matter, it might be good, and run open BGP D. There is URL ever since then. We have been meaning to run route servers when we started, but with three members they had the direct peering agreements if they were going to peer so setting up route server didn't really make sense but now we have suddenry doubled or quadrupled our membership it's starting to make sense to set up route servers.

And now it's time for questions. I tried to run through these ?? oh two announcements: Yesterday ?? no, not yesterday ?? on Wednesday, we came ?? became the 59th member of EURO?IX, thanks to Serge, and another is that we have actually, knew there is a list of these research projects, we are actually right now in the process of signing few ?? Internet test bed Finland memorandum understanding with computer science centre in Finland. So a continuation of all these. Questions?

ANDY DAVIDSON: Congratulations on a very busy schedule. Are there any questions? Are you willing to share some of the R and D projects you do with other exchanges so that other exchanges who want can install your ?? or similar?

SPEAKER: Many of these are pretty much just install DB and install this packet and put that configuration file in there, so ?? but, yeah, we could, you know, make the documentation presentable and then present it.

ANDY DAVIDSON: Thank you. Are there any other questions for Aleksi Suhonen? I can't see any so thank you very much.

(Applause)

At the beginning I promised that anybody who hadn't submitted a paper before the meeting, could present ?? could stand up and present their exchange and I know that that was taken up by VIX and 6 and Equinix so if you ?? are there any other exchanges in the room who want to do a quick one minute update as well. We have time for maybe one or two more after that. If not, shall I hand over to Harald first from VIX. Dave will present for TELEX. Harald first from VIX and then we will progress further.

Harald: I try to make this presentation as short as possible so I had to decide whether I take one slide or two and I decided to take the long version to have room for improvement. This is very short update of Vienna Internet Exchange. This year, we celebrated our 15th anniversary and we want to celebrate it with you and with our customers, and there is a little ambiguity in this slide; we decided to reduce the 1 gig ports by 30 percent but not the bandwidth, the cost is reduced by 30 percent by the first of May so that was question I already got so I am sorry for that, it was very late when I made this slide.

We plan to do some more birthday offers coming up this year but don't want to say too much about that currently. And as we are local host for the next RIPE meeting in autumn, we kindly invite you to get in contact with us if you have any questions, perhaps not regarding the RIPE meeting but surrounding, if you want to have some information about Vienna, the site, what to do, where to go, where you can buy the second best beer in town, just drop us a line and we will be happy to help you with that.

ANDY DAVIDSON: Thank you, any questions?

(Applause)



SPEAKER: Slovenia is not a big country, there are two million inhabitants, but we do have ?? we do have tradition not only in brewery, we also run Internet Exchange for 17 years now since 1994, and it is a small Internet Exchange with 18 members but half of them are v6, I will not say ready, half of them peer with v6. Don't ask me about the traffic, please. We have two locations, mostly on Cisco, Catalyst, 64900 series and a single VLAN. We are simple it, we don't have no route servers and no portal yet. We are planning to install two BIRDs this year and we will have to renumber the v4 part of the Internet Exchange.

To end with, I have actually one question for you, here: We run into a dilemma, namely two numbers are demanding for quality of services in Internet Exchange, they run voice?over Internet Exchange and they don't have money for separate pinnings for separate equipment so we are running into this dilemma: Should we run quality of service on an Internet Exchange and if we do it, how would we do it on layer 2 or 3, should we use cost or task fields or traffic class fields in IPv4 ?? IPv6 or should it interfere with traffic, should we remark the, excessive traffic or police the excessive traffic, etc.

Remco: So I have got two comments on this. I want to remind you that quality of service is actually a fantastic market willing term because it is all about dropping packets, not improving quality of packets. Second is, if you have got two members who want this is there is a fantastic layer one solution, it's a patch cable.

SPEAKER: I know but there might be more of them.

Remco: Then run more patch cables.

AUDIENCE SPEAKER: My view is that it should ?? the exchange point is a layer 2 service so, my my view you should be honouring the layer 2 quality of service bits and the operators then should ?? and you should make your standard way you treat those and make it available to members and then it's the members' responsibility to map their internal quality of service or class of service or whatever into whatever you provide over the shared medium. That is my view.

SPEAKER: For use layer 2 we should move to trunk pours then, that is our dim ma, we don't want to move to trunks. Have no simple access ports with no trunking.

AUDIENCE SPEAKER: We at TREX we provide, since we made these additional services part of our basic service model so that it wouldn't be just peering; we have a VLAN 4 which is the basic peering and then other VLANs, so our basic model is that the member can decide whether they want trunking or not and, so far, I think we don't have a single customer port without trunking, even though we specifically ask them every time do you want trunking or not and they say yes we want VLAN 4 in trunking so from my experience I don't think moving to trunking is going to be a problem.

ANDY DAVIDSON: It sounds like as part of this work, you will have a conversation with your members about this and reach a decision so please make sure you come back and tell us how that goes, the whole process and what you ?? what you discuss and how, that will be interesting.

SPEAKER: I might do that in a year or two.

ANDY DAVIDSON: That would be really good thank you.

Remco: Good morning from the other largest Internet Exchange in Paris, the Equinix Internet Exchange. Just a quick update. This is what it looks like right now. We are in the process of upgrading our switches. We run out of ports so we double the chassises and that is already happening in our own facility in Telehouse, we will be doing that at the end of May. There is, and this is a worldwide unique for Equinix, there is a new separate website for the Equinix ex change in Paris, and on it you will find a lot of information about the exchange including member lists. It doesn't replace the IX Equinix dot com portal which is the customer?only part of our website. We are growing, yes, we are. This is the only graph I will show. It's up and to the right. Looks like we are going to be hitting 150 members somewhere in June. And that is it. Thank you very much.

ANDY DAVIDSON: Thank you. Any questions for Remco?

Applause)

Other than Dave from telex, is there anyone else in the room from another exchange to wants to give a one minute update.

Dave: From telex, I just wanted to let everyone know where we are at, kind of, obviously we are in the US market primarily in Atlanta, Dallas, New York and Phoenix. We are ?? we've undergone significant expansions of the exchange in New York. We think we have added something like 14 new members in the past three months and we are constantly looking for other ways to bring cost?effective connectivity to people within New York, because as a lot of, you know, the 25 broad way can be expensive to connect in. So, the other markets, Atlanta, has been doing phenomenally, I am not sure how many people in the room, would have reason to interconnect there but would love to speak to anyone if you do because we are trying to figure out how we can help this community better interconnect in those tier 2 markets in the US. Thanks.

ANDY DAVIDSON: Thank you, Dave. Any questions?

Applause)

Thank you. Okay well, this will conclude the EIX Working Group for RIPE 626789 thank you for all attending I hope that was useful. Please make sure that some feedback, if you have any feedback that you contact Fergus or I and make sure to join the mailing list, if you want to present at the next meeting in Vienna you can all the information in good time and also further conversation in between RIPE meetings.

Thank you particularly for ?? thank you very much to the stenographer and scribes and Jabber liaison as well. It's always very useful and very useful in the feature and tech support team so we will see you at RIPE 63. Thank you.

(Applause)