 Yeah, hello everybody. I wish you a dry welcome here in this dry tent and Let me introduce to my right side. There's a system architecture of the RIPE NCC Emil and also he's a data addicted and he makes some cool Experiments is with internet roots. So give him a warm applause and yeah, enjoy the talk Hello Thank you. Hello nice crowd. Hello remote crowds I'm Emil and what I would like to talk about is something called hacker spaces Jedi everything is interconnected But how and this is actually a play in three parts. So we begin with episode four So in the beginning there was the internet right so you can all play the song that is under this I wasn't sure if the audio would play so I I didn't there but Like a while ago the internet was was developed and it's all nice and we all know it and we all love it But we all like it to be better So how do you make something better? You try to measure it? And That's a rhetorical question, of course Do we want to have a better internet and there's a nice Dutch saying which is made is weighted to measure is to know so that's So how how do you make a thing like that better? And we have lots of tools to measure the internet and How many of you are familiar with trace route? Okay, so I don't have to really explain what trace route is so you run it from your local box to the rest of the internet Simple so say your box is in Rome, which is this this I mean there's lines here. It's supposed to be a map of Europe So you have a box in Rome and all roads lead to Rome, right? But what if you want to measure from the Netherlands to Paris that's that's you cannot measure that from your local box So if you want to do trace routes as a measurement thing But want to measure all of the internet your one box and trace route won't Just won't do it. So what do you do? Oh? yeah, the other thing is that Internet paths are weird. This is not Like your regular car system. It's it the traffic system that With a car you typically take the shortest path, but in the internet these could be wildly inaccurate for instance friend of mine is trying to establish Internet exchange point in Albania And he actually measured from two points in Albania. He measured this weird path. So that's that's Yeah, way way way way way way longer than than a normal path He used that with the system that I'm gonna talk about You measure that with it But path from a to b so the path from me to you Could be different from the path from you to me, right? It's there's a symmetry there and actually as somebody tried to measure it and he came at only 12% is symmetric So doing it from this one point out doesn't guarantee that all the roads back are are the same So there's there's a lot of weirdness. So how do you deal with that? Well, the cloud is a thing nowadays. So you're just use lots of other people's computers and in our case these computers are Something we call right path loss. So that's a system some smells very small boxes There's one hidden on the stage actually Can you see it? It's right here. So It's these these types of boxes that we put that we give out to people To help us and other people collect data on the internet to make it better to measure it so All these dots are where these devices are. So if you look all the way up here Schwallbart There's actually a probe there and it we call these devices right path loss probes and this is what it looks like so Physical devices worldwide we have almost 10,000 deployed now And it helps people measure the internet so I'm not a hardware guy. So I I'm just gonna read this out for the people who are and Philip there. Can you wait Philip? He's the hardware guy who knows all about this type of stuff So ask him all the the hardware questions version one and version two which is these nice small devices Which we don't give out anymore our Lantronics export pro version three is the nice This nice box that I Just have here Which is based on a TP link? It does not work as a wireless router that functionality was disabled and There's something there's a one new box that we put in in like a stable places data centers And that's based on sucre's or was based on sucre's And we might in the future have virtual devices as well So but the main point here is I think they are everywhere these are measurements for a community by a community How it works is we give these devices out to you you plug it in wherever you Think isn't is a nice place to measure from and by that you Become part of this look of this global System and you can do measurements on all the other devices So this community is everybody interested in the internet routing performance, so could be you There's a URL there if you want to become a host My colleague Vesna who's also here Gave it gave a talk yesterday and was so successful in giving out probes. We have one left. So this is the only one left Other otherwise you'll just have to go there and Follow the instructions there on how to how to apply to get these devices and become part of this of this thing So there was episodes or there was the first part second part is something we developed on Top of the data that comes out of this. So now you have all of these devices that you can do measurements from everywhere So can you Measure what the internet in a country looks like based on on on on the data so What a call this project IXP country Jedi And it's measuring the path between these broke these devices in a country. So here there's two devices in or three A group of devices in a country and you just do a mesh of these trace routes in between So the trace routes will have the path that is taken and if you do a mesh it also has the back path And it will have all the intermediate hops so you can determine all kinds of interesting things about About that path for instance Two points in the Netherlands can we find points in the Netherlands Where the track where the path goes out of the country and comes back again because that's well self-optimal you would say So what does it actually look like are there any any interesting things we can see into in this and can we improve the internet by doing this and of course It's it's fun to Measure the internet like like like I'm doing right there And this is a prototype There's a this is a short URL where you can find all the information There's a github repo and there's the latest data that we have on each of the countries that we do this for So why the name IXP internet exchange point is everybody familiar with internet exchange who doesn't Who does know what an internet exchange point is? Okay Sorry Almost yes, so basically switch Or something far complex than a switch, but virtually switch where you where networks Inter interconnect and have a lot a lot of internet connect connection density. So the Well the thing that belongs with exchange point is they keep local traffic local and so this is a means to actually measure that Countries is an instance of local, but you can do this for anything So you can do this for all the countries in the Middle East or you can do it for a city Are is stuff kept local in a city or in a region or in a? In a language on a language boundary and Jedi Actually is because my son loves Star Wars, but you can also say that Adding metadata to this which is another part of the project makes It do some magic and I'll show the magic So digging into this for each IP we see in in the trace route we Capture geolocation data and that's tougher than you think maybe and network information Do we find these IXPs in the path do what autonomous systems are there? And as that that if you look hackers metadata metadata as a hackers best friend So problem with infrastructure geolocation is that it's hard This is just an example, and I don't know if you can see the IP address here It's a level 3 is a very big transit network If you look that up in any geolocation database you end up Here in the middle of in the centroid of the US And there's some nice articles how an internet mapping glitch turn to Kansas farm into a digital health There's actually a farm there that receives police visits Angry boyfriends Whatever Because that's the centroid of the US Whereas if you look at the host name that belongs to that it's pretty clearly says this is Amsterdam and this network is pretty Tidy in naming their names. So where people Think these databases for this are pretty accurate Which they are for the the edge of the internet where the users are for for where there's the infrastructure These can be wildly wildly inaccurate. So that's that's actually quite a problem for for doing this type of thing. So What we thought of is Crowdsourced this information. So now we have an interface where Experienced users can actually they see these names and and they see them in the context of this trace route and they can see Ah, that's in Amsterdam. That's all that's the name of my data center here. So they can actually feed us this information and There's researchers have done research on it and I've actually shown that the data that we have in here is far Accurate than the traditional geolog solutions. So solutions called open IP map. We're currently making a Production version of this because this is this is a quite prototypy at the moment So if you're interested in this type of stuff watch that name watch our announcements And of course it's made a source be with you it's on github It's in python and we run this monthly for each country where we have enough devices in the country. So This is a real way you can go to you can also go to that simple URL I showed earlier like I like below there So how Digging into the details you find the networks and IXPs in the country It's like the one or two probes per network We have networks like Deutsche telecom which we have three or four hundred probes in so you don't want all of these probes measuring to all the other probes. You just select one or two to get a representative Sampling and then you do the mesh between these so All the probes measure towards one and then next all to one so and until you have a a total mesh Then you combine that with the metadata. So that's where the magic happens and the output is like a geographical view What does the country look like a? Matrix view and a graph and if you have other ideas, this is open code So be happy to for people to actually contribute to this So this is a snapshot one of the monthly snapshots for the Netherlands a couple of months back And what you can actually see pretty clearly is London and Frankfurt. So that's where big exchange points are So Path are not completely staying in the country some people are surprised by that because there's very large exchange points in the Netherlands already But actually it's it's I'm being a politician here. I'm not telling the full truth because And it's probably the screen resolution Because if you zoom out of this, this is actually what that looked like and you just Didn't see this there was actually a path that we measured that went all the way to the US east coast and I Find it so weird I contacted these networks and they said hey Yeah, that's not what we are supposed to be doing But this is just BGP and there's not a lot of traffic between us. So we'd never we never noticed So that hairpin was fixed and this is something that People can use this to make the internet better like this I have a I have a colleague in the Middle East for instance who does this for all of the Middle East and Shows it to people that their path between their countries go through Europe through their big internet exchanges And that's now being fixed due to doing these types of measurements with these devices So we have a difference between IPv4 and IPv6 Well, that's a graph from a big exchange point This is the v4 traffic v6 traffic But what you can see here is there's a bit of a geek bias in into our Deployment because there's actually quite a lot of IPv6 Where we have this this data collected And you can of course see small differences between the two So this is a matrix view. So if you color These are the sources Of the measurements and the columns are the destinations of the measurements and then you have each of the cells You can color them by whatever you want and in this one I chose to color them by does it go through an IXP is the IXP keeping local traffic local That's so And the other coloring is based on does the path stay in the country or go out of the country. So the green is IXP Path stays in country. So that's IXP keeps local traffic local and that's pretty much the case in the Netherlands Yellow is it's local, but it's not via an IXP. So it's a direct interconnect or via some local networks The red is where the path go out of out of the country. So If you go to the URL, it's actually interactive so you can hover over you see the actual trace route So you can see if the geolog is correct you see as is and You can actually see what what the metadata That's included. So if stuff is going out of country and if it's a surprise for you, you can fix it and Yellow if if if you wanted traffic to go through your IXP and it didn't that's also something you can fix We also have a network graph review So here the dots on the outside are the ripet loss probes and any insight is where the networks are but this this one is quite complex So what all this graph now says is it's quite complex. I have a nicer example later on So now on to episode six the last one the hacker space is Jedi so Okay, you can measure countries, but you can also measure what's between hacker spaces. So First of all, it's where the magic happens, right? And well, there's a nice picture there of the Dutch hacker spaces so what if we would just put probes there and My colleague Vesna She has a project to actually get these probes into these in into the hacker spaces One thing that of course will happen is that they will be hacked on And we had a case where people actually did a responsible disclosure to us of vulnerabilities which were which were fixed There's a URL there for More on this project and and and on the hacker spaces Jedi And this is sort of the result of that of that effort right now There's when I took the screenshot, which it's not too long ago. There's 60 connected ones But if you do this for all of the 60 you'll just see Well all the path go all over the world. So that's well, that's that won't be a surprise. So I I'm not gonna show that one But what you what you can do actually is Figure What we need to figure out or to do these measurements is actually have them tagged or have them grouped in in some means So what we actually thought of is there is something called tagging of these probes That's so I have to dig into a little bit of detail on on how that works Here's a bunch of Tags that users put on these probes. So if you have a probe at a hacker space you can add the Hacker space tag to the probe so people will know this is a hacker space so they can do interesting stuff with them There's a lot of interesting Tags in there DN 42 There's universities that tag so you can do all kinds of Measurements between all kinds of different communities using these tags and that's actually What what the modification to the ISP country Jedi is that the the hacker spaces Jedi is we do only select probes with these specific tags So we select all the probes with The hacker spaces tag and see how they're actually interconnected and What you can also do is limit this by country because they're they're all all over the place So if you limit it by country, you'll you'll get are these hacker spaces. How are they connected? What are the networks in between? Does this traffic go outside of your country? So if you're interested in connecting up hacker spaces fear some kind of overlay or fear some some kind of experiments because these are the places where experiments happen what's in between and Of course things like Foreign nations surveillance these types of things that that I mean Locally you cannot escape that but if you can if you know if your traffic goes outside of your your locality That's that's potentially useful information to know especially If you're not Encrypting your traffic because everybody encrypts all of their traffic, right? All of my traffic. Wow So the rest of this is The same same type of methodology applies So I wanted to do this for Dutch Hacker Spaces and Then I discovered that These were not tagged very well, so we don't have them the users did not add tags yet So we've asked But then I thought what what what what can I do then? and then of course I thought of the Germans because the Germans are Gründlich are very thorough and actually I found that they they are They were quite good at Attacking their Their Hacker Spaces probes, so here's a visualization the the same geographic visualization of The German Hacker Spaces probes so in German Hacker Space you can see it Traffic or path will go through the Netherlands. There's a path through Denmark and there's also a path through Sweden, but that's actually because that space uses a VPN that ends up there, so And in IPv6 it becomes even more interesting because there's a path I don't think you can see it very well, but it goes to Warsaw or via Warsaw, which I find quite interesting But now the Geograph or the the graph of what networks what net what what Things are in between the Hacker Spaces becomes quite interesting I think Because on the one hand you have geography on the other hand you have ownership of the network and if you just look I've pointed out a couple, but there's far more for the German Hacker Spaces. There's a Russian or Ukrainian network I'm not sure there's US networks Swedish networks. It's it's it's all over the the place, so It's interesting to see that that actually happens and also interesting to be aware of it. I think So I'd like to conclude with some actions or some potential actions Putting more probes and Hacker Spaces. Yeah, unfortunately we have one left here, but if you If you're involved in a hacker space, please Consider having having a probe there and If you have them, please tag them and if it's not online, please get it online again because then we can do all these these types of fun visualizations for all countries or What you can also do is you have a bunch of points in a network That you want to do these same same types of visualizations between You just tag them with something and you have a new community of things that you can do this visual This this is all open code, so you can just run this anybody can run this And of course use the tools of visualizations Again that URL If you're a visualizer, I am not as you could see You can make far prettier visualizations far more interaction if you're interested in that I'd like to talk with you Another thing you can do is we have hackathons and you can hack away on these types of Of things so there's do or not do there's no try and That concludes this other any questions If there are any questions, please goes to the microphone that is also in the stream No, no questions Yeah, then another warm applause for the speaker and for the project at least Thank you