 So this presentation is from RIPE NCC. It has a lot of details on all these other tools RIPE NCC has been producing. But the one I'm going to talk about is RIPE Atlas. I should point out that I don't actually work for RIPE NCC. I don't even live in the RIPE service region. But I will greatly accept their slide. RIPE NCC is one of the five regional registries. Regional registries assign internet resources to entities in their service regions. There are five of those. RIPE NCC service region is Europe, the Middle East, and Russia. Other service regions are ARIN, APNIC, LACNIC, and AFRINIC. If you're in India, APNIC is the one you're most likely to be interested in. APNIC is the Asia Pacific Network Information Center. So if you have a network, a real network, not a home network, you need an autonomous system number and you need some IP address space. The people you get this from are human regional internet registry. Usually, you don't get them from the regional registry directly. You get them from a local internet registry. And it's the usual computer science tree with the roots at the top and the stem at the bottom. The RIPE NCC is a not-for-profit membership association that debates in Amsterdam. In addition to assigning internet resources, they also make some tools. One of these tools is the RIPE Atlas tool. I actually have an Atlas here with me and it bends in neatly in the background. It's not very big. The RIPE Atlas is a global network of probes. This is a probe. It's not a global network, which actively measure internet connectivity and reachability from different parts of the planet. The time these slides were written, which I don't actually know, they reached my mailbox a couple of months ago. The RIPE Atlas system was collecting just shy of 10,000 results per second, which is a reasonable number of results to collect. And these results allow the Atlas system and its users to understand the state of the internet in real time. The way this works is that volunteers host a probe in their network. And the more probes we have around the world, the better the Atlas system will be at measuring the state of the internet from anywhere to anywhere. The point of the internet is end-to-end connectivity. So the more probes you have, the denser network of probes, you have the better quality or the more valuable the results will be. If you've only got a probe in deeply organized areas, then you're only going to be able to measure those paths. It's very useful or a lot more useful to have probes in more distant parts of the world. Before the world broke, a bunch of us crazy people flew around the world carrying boxes of these things and explaining to customs that they're wireless access points. And we get volunteers to host them in their part of the world. Most of these probes are in Europe, as you can see from this map on the screen. That's just because that's where they were born. And several people have been distributing them around the world. The system subscribes to the gamification theory of modern internet technology. You earn credits for hosting a probe. You expend credits to perform measurements. So let's talk about these measurements. Your measurements can be a different type of measurement. You can do a ping measurement, which measures latency on the internet. You can do a trace route, which shows the path your packets take around the internet. You can measure DNS, TLS, HTTP, several different kinds of things. You can also set up alarms that will work with your own monitoring tools. So if you're running a service on the internet, say a web page or a chat system or some app or what have you, and you want to know how well the service performs from different parts of the world, you can use the Atlas network to keep an eye on your service. Say, I'm hosting this app or I'm hosting this web page. It's sitting on a server somewhere in Germany. And most of my users are in India. I would like to know what's the performance of this server or how reachable is the server from India? Can users from India even reach my server? Can they resolve its DNS? And you set up a right Atlas measurements to check from all the probes that are hosted in India to go and connect to your server in Germany. And the right Atlas system will tell you, right, latency is several dozen milliseconds. The trace routes, for some reason, goes to Singapore before going to Germany. And the DNS is full of errors. And you can put all of these details. You can put in your own measurement in your own monitoring system. And it will scream at you when things change. That's very useful. You can also test IBV6 connectivity because that's obviously the future of the internet. The probes themselves, I briefly talk about, they're very small. They're a couple of cubic centimeters these days. The older probes were a bit bigger. Currently, there's more than 10,000 of these things out there. And under ordinary circumstances, I would offer to throw them at you in the audience upon extracting a promise from you that you will plug it in and take care of it and actually host it. The current generation of probes is built around a nanopie with a bunch of memory. And I think they've all run out. Leah's just disappeared from the call. But also we're not actually going to do events anymore. So I can't give you a physical probe, but I'll come to software probes in a moment. Another potential way to participate in Atlas, if you have a large network, I'd say a data center, you could offer to host a naturalist anchor. This is a rack-mountable thing. And there's about 600 of those. And 100 of those are virtual machines. But I think those are no longer quite as relevant because the world is moving to software probes, which I'll get to. But the goal of the Atlas system is to get as many probes and as many ASNs as many autonomous systems. So as many ISPs as possible in as many countries as possible. India is currently reasonably well-represented in the Atlas system in terms of absolute number of probes, but in terms of population and in terms of number of networks, it's actually still fairly underrepresented. As I said earlier, better coverage means a better understanding of the state of the internet. The more probes we have on the more diverse set of networks, the better the end-to-end measurements are. There's, you know, we have plenty of coverage from, say, the US to Europe. We know just about every path from just about every meaningful ISP in the US to just about every network in Europe, but from a rural network in South Sudan to a rural network in Papua New Guinea, the coverage is not so great. Whether measuring such routes is meaningful or not, you know, it's open to interpretation, but for the, you know, you can't judge people on their measurements. So we want to make every kind of measurement possible. So in order to make it possible to continue to deploy probes, even without us traveling and also in places where we could never travel in the first place, there's also software probes. Unfortunately, these only run on Linux. So if you are running previously, you need to serve a virtual machine. It's all a bit fachy, but it can be done. And basically this brings the right Atlas probe into much wider potential deployment. If you have a machine somewhere which can run a Linux virtual machine, or if someone finally gets a number of round two it's together to port the thing to run natively on any previous machine, you can run the right Atlas probe. So if you've got a machine somewhere with space for a VM, you can spin up a virtual machine and install the right Atlas probe on it. And you can have a probe just as if you were in an event and someone threw a physical probe at you. So that's actually me out of slides. There's a bunch more slides in this presentation about other tools the right as you see folks have. I'll stop sharing the slides. But that was the end of the presentation. I think it's more fun if we actually go into some real live demo of how this thing works. So I will try to find the web browser window. If anyone has questions, you can shout them at me while I look for a web browser. Yeah, sure. If anybody has questions, feel free to unmute while... It's over. If you're feeling like you've been hit by a machine gun, can I share this window? I probably can. Share, screen, click, click. Okay. And I have to find the window. There we go. Right. So this is the right Atlas. Welcome to right Atlas. I've helpfully logged in and I have a little link to my Atlas and I can poke at measurements because measurements are nice. So if I wanted to... Well, these are all the measurements that are currently ongoing. Someone is doing all of these, all of these tests live in the system. That's all great fun. But let's try to create a measurement. Let's try to... For just for fun, let's do a ping measurement because it's the first one in the list. The target of this measurement, let's take my machine here, which machine shall I go for? Let's take the machine that's sitting in the room next to me, which is called Truth Ferry. So you give it the target of your tests. This one is a host name. You can take an IP address in there. You can take a host name in there. Doesn't really matter. I'm interested in IPv6 because I like IPv6. I'd like to run this test every 240 seconds. So it's just the defaults. But actually I'm going to configure it to just run it once because I'm not interested in this test more often. So I'll send... Well, actually I'll send 10 packets of 64 bytes because I like nice round numbers. Where would I like to ping from? Well, I don't want to ping from worldwide. I'm actually only interested in India. And let's use a wizard because we like wizards. And where is India again? It's over here now. India is over here. So I just give me all the probes in India. I can do this. There we go. I think I've selected a little bit more than India. Let's go for greater India, shall we? That's the kick-tie of India for life. Right. So all of these probes in India and other Indias are going to run a one-off measurements and just run it right now. Create my measurements. Yes, yes, yes. Now let's look at my measurements. So it's instant gratification as things should go. So it's now running. Click, click, click, click, click. So let's look at my measurements. And of course the demo effect is real. If I click the measurement, nothing is going to happen. Yeah, I've made it invisible. Okay, I'm not very smart. That was not to show me the measurements. That was high by measurement. I have now lost my measurements. Clear my filter. There we go. It's still running. Well, I could see it in real life. But this is going to run my measurement and eventually it's going to return results. So instead of waiting for this, let's look at the previous measurements, which might actually work. Okay, it doesn't want to work. I don't know why this doesn't work. It doesn't work because it's a demo. But things never work when you're doing a demo. So I'll stop sharing this screen and listen for questions while I try to debug in private. Right, sorry about that. Okay, there are a bunch of questions. Nemo has a question. Yeah, Nemo has a question. I'm here hanging out on YouTube, which is, is there a way, is there an easy way to get probes on cellular networks? Most of India connects via 3G, 4G networks now. That's two questions. Is there a way? Yes, is there an easy way? No. Swapnil, a couple of days ago, showed me a way to do this. And I'm going to put up his webpage if I could find it again. Yes, there we go. I can do that. And I can share this screen. If I can find the appropriate button again. There we go. Here's the button. There we go. Swapnil would have showed you this webpage. So he has managed to make a right pathless probe connects to a geo widget. So he's got the same little black gadgets I have. He bought a little rotary thing, a GL MyFi gadgets connected the whole thing together. And that's how he runs measurements locally. The thing to be aware of is that many of these mobile networks, you end up paying for bits and the right pathless measurements, you consume bits. So you probably don't want to set up a postpaid plan and then get a bill at the end of the month for the value of your house without all your neighbors. You want to use prepaid plans so that when you run out of bits, you can decide whether you want to invest in more bits or whether that's enough bits. So I think Swapnil does a prepaid thing and every now and again, he goes out and gets a two gigabytes or five gigabytes top up for the SIM card he's using and that works. So you can hack something together. You could, you know, that's with a physical probe. You could presumably also hook up a software probe to the same kind of gadgetry, but it just becomes a lot less portable. This with a five, with a last generation probe, a little MiFi gadgets, you can still walk around with your mobile measurements if that's your thing. Does that answer your question? Yeah, I suppose so, but let's see if Nemo has anything more to ask. Trouble usually have a demo, right? Sure. Yeah, it's not working because it's a demo. If I click the button, the button says no. Well, the button doesn't say no. The button just doesn't do anything. So it's still running. Well, it claims to be running my tests, but it's, or my measurements, but the results are not visible and the results of all my previous measurements aren't visible either. So one of the faces in this call mentioned that they have been doing some measurements for a while. Maybe their measurements work better than mine. I think that was Gurschen Butz who said he had some measurements. Hi. Yeah. I mean, at CS we've also been running the Atlas probe for a while. So if anyone's watching and wondering if this requires too much work, I've only received two emails regarding that in two years. So it's not a lot of work. And what you get out of it is a lot. As Trouble was saying, you can do things for yourself maybe and there are blog posts on the website that talk about these things, just network connectivity in general. And you can, to me, Atlas is very customizable. You can run tests repeatedly. And therefore, I mean, check whether your ISP is blocking a particular website that you can connect to it. You can catch your ISP doing other shady stuff, including like throttling connections to a particular websites, et cetera. And as a researcher at, so I work at the Center for Internet and Society and we've been also doing some work on network measurements and measuring web censorship in India, which I'll come to in a bit. But these measurements are awfully useful for research as well. So on using the data, you can detect internet outages which may be intentional or otherwise. Routing the subs that keep happening regularly. And just generally because internet traffic has all gone up because they're all stuck at home using the internet for everything. So I guess there are some insights to be drawn there about whether there's more congestion network delays, et cetera. And you can find great examples at the RIPE Atlas website. And what I'd also like to talk a bit about is the larger landscape of measurement tools. So you have the RIPE Atlas which does a variety of measurements. I think it's capable of doing the maximum number of measurements out of the landscape, but there are specific tools for certain things. So you may have heard of the open observatory for network interference, UNI, which is perhaps a go-to tool for measuring web censorship. So it tries to connect to particular websites that may be blocked in your country and their data set is also open and that is available as a mobile app and desktop tool as a CLI tool, et cetera. And similarly, there's the Center for Applied Internet Data Analysis, which is Kaida and that's primarily aimed at measuring internet shutdowns and outages. And just before this talk on the event page, I saw a very interesting comment from Zainab which is about a blog post Kiran had written, I think around two years ago. And it says that I think Kiran and others wanted to build a censorship detection service software for India. And I think that comment spoke to two things. One is about accessibility, which is a concern Nemu has already raised. Like since most of internet users in India are mobile phone users, how do we get these tools in their hands and get more measurements from areas that we haven't seen? So for example, in any such tool, I've seen that representation from smaller towns or villages in India is abysmally low and where mobile phones are largely more popular. And the second thing which Kiran's comment spoke to was about how measurements can be used to make arguments that extend to the legal and policy world. And that specific comment was about net neutrality. So in fact, you can use the right parallel tools. You can use uni et cetera to find out whether your ISPs violating net neutrality. And this is a point that also is made by Fenmick McKelvie in his book, Internet Demons, where I think he's also calling for more network measurements to make these policy and legal cases. And this is something we've been also doing at CIS and recently we've launched SensorWatch, which is available as a mobile app and it's on the Play Store and you can download, it has an APK et cetera. And what we're trying to do is run measurements from mobile devices and they try to connect to a list of websites that are potentially blocked in India. And what we've seen in the recent past is that different internet service providers in India are blocking entirely, widely different websites. While they're only supposed to block websites when the government tells them to, they're blocking websites arbitrarily out of their own volition and sometimes it's not clear why, are they not complying with orders properly or is there something more sinister going on as in they want to block certain services because of business reasons et cetera. And really, and I mean, some of you, including I know some people from Hasgeek who are involved in the fight for net neutrality in India, which legally one may think that we have one, but how do we ensure that the laws are complied with? And this is a monitoring problem and this is squarely where such network measurements come in because in our recent tests we found if they're blocking widely different websites, then net neutrality is being violated currently in India. So therefore, even if you have it in law, you do not have it in practice. And by running these variety of measurement tools and I also encourage you to of course, check out sensor watch which we at CIS have released recently and you can help us sort of monitor net neutrality violations, web censorship in India and possibly make a case for the telecom regulatory authority or et cetera to start monitoring on net neutrality violations and otherwise. Thanks, thank you so much for this. I think Trubble's got his demo also working. So this is- Well, I haven't got the demo. I wanted to do working because web pages are evil and never work what you wanted to do. But Ripe Atlas also has a command line interface which is written in Python and you can find it on GitHub somewhere. I'll find the link in a moment. But there's not a CLI called Ripe Atlas and you can just tell it to reports on the results. So this, if you recall from when I shared my screen earlier, this number is the measurement ID I've just set up. So if I just run this, remember it was an ID of six ping from everywhere in India, from any probe in India to the machine in my kitchen over there. If I just, you know, please spit out the results. This is a bit painful to read. Maybe if I make this window wider because IPv6 addresses are very wide. This might make a little bit more sense. Right, there we go. Let's pipe this into more because there's more data that I'm willing to consume in one go. Bites from a certain probe number, probe number 100278, which has IP address mumble to, this is the machine sitting in my kitchen, has succeeded, typed a little 50, and these are the ping times, so the latency is about 100 and, you know, I grep this is 110ish milliseconds. This probe has an IPv6 address and can resolve tooth fairy, which is in my kitchen to this address, but it can't actually reach it. So it has, presumably it has no IPv6, excuse me, to route to me. Same thing here, same thing here. For some reason, this guy here, probe 29814, takes twice as long to get from India to Hong Kong as probe number 100278. I don't know why this is. You could then go and do a trace route to find this out. Speaking of trace routes, a couple of months ago when I was last doing this presentation or one like it, I think I was handing out things in February. I don't remember where I was in February, somewhere else. I did a trace route measurements from various parts of the internet to somewhere else and then I get the trace route outputs if I tell the thing to report on this result and then I could go and analyze what routes are these things taking from different probes and that's all very useful. So trace routes and ping are just two measurements you can do with the right battle system. You could also do HTTP requests. You can do TLS requests. You can check NTP coverage. There's several other things you can do with the right battle monitoring system and you can get the results either through the web page if web pages work for you or you can programmatically get them using the right appless toolkits which is what I've just done now. And in my experience, that's just because I'm a troublemaker, the command line works a lot more reliably than the web page but that's just because web pages got it for me.