RIPE 82

Daily Archives

18 May 2021
.
14:30 (CEST)
.
Connect Working Group
.
18 May 2021
.
14:30 (CEST)
.

REMCO VAN MOOK: Hi kids, do you know what time it is?
.
I can't hear you!
.
Who lives on some fibre glass under the sea?
.
All together now:
.
Connect Working Group!
.
We are virtual so we might as well be under the sea somewhere.
.
Welcome to your second most favourite Working Group of all time. Connect Working Group coming to you on a Tuesday afternoon, just to mess with everyone's schedule.
.
We have a number of interesting topics for you today. Let's just get to the agenda.
.
I'm not going to talk about the emergency exits...
.
The agenda: Opening. Well, this is an opening. Welcome. Hi.
.
Scribe: We have the lovely Hans Bakker, who has been volunteered by the RIPE NCC to be our scribe for this session. Welcome, Hans, thank you very much.
.
And the format for this session is, we're going to have a couple of presentations, then we are going to share with you. After some housekeeping stuff, and then we'll have a bit of open microphone session time for bits, a few bits of pieces and open discussion.
.
Moving on to the housekeeping bits.
.
I trust you have all read the minutes from RIPE 81? And you are all in violent agreement with these very carefully drafted minutes. That is excellent. So, those are approved and shall be published as such.
.
Then we have the Working Group Chair selection process. Will, do you want to talk about that now?

WILL VAN GULIK: Basically, we were reminded that we need to mention again what's our selection process, in case it wasn't clear, and so what we did as a reminder was shamelessly copy the one from Cooperation, and that's still in effect because we didn't change that, obviously, and basically, what we will do is publish that on the mailing list as a reminder, and I think that is basically it. There is not much to say about that.
.
If anyone has something to say about it, just reply to the mailing list when we publish that, but I think, yeah, nothing changes there.

REMCO VAN MOOK: Okay. So, just to remind everyone. We discussed this in past, the Working Group Chair, in RIPE 72, which is five years ago now ‑ my goodness ‑ all right. That's it for that.
.
Code of Conduct policy development chat logs. Flo, can I tease you for that one.

FLORENCE LAVROFF: Absolutely. Just let me start with the chat logs. So, for the chat logs, it has been decided that those were considered as being part of the meetings material and that those should be registered as well, so they should be saved as well. So, just for you to note this. I think it's quite important while we progress this session. Guys, I will let you continue on the COCN policy development.

REMCO VAN MOOK: Okay. Then we have feedback about the virtual Connect sessions. I mean, this is the third time we're doing one of these. I don't think we have had a lot of feedback outside of the meetings. Is there anything that anyone would like to share at this point, how they think ‑‑ what they think we should be doing differently? If not, we are just trying things out here as well.
.
I am just going to leave that open. If anyone wants to comment, we can come back to that later.
.
Then, the next morning...
.
What's next on the agenda? We have IXP traffic in the Covid era by Simone Morandini from MIX‑IT. Simone, the floor is yours.

SIMONE MORANDINI: Good afternoon to you all. I hope you can hear ‑‑ I hope you can hear me loud and clear. The slides I am about to present are one of the ways basically to describe the last 16 to 18 months by looking at how the traffic evolved on our IXP, and on our IXPs as well. So, what I would like to share with you is basically an analysis on traffic data seen by multiple point of view.
.
Unfortunately, Covid is still here, but perhaps we are now able to see a little bit of light at the end of the tunnel, and so maybe we can cautiously start to think about it as something of the past. Let's everybody hope that.
.
We collected our data. We extracted some macro information and we also asked our neighbour IXPs if they could confirm the same trends, or if they observed something different.
.
Well, on our part, we surely identified it ‑‑ identified the three main phases: what he we call the vertical wall during the first lockdown; a second phase when we thought the pandemic was already behind us; and a third phase when we realised that basically we were wrong.
.
There are new takeaways that we can share, but I'll ‑‑ I'll get to this later on during the presentation.
.
So we can really start from the beginning.
.
So, these 3D graph you see on the left‑hand side of the slides represents the weeks before the first lockdown was announced, and then the last one, the one in red, is the actually the very first week of lockdown. At that time, it was not a matter of if it would have happened, the lockdown, but rather a matter of when it would have happened. The formal announcement in Italy, it happened on a Sunday, on March 8th. And again, it's the week depicted in red in the graph. And after, let's say, undecided Monday; on Tuesday, March 10th, for the first time we exceeded the one terabit threshold and for us it really happened overnight. That's why we referred to it as a sort of vertical war.
.
At first we thought it could be an isolated episode, but, as you can see, it turned out to be the start of a new phase, where a growing number of people started to use the Internet as they never did before, basically, and the graph on the right‑hand side illustrates the general change in the behaviour. The red line is the traffic average over the six weeks of lockdown, and the blue lines represent the traffic average during the weeks immediately before.
.
And the shape of the traffic clearly highlights how, instead of a single daily feed, we were facing almost three identical peaks throughout the day, probably for three different reasons, but, nonetheless, the shape changed very much.
.
By focusing a little bit more, another thing that can be noticed is that weekdays, in the top graph, were not so different from weekend days, in the bottom graph. My apologies for the notes in Italian, but I guess the graphs are pretty clear anyway.
.
Okay, there is generally a slower start of the day during weekends, but you can see that daily peaks are exactly identical. Again, here, the red lines indicate the average traffic during the lockdown weeks and you can see how the traffic inflates week after week.
.
Then, the summer came, and, as always happens, the traffic decreases. It's something physiological that happens every day, and 2020 has been no exception to that. However, what we notice is that the holiday period was somehow shorter, and, all in all, traffic did not decrease that much compared to what happened in previous years. This is probably due to the fact that people worked more, generally speaking, and possibly when they actually went on vacation, that vacation was shorter than usual.
.
As I said at the beginning, the idea that the pandemic was over was clearly wrong. So, after summer, we basically had other lockdowns. Italy got divided in different areas: Yellow, orange and red, depending on the the level of the pandemic, but, all in all, again we observed other lockdowns where we basically saw that the behaviour was exactly the same as the first one.
.
So we arrive at Christmas time, actually this graph embraces a much wider period of time around Christmas, but again, what we can observe is something a little bit peculiar, because one of the problems of the home schooling during the first lockdown of the spring was actually the shortage of tablets and laptops that families had available to study or work because basically every person in the house needs a device. Who knows? Maybe Santa took care of that by bringing a lot of new devices or maybe the offer of Christmas movies was particularly rich, but you can see how many times we had to update our traffic record during those weeks around Christmas.
.
We surely celebrated new year's eve as well, maybe not for long, because the traffic drop around midnight is pretty visible, but it is clear as well as short. So maybe not so much to celebrate of the 2020.
.
All in all, this one at the top is the shape of traffic that we now see as being the canonical data mix, at three well‑defined moments of the day.
.
One thing has never been modified, it's our pasta time that is sacred to us, and not even the Coronavirus is going to change that. And the other fundamental point is that the daily peak is always around 10 p.m. and there are really no doubts about that.
.
The graph below, instead, represents the traffic exchange mix by a content provider connected to those access providers, and you can clearly see that, here that the pattern of three hours of lessons in the morning separated by two clearly visible breaks.
.
As I mentioned at the beginning of the presentation, we asked our neighbour IXPs whether they could confirm or deny our findings, so basically to see what they saw from their point of view. And many thanks to ESpanix and France IX for giving their contribution.
.
Spain had a development in terms of the pandemic that was quite similar to the one we had in Italy, and, in fact, their graph basically tells you the same story. As you can see, there is a good portion of 2019 here as well, and, in order to appreciate even more their vertical war, and the other similarities that you can observe here are again the slowdown during summertime and the clear growth in autumn and up to the new year.
.
Now, ESpanix as well, it doesn't show much difference between weekends and weekdays. I don't know if you are able to see here, but this one on the left is a Sunday, and the other one is a Monday, and they basically overlap to the minute. So, again, very minor differences here.
.
Our friends from France‑IX are actually in a position which is a slightly different. They do not connect that much eyeball networks to CDNs, so their growth has been more organic, let's say. It didn't show sudden spikes in their statistics, and it was mainly due to the increase of remote working.
.
This is also the reason why they are not influenced that much by sporting events, but rather by the distribution of software updates. Anyway, they registered a 40% increase in traffic during 2020, which is certainly remarkable.
.
Let's go back to MIX MMR for just a moment, because we are in the position to observe this whol phenomenon from a slightly different angle, because we manage our own data centre as well in addition to the peering LAN. I showed you how the traffic increased on the peering LAN, but the request of new Cross‑Connects actually follow the same path and in 2020 we had, I have to say, a spectacular increase for that. When we built the second data centre in 2016, we have been smart enough to prepare a room in order to be able to accommodate new requests much more easily than we did in the past, because in our first data centre, we had the tradition of using, let me say, tailor‑made fibre cables to keep the cabling as organised and as neat as possible and to have exceeding of fibre in the racks as little as possible. But this would not have been a sustainable approach with these numbers. And, in fact, we now are able to provide a meeting‑room facility for our first data centre as well.
.
Everybody in this industry is very well aware about the impact of the distribution of software and game upgrades because it involves many players. It involves the CDNs, it involves access networks and it involves, of course, us, IXPs, as well. And we were able to see some difference here that we were able to notice.
.
In this graph, the orange line represents the day when the content is distributed, and it's compared to the very subsequent day, in blue, which is, let's say, a normal day ‑‑ a usual day. You can clearly see how different it went on these two occasions, with the second one evenly distributed throughout the entire day.
.
And finally, in terms of content, there is no doubt about what end users want to access? Football is still the number one sport in Italy, and our graphs state that very clearly. These are just a few examples between the end of 2020 and the beginning of 2021, but it happens basically every week, and it doesn't even have to be a big match any more in order to be seen in our statistics, but of course when it actually is one of those matches, the result is more visible.
.
This is the traffic graph of our friend and colleagues at NaMeX in Rome during the last Juventus‑Lazio match. They registered their all‑time peak in traffic and it happened for a number of different reasons. First of all, it was prime time of a Saturday evening. Juventus, as a team, has a very large supporter base. Lazio, on the other hand, has the majority of their supporters in the surroundings of Rome, and on the technical side a content provider is actually peering at the IXP, and so all these factors have to keep local traffic local, which is something that we are we are happy to see. If you want to know more about that, we publish a joint article with NaMeX and topics as well. Here is the link so you can read on if you you are interested.
.
And that is actually all from me. So thank you very much for your attention, and if you have any comments or questions, I'm happy to take them.

FLORENCE LAVROFF: All right. Well, thank you very much for this presentation. Really, as always, long live football, really. So very nice.
.
And now I am wondering do we have any questions here? We do, we do have a question from Rob Evans: "Thanks for the info. What date did the ESpanix graph comparing Sunday and Monday cover? Was it Easter weekend? And would the Monday have been a holiday in Spain?"

SIMONE MORANDINI: I don't know how to answer. I mean, I have been given these graphs out of courtesy from them, but I actually don't know which exact days they represent. I get that, in Spain, the Easter Monday is a ‑‑ is not a working day there as well.

FLORENCE LAVROFF: Okay. That would make sense to me. I hope it answers your question, Rob. Just wondering do we have some other questions now for Simone? One round, two rounds, three! I don't think so. I think that was it.
.
So, Simone, thanks a lot and I say to the crowd, please do not forgot to rate this presentation. It's always nice for the speakers to have some feedback. So, thank you.

SIMONE MORANDINI: Thank you.

FLORENCE LAVROFF: All right. So now we can move to the second presentation that we have for this Working Group, the presentation from Bijal. So this is a tradition that we have here in this Working Group, to have an update from Euro‑IX. Bijal, always nice to have you, welcome, and feel free to start your presentation.

BIJAL SANGHANI: Hello. Hi everyone. Thank you for having me again at this RIPE meeting.
.
So, what is EURO‑IX? It is an association of Internet Exchange points, and today we have 71 members. I'd like to welcome our newest members NIXI and Eurasia Peering, who joined us in 2021.
.
We also have a number of patrons, and patrons are organisations that typically work close with IXPs, so you can see a number of those there.
.
Usually during this session, I talk about some news from the membership, so I'm going to do that first today and I am also going to talk about some of the EURO‑IX activities.
.
Let's start with the member news. First up, we have NaMeX, so, mark your diaries, 9th June, NaMeX has its member meeting.
.
GR‑IX reached a new traffic peak which ‑‑ and they completed their platform upgrade, which now they have fourteen 100‑gig ports connected.
.
ANIX was created back in 2018 and is managed by NaMeX. They have doubled their members in the last twelve months and are also looking to upgrade their hardware.
.
LUCIX, on their road to becoming MANRS compliant, they have activated route filtering and invalids are now dropped and about 35% of the accepted routes are RPKI valid. So some good news there on the security front there.
.
VIX has reached 1 terabits of traffic. So this week they are in the middle of their hardware upgrade. So good luck with that as well.
.
Now, I am going to talk about some of our activities.
.
First up is the IXP database. The IXP database is a database that collects data, basically, from IXPs. It uses a standard JSON format called the IXF Member List, and it contains data about the participants which is the IP address, ASN and locations, but also it has details about the hardware and route servers. We have got documentation available at the URL on the page, and I'm thrilled to say that we recruited a product manager this year who started in March and he'll be working on the next stage, which is data visualisations and the analysis, which I think is the exciting bit now is to kind of visualise the data. So Leo Vegoda joined us in March.
.
Today, we have 271 IXPs reporting to the database through the APIs.
.
We are looking for sponsorship, so if you are interested you can check out the link there, but I'd also like to say a big thank you to our current sponsors because obviously we couldn't do this work without them, so thank you all very much.
.
A quick update on the peering toolbox, we presented this at last year's RIPE meeting. Unfortunately, due to various reasons, the development in 2020 was slow. We did conduct a survey and we got some really interesting feedback which we have analysed and the plan going forward is to have a panel. So, look out for that. The website is also now live. So you can go and have a look at the website and if you have any content that you'd like us to include, then please contact us and send that over to us and we'll happily add that to the page there.
.
We are also looking for volunteers to join the peering toolbox team. So if you are interested in joining the team, then please get in touch. If you want to know more about it, also get in touch and we can explain what's involved.
.
With obviously us not being able to meet last year, we started doing a number of virtual events to first of all keep the community connected but also to continue to learn and grow, you know, things that we have not been able to do over the last year.
.
So, we had a number of presentations. This year, in 2021, we have learned about the basics of governance and best practices for non‑profit boards, which is a very interesting presentation for our community. So if you are thinking about joining a board or work with a board, I think it's quite an interesting talk to listen to.
.
We also heard from Todd Parker and Ryan Gunter from Switch who presented on their proposal of best operational practices for maintenance notifications. So, this was also really interesting, so if you manage a lot of peers or working with a lot of Internet Exchange points, I really recommend you watch this presentation, or if you know Todd or Ryan, get in touch with them directly and they can talk to you about their proposal.
.
We also heard on the NIS 2 directive and the impact to IXPs, which, again, is, you know, something that's impacting not just IXPs, but other networks and hardware vendors and all sorts in the community. So it's important to keep on top of what's going on there.
.
Coming up, we are looking forward to a BIRD route server tutorial, a presentation from one of the ‑‑ our patrons on the future of inter‑connection and data centres and an 8‑route server tutorial. So these are actually open not just for members, so, you know, anyone in the community can join.
.
Another series that we started last year was the 'Let's Talk, Let's Act', and the idea here is to talk about things that can ‑‑ talk about topics that can help inspire and motivate people to, you know, make a change or take the next step that they are not sure about taking. So we had one on engaging with youth ‑‑ more with youth in the industry last year, and we also heard from women leaders at IXPs. This year, we have ‑‑ we have heard ‑‑ we looked at ‑‑ sorry, I was just trying to read ‑‑ I saw my name and I was trying to read the comment and got distracted. This year so far we have looked at exploring the power of mentoring. When Florence asked me for this presentation, she asked me to speak a bit more on that, so I'm just going to talk about that.
.
So, mentoring is a topic that's come up and, you know, there are many industry organisations that are offering this. But what we wanted to do was kind of have a discussion and look at and hear from different points of views. What does it mean to be a mentor, what does it mean to be a mentee, what is your role, how can you even set up programmes?
.
So, you know, what did we learn? So we learned that mentoring can be rewarding and powerful. What does it mean to be a mentor? We have lots of people asking, if I want to be a mentor, what do I need to do? Do I have to know ‑‑ you know, all the answers? Do I have to, you know, train? Or, you know, what do I actually have to do? So, you know, what we learned was that it's really just about sharing your experiences and point of view and being able to give some direction and support. It's important that if you do decide to go into this relationship, the mentee, you know, should be doing most of the work. The mentor can also learn by hearing different perspectives.
.
Setting up a programme. You can start small and grow, reach out to other groups that can also be mentors, so you don't have to just do it, you know, in a particular group you're in. So, it can also, you know ‑‑ something that more industry events could start doing as well, something your company could offer, and I think also the NOGs are also a good place to do this.
.
One of the challenges that came up was finding a match. So, if you already know somebody that, you know, you look up to or you think could be a good mentor for you, then why not ask? Because, you know, I think everyone likes to hear a compliment, so if you are ‑‑ if you go and say, you know, you like what they do, and you would like them to be a mentor, you know, it's hard to turn that down.
.
And, you know, I think another thing we learned was that if it doesn't work, that's fine, you can try again. It doesn't mean that, you know, the first time you get matched with someone, that that, you know ‑‑ it's going to be perfect. There is a road and a journey that you go on for this.
.
That was a really inspiring discussion. So if you are interested in hearing that, you can see that on the link there.
.
And that's it. Thank you. Are there any questions?

WILL VAN GULIK: Thank you very much, Bijal, for that. That was very, very good. I think the mentoring is really something we need to have and to be able to push because we need a future, I guess, so it could be helpful for that. I see we have one question.
.
Kostas Zorbadelos asks: "How does IXPDB compare to PeeringDB; do they have overlapping informations?"

BIJAL SANGHANI: Yes, so it does have information that is the same as what is available in PeeringDB. However, the IXP database source is only the IXPs. So that I think is the difference, because the source of the IXPs ‑‑ the main source of the IXP data and the PeeringDB is from networks unless you have the IXF JSON configured in PeeringDB, which is also available, by the way, so, you know, IXPs or networks that want to enable that, they can do that through PeeringDB as well. Does that answer your question?

WILL VAN GULIK: We'll see. I think it does. We'll see if he reacts. But I think it was clear enough,yeah. It's really simple to go and implement actually, so...
.
Okay. So do we have any other questions? Any comments? We are actually ahead in time. So that's really good, which is amazing. No other questions. Oh, yeah, we have got one question. Another question from Farzaneh Badiei from Yale: "Other than presentation about MANRS, does EURO‑IX have any special programmes that incense advise the members to adopt MANRS? Do you know how many of your members have adopted them?"

BIJAL SANGHANI: That's a good question. No, I don't know how many of the EURO‑IX members have actually adopted MANRS, and we don't actually have a programme per se, but we very heavily support the existing MANRS programme that ISOC are running, and, you know, we have had many presentations, and actually the IXP programme came out of one of the EURO‑IX forums. So, you know, we are ‑‑ we work closely with the MANRS team and ISOC in that regard but obviously we encourage IXPs to adopt MANRS, yeah.

WILL VAN GULIK: Okay. Thank you very much for that.
.
Okay. So I think we covered all the questions. And now we have got the right font with our slide deck, thank you, Remco.
.
Okay. So what's next?

REMCO VAN MOOK: Much, much later...
.
Of course I wanted to do that voiceover. My French is by far the best.
.
But, now we come to the open discussion part, and there is one point I would like to bring to your attention; we're not going to have a discussion around it right now but I do want everyone to be aware.
.
So there was a presentation submitted to the Programme Committee about GAIA‑X, which is a new European project set up by a number of very large companies that proposes a European sovereign data infrastructure, is I think the phrasing they are using. The Programme Committee did not accept that presentation and we decided, you know what, we find this very interesting because there is a huge ‑‑ on the infrastructure side, there is a huge play in terms of network connectivity and the inter‑connection in there as well, we'd very much like to hear what's going on.
.
Sadly, the GAIA‑X people didn't accept our invitation, which I'm hoping and looking forward to hearing from them in the future. But it is probably good to have a look at what's going on there. There are some interesting founder members and some interesting choices, I am just going to put it that way.
.
I'll leave it at that.
.
Then, other topics. This is your chance to run up to a virtual microphone that's in front of you, and let us know.

WILL VAN GULIK: We have got the link for some references in the chat.

REMCO VAN MOOK: Yes. That's indeed ‑‑ thanks, Rob, that is indeed the link.

WILL VAN GULIK: Do we have any comments, questions? Well, tomatoes would be a problem because we are a bit far away.

REMCO VAN MOOK: You are welcome to throw a tomato at your own screen, that's fine, we don't have to clean that one up. It's very quiet. In that case, I am ‑‑ oh, Daniel wants to say something.

DANIEL KARRENBERG: This GAIA‑X thing has me intrigued because I see sort of, as a networker and as a citizen, I see a good intention to develop an industry that can provide sort of services that the big, you know, Amazon, Google, big four, big five, private sort of outside our region and that realises the necessity for our regions to develop their own industry, and I am wondering ‑‑ but on the other hand, it has, to me, a little bit of a smell of, you know, government industry policy, not really realistic people in it, mostly for the money, no concrete results, so I wonder if anybody, especially since I see that some of the IXs are taking part of it, any feeling about the real relevance and the real difference it might make? Since we have time, why don't we discuss that.

REMCO VAN MOOK: We need to discuss that actually with the GAIA‑X people actually in the room, and, since they are not, I'm hoping that we can do that another time. I do think that there is a bit of homework to be done for all of us. I have already seen an offer from Alex Yoda (phonetic) in the Q&A section to ‑‑ who mention that is both DE‑CIX and AMS‑IX are part of GAIA‑X and is going to liaise something for RIPE 83. I would be delighted to have that content for the next presentation ‑‑ for the next session, Alex, so, consider that already accepted. So, yeah, I mean, I don't know how to discuss this without actually having like the proper, like, insights and preparation. I have somewhat mixed feelings around it, like you just expressed, Daniel. I do think that we need to be aware of what's going on there and that's why I brought it up.

WILL VAN GULIK: I see that we have got Erik that was queuing, so I will grant you, Erik.

ERIK BAIS: Thank you. I would like to ask the audience an interesting question here in regard to anti‑abuse. There has been some discussion in the Dutch communities about how far can we ask the Internet Exchanges to either stop serving us routes from networks that are doing bad things, let's put it that way, specifically the typical hosters that, you know, do ‑‑ are popping up on all the lists, and one of the things that I would like to ask the audience is: How can we provide a non‑discriminatory list that on one side provides a good option to say, you know, these are the bad hosts that we don't want to peer with, and, as a community, say, we want to have the Internet Exchanges and we obviously know that, you know, you can't just disconnect them from the member organisations or as a customer, but you can actually tell, you know, if you are on this list, you need to do additional work with the route server to actually get the peering done with that network so that, by default, you are securing the rest of your members from all the shit that you get from the various networks.
.
Now, the question is: How do we come to such a list? Are we looking to things like the 'don't routes list' from Spamhaus? Are we looking at the things that the guys from HostExploit did in the past and see, you know, what they have in the top 50 list, for instance? But this is something that we're looking for and to see if we can actually push that forward, Internet discussions with the various Internet Exchanges.

WILL VAN GULIK: That's a really, really, really good question, a really good point, I think. Let's see if we have got comments like that right now. Otherwise, I think that should be definitely taken to the mailing list or so some discussion, further discussion, because that's a really, really good point, Erik, I agree.
.
And I don't see much there. And I don't hear you, Remco.

REMCO VAN MOOK: So, let me take that to the mailing list. Thanks for bringing that up, Erik. I think that's a very good point. I don't see ‑‑ I see Harry Cross making some comments. Would you like those read out, Harry, or is there something I can do with them or... do you want to speak to those points yourself or...?
.
I'll take that as a no. Well, okay, I'll read them out.
.
Harry Cross says: "I find the concept of deep platforming and ASN like that worrying. We'll need to make the thresholds for this very apparent, providing clear detail of what offences runs an ISP placed on that list."
.
That's solid feedback, I think.
.
"Considering the majority of the abuse my network receives from Google, I don't think it's realistic to expect them to change their behaviour and care", is another interesting point.
.
Right. Okay. I would suggest that we move this to the mailing list. Erik, is that okay?

ERIK BAIS: Yeah, very good.

REMCO VAN MOOK: Thank you.

WILL VAN GULIK: Okay.

REMCO VAN MOOK: That was some interesting extra content. I don't think we have a lot more other topics that people are bringing on. So, unless my co‑chairs disagree, I think we're going to call it for RIPE 82?

FLORENCE LAVROFF: Just before we do that, I think that we should do another round of feedback about the format of the question, about the importance of having this discussion slot at the end of the Working Group for next time as well. So, if you have any comment about that at the right moment, feel free to raise your voice, and, if not, indeed I would agree that this is the end of Connect Working Group at RIPE 82.

REMCO VAN MOOK: All right. As always, we very much appreciate your feedback. Rate the talks. Rate the presenters. Rate everything. See you all again, even though it's just on a list on my screen, and until the next one, bye.

FLORENCE LAVROFF: See you later.

WILL VAN GULIK: See you later.
.
(Coffee break)