CHAIR: Well, well, well, welcome back, we're just in time. So, thank you for the amazing slide, Remco.
So now we are starting our second session, second part of our session. We will start with some interesting lies and not damn lies from Remco. Then we will have Massimo, who will get us also some insight about measurements he got. Then we have got the EURO‑IX update from Bijal and then a short summary about the peering toolbox from Nurani.
I guess, with that, you could go and start with your part. Also remember to rate the talk because we will need that, obviously. And besides that, I think we are ready for you, Remco.
REMCO VAN MOOK: Thank you, Will. There is ‑‑ I don't have slides for this particular thing. It is part of a discussion that I noticed. And it has everything to do with the public visibility of what we do. Not just within this Working Group, but as a community, as people who run inter‑connection for networks. What is the result of that? The result of that is traffic. Traffic between networks. And originally, that was very visible. I mean ten, fifteen years ago, if you wanted to see how the Internet was growing, you would go look up the website of your nearest friendly neighbourhood Internet exchange and then you could have an idea of what the Internet was doing and looking like.
That's not really the case any more, because, as we have all evolved the way we connect our networks, a lot of that traffic has actually moved to PNIs, or has actually disappeared off the inter‑domain altogether, but especially with the whole Covid‑19 and people being very dependent on the Internet, it's come under some public scrutiny and a bunch of journalists who show up and start basically looking at the statistics that they know about and then start inferring all sorts of ideas of how the Internet is doing.
This is basically, I don't have an answer to this. I would love to hear some ideas and feedback from all of you. Is this a problem at all? Is it okay if there is no sort of public visibility left any more of what's going on with the Internet? I think Massimo has, his presentation that's coming up next, could show a couple of things. And given that entire countries, especially in the lockdown situation that we have been through, are now pretty much dependent on us doing our jobs. How do we marry that? How can we demonstrate, show as an industry, what it is that is going on? And of course, we have all wrapped ourselves in NDAs and God knows what else.
So, really just open for questions, open for suggestions. It doesn't have to be here live right now. It could be anywhere. But I just wanted to desktop that thought in here. I am very eager to hear your ideas and feedback. Please let me know.
And I see one question by James Blessing.
Is the problem a lack of information being published or a lack of knowledge in journalists?
In all fairness, I think it's both. Because, you can't blame a journalist for going to sources that he knows are there and have worked before for publishing something. I also don't think it's fair for us to say, well, these journalists shouldn't be doing this anyway, right.
So, I don't know. But if we want them to stop painting wildly inaccurate pictures, then we should give them something better, and I don't have the answer to that. So just an open question, really.
Anyone else?
WILL VAN GULIK: I don't see any comments either on my RFC or this chat. Blake says:
"Perhaps journalists are used to calling companies asking for comments."
REMCO VAN MOOK: Well, for sure. That's one way of doing it. But that never helped in sort of getting an answer in the aggregate. You can go and talk to your local cable provider and ask them how is your Internet working? And the answer you'll get from their marketing department is, we're fine, thanks for asking. Which may not be the exact answer you're looking for.
I don't know. It's part of it, right.
So, yeah... I don't have the answer.
WILL VAN GULIK: We have got another one, it's a comment more.
One of the issues is ‑‑ Eileen Gallagher says:
"One of the issues that journalists will sometimes use a graph to make judgments that shoots the narrative."
REMCO VAN MOOK: Absolutely. And there are fantastic graphs of most notably Internet exchange points that they can use for anything. I mean, any ‑‑ if they want to depict something bad, then you'll have somewhere where a measurement got missed and traffic dropped off a cliff or whatever.
So, yeah, I see another question from Nurani:
"It depends on how you define 'transparency'. Is the question how much traffic is sent where or the question is does it work, or will it continue to work even throughout the crisis?"
I think that the second bit of this is the question does it work, how do I know it works and will it continue to work? is a fair one. And I think, until now, what has been used as sort of a proxy for that answer is traffic statistics that get published. I mean, if you want to see how ‑‑ if you are going to end up in a traffic jam on the road, you have live traffic information, right.
So, yeah, I think the final, the question that everybody wants to see answered is: Does it work? How do we know it works? Will it continue to work? And we might be a little bit too secret scroll as an industry around that.
WILL VAN GULIK: There is another one, another comment from Daniel Karrenberg:
"That's not a new issue, look up FCC and some notes earlier. The Internet is going to die scares. If you want to aggregate something, you need some mutual place to aggregate and buy for the majority to buy data. Even if it appears to be set serving, we have a neutral place all the RIPE NCC."
I think that's a really good point.
REMCO VAN MOOK: That's indeed a very fair point. Thanks for making that, Daniel. There is another comment from Michael, I think. Can you read that?
WILL VAN GULIK: "In general, there is often lack of capacity especially it relates to the Internet policy‑related knowledge. Many are also struggling with funding, especially now. But even before, journalists are struggling, but there are plenty of journalists and media organisations that work in the sector that RIPE could work with. For example, I help the Dutch new dynamic coalition within the IGF that is focused on journalists stability and many of the organisations involved there could be a good mediator between the technical side and the media."
REMCO VAN MOOK: I think that would be very helpful, once we figured out what the technical side is and looks like and consists of.
WILL VAN GULIK: There is another one from Blake here:
"May I suggest that if anyone here is approached by a journalist, a large challenge has been the physical damage to our sector and threats to fail technician. However, it could be burning marble towers or booby‑trapping or vandalists cutting fibre."
REMCO VAN MOOK: That's a very solid comment. I find it striking that there is, that people are complaining that they are not loading their Facebook videos about 5G spreading Covid fast enough and then burning down cell towers. That's just stupidness that we see everywhere and I don't think that's a particular worry for this industry alone. That's a broader issue that needs to be addressed on a global scale.
I think Aaron has a Q&A, not really a question, but notable that: "Open IX is about to open an connection report which has all kind of measurements data about inter‑connection. We hope to generate something that's not just based on volume of traffic which should be a helpful tool for those who want a state of the inter‑connection perspective and consumable by a large audience without having technical knowledge."
I can only say, Aaron, if that gets published, please drop us an e‑mail on the Connect Working Group list. I would love to see that, thanks very much.
And Anita Nikolich pointed out a link to an opinion piece by Vince Surth about this topic. I think we should get that out to the list as well. So if we can make sure that that gets captured in the minutes, that would be fantastic.
And then I think Andy Davidson:
"In some markets, the media understanding of the Internet is shaped by complaints by end users about the lack of availability of high capacity services to pockets or knot spots or unreliable or under performing last mile. I think we can demonstrate in the inter‑connection world that our services help ISPs and consent networks deliver unusual and unpredictable traffic patterns really well and that the debate needs to move to the last mile."
I think that's a very fair point, and I think that discussion needs to be had and I think that mostly that we can make that go away by showing that we're actually doing our jobs on this side.
I think we should close it off at this point.
WILL VAN GULIK: I think so. And then now actually it's time for Massimo, who is going to talk to us today about the impact of Covid‑19 on the network. So the floor is yours, Massimo.
MASSIMO CANDELA: Thank you very much. So thank you very much also for accepting my talk last minute.
So, I am a senior software engineer at NTT. And the presentation of today is about the impact of Coronavirus on the Internet latency in Europe.
Basically, this is a research that I did in cooperation with these people here. And what we basically did is, like, we said, you are RIPE Atlas, you are the greatest and biggest network measurement available. Can you tell us what you know about latency? So we basically connected all measurements from both anchoring measurements which are essential sea measurements done by RIPE Atlas to what could be defined the core of the infrastructure, like IXPs and ISPs and data centres, and also whatever else the users were defining. And we collected measurement between the 10th February and the 27th March. These dates are like the 10th February is before any lockdown event or when essentially coronavirus was not yet considered super serious, and March 27, when after, basically, almost all countries in Europe experienced some form of lockdown.
We collected only periodic measurements or measurements that are scheduled to be autonomously repeated after a certain amount of time. They sent a list at least three packets every ping, so three RTT times every time. And they have bot sets targets in Europe, thus because the datasets is big enough and we already struggle with that.
So, there are like, for the anchoring measurement in the end of the filtering, we collected around more than 6 billion round‑trip time and for the user defined more than 500 million.
So, now we focus on Italy also because it was one of the most affected and also because I am Italian. What you see in this plot here is essentially you have on the left the anchoring measurement. On the right, the user defined measurement, and you see like the dashed lines, the vertical dashed lines, they are the lockdown events. They are like initially there was a partial lockdown in the north, after more regional lockdown. Up to the last lockdown where nobody can get out of the house. And the two grey areas that you see are the first week before any lockdown and the second week after the most restrictive lockdown, so what we want to do is compare before and after, essentially:
And to do that, what we did is what you see here plotted is not the absolute run time, because we could not have sense to do that, but it is the increase above the minimum for each target pair that we basically we collected. So what we did is we took for each pair, we ordered a measurement for the 45 days, selected the minimum and sorted this minimum to all the values for that pair in a way that we only check the increase.
And as you can see, there are at least ‑‑ you see, of course, that the general trend increases especially for the user defined measurement, but you see clearly that there is an increase of variability and there are these waves and they go more and more high.
Another thing that you notice is that the minimum value increases at the beginning and after cool down a bit and this was interesting also to try to explain.
So, we tried to understand why it goes up and down so much, and we think, okay, this is something during the day and during ‑‑ because everybody is doing remote working and remote learning, we could not have been more wrong, because actually when we plotted the ratio between the time slot of the day between the two weeks, we discovered that the increase on the times was evening hours, not during the day there is an increase, yes, but it's not big like the one in the evening, and especially there is a peak between 8 and 9 p.m. UTC which is essentially double the increase compared to what is the first week.
So, to better analyse this, we said okay, if it's like in this time frame it's possibly people going on watching movies or things like that on the Internet, so we started to take from this set, all the measurements that are targeting content providers that they offer some sort of entertainment.
Unfortunately, it looks like people it the RIPE Atlas only love to measure YouTube, for which you find a lot of measurement, so I am reporting here YouTube, which is one of the the ones we did. On the left you see only round‑trip time increases for the time slot between 5 and 6 a.m. and on the right you find instead between 8 and 9 p.m.
As you can see, while in the morning, 5, 6, there is not really anything, it looks almost ‑‑ nothing. On the right one the lockdown especially, it increases slowly, on the lockdown boom, you have a big increase. And in a way, this confirms that for sure content are contributing time slot in the increase of latencies that we see.
We did also a comparison in IPv4 and IPv6. And on the left you see IPv6, on the right you see ‑‑ the other way, IPv4, IPv6, and you see IPv4 follows the normal trend that we saw before, and there are an increase of variability. IPv6 is, in general, already viable enough independently from the lockdown, but we see that it doesn't not really increase during lockdown, it actually looks as it goes below, and this is again the same phenomenon that we are seeing before and we also basically discovered the same thing with other subset of this datasets, for example in the bottom right we see from Italy, to the rest of Europe, there is clearly a part between again 5 and 6 where it goes low, which means that the first week is actually slower than the second week while you are in lockdown.
So to summarise here. IPv6 is not really impacted so much by the latency increase, mostly I believe because there is so much IPv6 impacts this network especially in Italy. And IPv6 is actually looks for subject to this kind of improvement, that they are going on. And you can clearly understand that when you ‑‑ so, why IPv4 is, if you consider the variation is ‑‑ almost doubles, the IPv6 is just 0.2.
So, and these trends that go down are justified by, for example, the Italian operators, they start doing periodic meetings every week, essentially, like summarising what improvement they did on the network and there was a good amount of cooperation and I really like that. And I would say that one of the reasons also why this research was done is also to check if there is this. And indeed during night hours, so when the network is uploaded ‑‑ is not loaded you see that the latencies are going better than before, and I believe that that is a recent due to this improvement. At least this is the only explanation I came out with.
And just because the presentation is really short, I want to show only two other plots, which are Sweden and the rest of the entire region of Europe.
Sweden, it's interesting because it was not affected by lockdowns, or at least there was no official lockdown, but also in this case, even if less than for example Italy, you can clearly see that there is an increase of latencies during the lockdown period. This could be due to the fact that people aren't any way doing lock downs by themselves or at least since the greatest impact is the evening, night, it could be that they are anyway avoiding social interaction and so they entertainment is anyway online, and that's why we also see that for Sweden.
And if you see the same trends for Europe. Again, anchoring measurement, so core network and whatever else, we see in both cases an increase of course the values of various lock downs happen in the various countries in different times. In various countries they don't even have a lockdown, so there is a more smooth out the trend. But if you compare the various numbers you will find an increase of the second week compared to the first one of an average 13.6% in terms of milliseconds.
So, this is what we manage to come up for now. And a record will ‑‑ a complete record with all other countries will go online pretty soon. And I think my time is over and this is the moment if you have any questions.
WILL VAN GULIK: Thank you very much for your presentation, Massimo. That was really insightful. I have a comment while you were speaking that, we have got also some measurement that was done during, like on the RIPE Labs, and Vesna linked that into the current chat. I I will also forward those links to the mailing list afterwards, and ‑‑ yeah, I will ‑‑ okay. I am asked to copy that link up.
If you have any questions, please go ‑‑ I don't see any for the moment. So ‑‑ but that was really insightful. I mean, it's funny to see how the things moved and what we were expecting. Like that's another case of what Remco was saying before, what we have been told and what actually happened. That's an interesting one.
MASSIMO CANDELA: Yeah, I think it's particularly interesting, this cooperation of when in my case I am aware of the Italian one, but I think this happened everywhere in Europe, the cooperation among operators that every week we were reporting, oh, we improved this, we improved that. And actually it's really nice, I don't have any other explanation, to see that happening a night and to see that improvement. We are also collecting a bit more data to try to correlate this enters when the round‑trip time actually goes down at night compared to before, to correlate them with a specific event of network operators, that would be really nice to do. But in the end, if that is recent, really the kudos to the entire operator world.
WILL VAN GULIK: Thomas King asked in the chat:
"Do you have any measurements of this before the Covid‑19 lockdown? The difference would be interesting to see."
MASSIMO CANDELA: Actually, the entire thing is based on that, so this ‑‑ as you can see, that the entire ‑‑ from the 10 February basically in Europe there was nothing more or less going on about the Coronavirus for real, nobody was in lockdown. So that's why we took this first week, so this first grey area is like the normal status. And the second week the other grey area is when there was complete lockdown and we did the same for each country related to one lockdown we have this dashed line. As I said, are lockdowns happening in time. So we did this comparison. For example, this plot that you see here, the ratio, are the ratio of after and before. So, if it goes below 1, it's better. After, if it goes above 1 is worse after. In this case, it's basically all worse, almost double. But there is cases like this one where actually a night gets better.
WILL VAN GULIK: And I have another comment and I have got a question.
I have a comment that was interesting.
REMCO VAN MOOK: It was asked: "We see enormous growth of NetFlix traffic in the lockdown, but after NetFlix some changing of their video profiles. We saw 50% decrease of the NetFlix peak. Another thing we saw was the 9 a.m. to 10 a.m. upstream peak from home. Is that something that you observed as well? I mean, there was ‑‑ I remember from a few decades ago, so probably end of March, that there was ‑‑ that the European Commission actually asked NetFlix and others to reduce their traffic profiles. Has that been reflected in the stuff you have seen?"
MASSIMO CANDELA: In general, I would say it could be. We don't have specific measurement for NetFlix, because as I said we based this on whatever the RIPE Atlas community was measuring at that time and we are basing it on latencies and we don't have traffic data. We can clearly see that, for example, in the plot on the right, which includes also content, we can clearly see that there is a going up and after a bit of going down especially in night hours. So, both the network improvements, both the content providers reducing the quality of their video definitely could have impacted this. Another important thing that you said about the peak of between 9 and 10. We can see that a bit here also in this chart, where you see that essentially there is a peak between 9 and 10 after it goes down again and after it grows again in the afternoon to be extremely high in the afternoon. So, I totally can, let's say, correlate to what is said.
REMCO VAN MOOK: So, what I can see in this graph is that people take extended lunch breaks. So, I think we need to wrap this up. I got two Swedish comments clarifying that even though the country was not in a forced lockdown, it was in a recommended lockdown. And a lot of Swedish people observed that. That is, of course, exactly what happened, and thank you for clarifying that.
I think, with that, we are going to close this off. Massimo, thank you very much for presenting this. This was very insightful. Thank you.
MASSIMO CANDELA: Thank you for this time. Thank you very much.
WILL VAN GULIK: I think next up we have got Bijal coming up with some EURO‑IX update during this interesting Covid time. Bijal, the floor is yours.
REMCO VAN MOOK: You are on mute, Bijal.
BIJAL SANGHANI: Sorry about that. Can everyone see my presentation? And the presentation is fine?
Hello. It's nice to be here even though we're under different circumstances.
I usually give an update on what's been going on with IXPs within the EURO‑IX membership and just new things that they may be working on. But obviously, you know, we're going through quite ‑‑ well, quite something at the moment with the virus, so, my update today focuses on what we have been, what's been going on and what we have been discussing.
So, the story of Covid‑19 and the impact on IXPs.
As we all know, it started in mid‑January, which right now seems a lifetime away, when we started to first see cases of the virus appear outside of China. By mid‑February, we were seeing more and more infections spread and the cities and countries started to react to this.
We were about to go into a global pandemic. Obviously at the time we didn't really know how serious this was going to be, but, you know, at times like this, it's really important to stay connected and also have ‑‑ share what you're going through with others in similar situations.
Sadly, we postponed, and then later cancelled the EURO‑IX forum. So, we decided to have some meetings online. And we had our first meeting on the 13th March, and since then we have had seven other online sessions.
So, what happened? So like I said our first meeting was on the 13th March which in the UK at least, we were not quite in the lockdown mode and the government had said this they were first of all going to try the herd immunity which is what was happening in Sweden. So, we heard from MIX and NaMeX and TOP‑IX which were around three weeks into the lockdown.
And honestly, I think that first session was really scary, because, you know, there was a lot of other countries on the call who were not there yet and obviously we know Italy got hit the most. So it was really interesting, and like I said, scary to hear what was going on, what was going on there.
At the same time, you know, I think many of you know Nick, so as soon as Nick heard about this, INEX also started to do lots of planning and they shared a lot of the documents that they sent around and shared with their membership with also the EURO‑IX community which again helped others to kind of plan and decide on how to do things. And I'm sure you all heard that also for the first time, the pubs were closed in Ireland for Saint Patrick's Day.
What was going on? IXPs, obviously they were operating fine. They were starting to work very closely with their members to see what was needed at this time. They, very quickly, carried out any upgrades that they essentially needed to do for their own backbone, and then worked with their members to upgrade port capacity. Many IXPs were, and still are, offering free ports during this time, and also offering remote hands, so that other people don't need to go to the data centre to avoid contact and to minimise the spread of the virus.
We saw ‑‑ this was, you know, something I think that was kind of beyond what many, you know, what IXPs needed to do, but, you know, a lot of IXPs did this to help out the customers and members.
We also realised that kind of government relationships came into play because IXPs needed access to data centres. The governments ‑‑ some IXPs, as you may know, are already considered critical infrastructure, and I think this highlighted the fact that some more IXPs needed to be reconsidered as critical infrastructure, because they also needed access and so forth.
By the end of March, more and more countries were under lockdown and working from home. We started to see local NOGs coming together. You know, ITNOG have had numerous meetings and with, you know, over 200 attendees and sharing what's going on in Italy, and I think that's, you know, that's been really really good for us. And also to keep connected and know what's going on with other networks, and share experiences as well.
The UKNOG, they put together a spreadsheet with, you know, a list of people's names and people that could be available to visit a data centre again to avoid the risk of too many people going out.
In terms of traffic, you know, most IXPs saw an increase of traffic from between 30 and 50% and again, you know, like Massimo was saying, the traffic ‑‑ the day and night traffic was starting to kind of even out. And this ‑‑ you know, there could be many reasons for this, and Massimo mentioned some. You know, working from home; schools are all closed, so children are all doing their learning programmes online. But not only that, they are doing a lot of video calls, you know, with their teachers and other students. So there is a lot of additional services being used on the Internet which weren't necessarily being used previously during the day.
Gaming is another one. You know, normally people are working during the day and gamers kind of tend to come out in the evening, but in everyone working at home, this could be quite easily something that's done during the day as well.
So, other things that we discussed during March were data centre access. So, you know, were data centres going to allow people to enter? What criteria were they going to put in place? You know, we have heard of some data centres who were taking the temperature of people going in. You could only go into the data centre one person at a time, which obviously, if you're carrying heavy equipment, can ‑‑ well, it's not ideal.
So ‑‑ but in most cases, I think the data centres were very understanding and I don't think in the end there was any major issues with access to the data centres.
Another topic that we talked about was cable cut. You know, if there is a fibre cut, whether it be terrestrial or sub‑sea, what's going to happen? How long is it going to take to repair? You know, these are all things that you hope don't happen when you are already in a crisis. Things like deliveries. Some IXPs are co‑located with universities, for example. So, you know, deliveries are delivered to the university, with the university being closed, many of the deliveries were then being sent back which, when you are waiting for a critical hardware, you know, isn't again ideal.
And then cross‑connect prices. So with the bandwidth going up, you know, IXPs were offering free ports. But what was happening with cross‑connect prices, you know, nobody predicted this was going to happen, so, you know, maybe these costs weren't budgeted and potentially they could have stopped people from actually upgrading because they couldn't afford the cost connect price and so forth.
So all these things were kind of discussed and, you know, in the end there was solutions for everything, I think. But it's just the kind of things that you think about when you are in a crisis or you don't think about before until you are in a crisis.
And then, of course, there was the cable cut in South Africa and an earthquake in Croatia. I am happy to report that the IXPs, you know, there wasn't any interruption to the operations of the IXs there, apart from maybe a little bit more traffic, a traffic increase.
April:
So, we're only at the beginning of May and April seems like years ago.
So the big news I think in April was that Telecom Italia joined the three IXPs, so MIX, NAMIX and TOP‑IX who were ‑‑ and I know that they all worked very hard together on making this happen, which I think is you know quite an achievement. So that happened at the beginning of April.
Also in April we were now seeing more and more people working from home obviously, but also we started to see community events being done virtually, like we are here, or either being postponed or cancelled because... because there is no flights to travel anywhere at the moment.
And it was becoming clear that the situation was not going to improve soon. So, I think you know, in March, while there was, like, shock and fear; by April, you know, I think we started to realise that this is not a short‑term thing. This is, you know, we are in this for the long run ‑ well, at least until there is a vaccine or some sort of solution.
We started to see again strange traffic patterns, and I think somebody mentioned earlier a comment was that, you know, we were seeing content networks making changes internally so that, you know, they weren't using up so much bandwidth and dropping the quality of the streaming to support that.
So, there were cases where some IXs who had originally seen a spike in traffic, were now seeing a decrease in traffic. And actually while we are just talking on traffic, I just wanted to add that not all IXPs around the world saw an increase in traffic. You know, there were some countries, Uganda and Pakistan for example, which actually saw a decrease in traffic. When people started to work from home, you know, the Internet that they were using was from their office so they no longer ‑‑ some people don't necessarily have Internet access at home. So, the traffic actually went from the Internet Exchange either on to their fixed lines or mobile, and also where the content was hosted. So, for example, if the ISP has the content on net and you are connecting from home, then you are more likely to go straight through them.
So, you know, traffic is one of those things, it depends on where you are and what the situation is. So, lots of more strange things in traffic patterns were seen during April.
But also, by the end of April we started to get used to this new normal.
In May ‑‑ so by May, I think, well ‑‑ so by April, the shock and fear and you were starting to realise that, okay, this is going to be ‑‑ we're in it for the long haul, and then in May it's, okay, how do we get back to some sort of normality? Things like IXs were carrying out essential upgrades, but they were putting certain works on hold that could wait. But then also that gets to a stage where you can't wait for it any longer, so, you know, IXPs are trying to plan and get back to business as this, as this is, you know, as a different kind of normal, but starting to look at upgrades that are maybe not non‑essential or will be essentially soon.
But planning, and you know I think that's the key thing.
Another interesting thing that we talked about on a last call, which was last week, was, you know, how is this going to affect the people at the IXs, and this isn't just about how this affects people at the Internet Exchanges, it's how is this going to affect people in general. You know, I heard a really good presentation from Niall the other day, Neil McRae, from UKNOG, and he talked about the challenges of working from home and people that he thought could work from home very easily are struggling and people that he wasn't expected to work well from home, you know are doing very well, and how do you keep that kind of team spirit going? How do you keep people motivated?
So all these kind of I think the more human and the people aspect of things is now starting to really kick in and ‑‑ yeah ‑‑ see where that goes.
REMCO VAN MOOK: Bijal, can you finish up, please, we are running rather late all right.
BIJAL SANGHANI: Sorry. Just very quickly then, on the traffic. I didn't want to put up a bunch of traffic BoFs, but we have some tweets from a couple of IXs and you see that the traffic has been increasing.
Some other information I thought was nice to share was again mentions that some IXs were offering free ports. I mentioned ‑‑ well sorry, I mentioned Neil's talk earlier and one thing I thought was interesting in the UK here on Thursday evenings we clap for the NHS and care workers. And, you know, we see that there was a dip in traffic when we did this for the first time in March.
There was a new version of IXP Manager release and this was to help IXs identify capacity issues that may be brought up during this time.
We saw more online gatherings. AP‑IX got together, they invited Marrissio from APIX to talk about the situation they had gone through in Italy. Again, ICNOG are doing a great job and having regular meetings.
We also saw a decrease in traffic in India where the IX in India, in India the government asked everyone to turn the lights off and use candles. There was a traffic ‑‑ you'll see what happened there. And I think this was partly put together because, you know, like Remco was saying earlier, with journalists coming in and making assumptions from reading traffic graphs, you know, that this piece of work was put together from INEX to see kind of give the real story from an IX point of view.
REMCO VAN MOOK: I think I'm going to stop you here. Thank you for doing this. I think this is very ‑‑ I thought it was very insightful. So, again, thanks ‑‑
BIJAL SANGHANI: Can I just have one second for the last final slide?
REMCO VAN MOOK: Okay. One second.
BIJAL SANGHANI: So what we have realised is IXPs are central to the inter‑connection solution, they are resilient and scaleable and adapt quickly to the industry needs.
The openness and sharing has been invaluable and I can't say how proud I am of the IXs for being so open and sharing. And this was possible because of the trust in relationships that have been built up over time and I think during this time we also need to continue those and it's important that we find ways to do this, even though we're working remotely and from home.
Thank you.
REMCO VAN MOOK: Brilliant. Thank you very much, Bijal. A big round of applause.
(Virtual applause)
REMCO VAN MOOK: We are going to quickly go over to Nurani. Any questions for Bijal, please ask them offline.
NURANI NIMPUNO: Thanks very much. Let me see if I can get this working. So. My name is Nurani ‑‑ as I am giving this I am imagining all of my dear friends in the Working Group before me. I think a lot of you might know my face but might not know that I have recently joined Linx, but I am here in the capacity as a member of the peering toolbox which is an initiative in the EURO‑IX and I am here to tell you a little bit about this.
It started a couple of years ago with a few people from the EURO‑IX community IXPs, but also some of the networks seeing that there were a lot of new entrants into the peering market. And some of the networks that joined were so‑called non‑traditional networks, enterprises and others, that didn't necessarily all have a lot of experience in peering. So we agreed that it would be great to actually provide them with a little bit more support.
We noticed on both the IXPs that happen there was a lot of hand‑holding needed for some of these networks. But some of the networks and consent providers out there also said they needed to do a lot of hand‑holding. So we said why don't we see if we can do something about that.
We decided to come together and create something that we called a peering toolbox, and the idea really is that it should act as a researches for IXPs and networks with information on how to peer.
So it's really to be a resource for that community.
So, these are the members of the peering toolbox. As you can see a mix of ISPs and peering networks. And we thought it was a good mix of experience to create this resource.
So, what is it? Well what we started with was to write some best practices documents. And to sort of create a place where you can go and learn about sort of the basics of peering. The idea is not that we document in everything there is to know about peering, there is a lot of good work out there and we don't claim to be the only authorities on this. It's really just ‑‑ the idea is really to create something that's useful to everyone.
So we don't claim to be the creators of a peering vibe and we don't intend to go into any sort of discussions about the best peering strategies. But it's really more of a resource that people can use. And we saw that there clearly there is a lot of good information out there, but a lot of it is spread over lots of different organisations and websites. But there are also gaps that we thought that we could try to fill.
So quite practical information on peering.
If I'm going a bit fast here because I know that we're running behind.
First of all, we'd like your input. If you have some good material, then please send it to us. You might be, you know, you might be a peering manager who has had to put together a little handbook for a new staff member, or you might have had to give presentations internally in your organisation explaining what it actually is you do and how that works.
So if you are happy to share that, please send that to us. But also, if you have good material that you are maintaining and that you host, you can also let us know and we can link to you. So we want you, and we want your input. We know that there is a lot of good experienced people out there and this is a community‑driven effort and you are part of the community. So...
So just as one very quick thing. We put together a super simple survey. This is not meant to cover all aspects of peering. It's really to get a sense of some of the simple challenges that particularly new entrants have. So, if you are a network, please fill that out, I don't think it will take more than five minutes to do. Or if you are an IXP, please fill that out. Hopefully we can use that information to inform or future work but also to share that with the community.
We are also talking about putting together some other materials that not just documents, but webinars or other things like that.
If you think this is interesting and if you want to contribute, please contact us, and there is a brand new mailbox that has been created for this. So, it's hardly been used, so please feel free to start using it. If you want to help us.
That was it from us.
REMCO VAN MOOK: Thank you very much, Nurani.
NURANI NIMPUNO: Was that fast enough for you?
REMCO VAN MOOK: That was perfect. Thank you so much for speeding this up. We saw a couple of questions come by. The website is indeed not up yet. So if you try to find it or click on it, if you get a result, that's wrong. So, with that, I think I'm going to close this session. I would like to thank all 465 participants for sticking with us until the end. Look for us on the mailing list if you are not on the mailing list please subscribe to Connect Working Group mailing list at the usual address.
And that is it from us. Thank you all, enjoy your very quick coffee until your favourite Working Group, Services Working Group, starts in about five minutes. Thank you all very much. Enjoy the rest of the week and hope to see you again soon. Bye.
WILL VAN GULIK: Thank you everyone. Bye.
(Coffee break)
