 Hey, hello. Thank you. Thanks for the intro. You've heard from me before, like, 20 minutes ago. Anyway, the details of my life are inconsequential. So let's talk about what I've been working on, though. And for the last, you know, maybe a year and a half now, I've been playing around with a lot of experiments that involve HTTP2. At the beginning, it was still speedy, and slowly it sort of evolved into something a little bit bigger that I now call Common's Host. And it's sort of a CDN, and it's pretty cool to be talking about it here at Cloudflare. I hope there's no offense taken. I mean none. And let's get straight into this and why I'm calling it like hybrid and all this kind of stuff. So anyway, if you follow the Cloudflare CEO's Twitter, like you should, he talks a lot about this really hot topic today, which is Intel versus ARM. All right? Okay. Anyway, so Intel and ARM, two architectures. And what does that mean when you're running servers, right? Two different CPU architectures. When you're paying for servers, and for those who are not, you're still kind of are using those service if you're routing your website through them, right? And so we're consuming all the energy. And the big selling point that is being highlighted right now is that ARM CPUs use a lot less power than Intel CPUs. And that's kind of interesting. I mean, if you're running a service like a CDN, like Cloudflare, you're probably switching over to ARM is kind of an interesting thing. So Matthew Prince, the CEO of Cloudflare, sort of pointed that out recently. And I was very interested in that because, you know, the numbers don't really lie, even though that is probably a lie because that's some kind of sales pitch at a sales convention. So, you know, grain of salt and all that. But still, there's something to it, I think. And I was, when I thought about that, I felt like this sort of like a deja vu, right? Sort of around the time that this movie, The Matrix, was created about 20 years ago in 1998. This thing came out. Ooh, cool, right? And that was not retro at the time. That was like state of the art design. So, you know, these people came out with this really cool idea that you could search the entire web and had this, everyone knows a story, right? The page rank thing. And, you know, they've kind of improved on everyone else. But one of the things that I remember from back then, and I'm dating myself a little bit here, is that it was first hosted on this cool system here. And in fact, these two boxes here, those were the ones that ran Google search, okay? You know what Google does today in terms of scale and servers. I mean, you can imagine. We don't know. The crazy thing is that these were commodity hardware at the time. This is what everyone was running at home. You know, this is what you would play your games on. That's what I was doing. Intel just donated them just to see what they could do with it. Because it's like a cheap thing. Nobody seriously is going to run an actual company on, you know, this desktop, you know, mid tower thing. But they did and it worked. And then, you know, IBM was like, wait, wait, wait, wait, we want to be part of your company because we see you taking off. So they donated some, you know, fancy boxes, all the ones that you see underneath the table. And that's sort of, I guess, where they ended up staying. They ran some ancillary services on those, but it's too late. The commodity thing happened. The big idea of Google was not just a search thing, but it was actually that you could scale giant internet services on commodity hardware because you need to go for the best performance per dollar. You don't care about the most powerful servers in the world as a single unit. You care about what is most efficient to run as a business. And so if we move forward back to today, I think that the commodity hardware of 20 years ago has sort of changed and has now sort of become the, you know, maybe a little bit of what the enterprise hardware at the time was. And so I feel like there's sort of a, there's an impression right now, the Intel versus ARM of this Goliath and this David story. Everyone who's familiar with Malcolm Gladwell, you know, the David versus Goliath thing, things are not as they seem, right? You think, you know, Goliath, the big guy, you know, the big warrior with the armor and the weapons is like super strong and cannot be defeated. And you think the little guy is going to get squished, but actually, you know, he's got the technological superiority, this thing here, this little sling turns out that has about the, you know, the kinetic energy of a handgun today. And, you know, he was, you know, the big guy had no chance, right? So right now Intel sort of feels like there is huge, you know, monster CPUs, these Xeons, 24 cores or what have you, you know, king of performance. Whereas ARM, you kind of think of, you know, that's the stuff that you have in your pockets or your hands if you're not paying attention. And it doesn't seem like a fair battle, but I like to make the case that it totally is. And when I look at today at the sort of the best performance per dollar, I think of all the phones up to last year. Okay? Because, yeah, nobody wants to buy the last iPhone, the last Samsung, the last whatever, right? They're gone, they're ancient history. But the thing is, that there were so many made last year of those CPUs that the left over capacity, the left over volume that was produced in these chips is probably still a really significant number. And they don't really go bad because silicon is sort of nicely protected in a dust free case and whatever. And so this is a screenshot from Alibaba, Aliexpress or whatever. You can buy these chips by the thousands of like any of the old, you know, Samsung phones or whatever. You can just buy them by the thousands. And this gentleman here who's not paying attention is, everyone knows is Mario, if you know anyone who's a really cool guy who organizes this amazing conference called Frost Asia, which just happened last month. It'll happen again next year, should totally attend. And this person here is called Bunny Huang. And he's amazing. He tells amazing stories about the sort of the hardware market in Shenzhen and what's happening there. And one thing that really struck me during his talk last month was that he goes into these little shops, these like individuals, it kind of looks like someone, you know, okay, you go into these little shops and they sell just reels of components and just like massive numbers of chips. And he was adding up like how many are there. And came to the conclusion that like a single one of these shops would have more like in their little shop than like you could find in all the American, you know, the distributor websites that you could order like in the entire North America, right? It's ridiculous. And it's not because they're like they really love these chips. It's just because they're really good traders and they knew where to buy the cheap stuff and then sell it to someone else who's got a project. You can go into these things with a bunch of cash and buy like a million chips. So there's a lot of these things still available, but they're all kind of like last year's performance stuff that is now dethroned. But in terms of cost efficiency, I think it really makes sense. And I've been looking at these things from a perspective of how do we run them? And what would it, you know, what else can we point out? Anyway, so if I look at servers, desktops, how many of them are being produced and shipped every year, you see that servers is kind of going up desktops, I think obviously it's going down, mobile phone, kind of exploding, now tapering off, of course, turns out there's only so many people in the world. So if we look at the actual numbers across a couple of different categories, there's about 11 million servers every year. Sounds like pretty cool, right? Let's assume that they're all Intel, right? Desktops are a lot more, though, and that's why we started with the commodity hardware. Desktops was a commodity and then you take from that market and you sort of create a high performance version of that and that's why servers made sense to run on Intel. Laptops sort of taking over now. Tablets actually bigger than laptops, that was surprising to me. But the real kicker is that phones, right? Okay, it's just ridiculous. 133 times more phones than servers. All right. Now, tablets actually, but definitely phones. Most of them are going to be running ARM. What's ARM was a stand for? Acorn, RISC, machines, you know, I like emojis. Anyway, it turns out Acorn was renamed to advanced. It turns out that that's the chest not the chest. So let's just call it ARM. So most of the mobile processors like the phones and even most of the tablets, just going to be ARM. So we all have ARM processors. In fact, we probably have a lot of them. There's been over 100 billion of these things produced. 100 billion is a big number. More than a number of people, right? Weird. So you find them everywhere. And the reason you find them everywhere and the reason there's so many is it's not just like this company ARM that produces them. There's a ton of companies that make them. And they actually look at these logos and this is just a small selection. A lot of these are actually competitors, you know, like Samsung, Apple, NVIDIA, AMD, you know, whatever. So how does that work? Actually, it turns out ARM doesn't even sell these chips. They license stuff. So two things primarily. One is the instruction set that these chips run, right? Current generation is ARM v8, you know, but they've been doing this for a while. So that means that you can create a chip that actually implements their instructions. And that's what, you know, sometimes like Apple, like if you have, you know, they'll talk about like their custom chip chips, but they actually just implement this, you know, not just, okay, it's a really amazing thing. But they implement this standard, you know, instruction set. And that ARM also licenses sort of was on the next level of abstraction would be like the actual cores. And so if you're a system integrator, and you want to, you know, create a video playback device that can do 60, you know, 60 Hertz, 4k HDR, right? You have a certain performance envelope, maybe a certain power budget, you know, you can talk to your arm and get like, okay, you need this many of these specific cores, and then go off and have some of these fabulous chips. And so that's, that's sort of their business model. And that's why there's so many companies able to produce these things in such volumes. Now, risk actually used to mean that these were very simple chips that were very low power and couldn't really do much. And you had to like write everything in software, which is when the pros and cons at the same time. Now, these days, they've actually kind of come closer to what Intel offers with a lot of, you know, very specific, very powerful and energy efficient instructions. And even in the latest generations of like the ARM V8 that I mentioned, there's hardware accelerated crypto stuff. So really cool for all of us who are rep performance people and, you know, want to use HTTP2, because that's the greatest thing ever, and future quick, and all these things are requiring 100% encryption of all your traffic. Otherwise, the presages will work with it. So you need to have, you know, these efficient hardware accelerated crypto stuff in all of your devices. Otherwise, you can consume too much battery or it's going to be slow. On the other end of the spectrum today, there's also like these really powerful ARM processors. Those are the ones that, you know, in the first slide that I showed, those are running servers. So those might have, you know, 20, 40, whatever cores and they can take on workloads that are equivalent to whatever the Intel Xenons can do today at reduced power budgets, right, because they benefit from all that shared investment into the technology of that platform. So let's get back on to what we're actually here for, which apparently is now the CDN meetup, because all the three talks are about CDNs. I will do a little comparison. First, I'll explain like sort of how a traditional CDN works. And this is, I don't need to pick on Netflix, they do amazing work and they're a really great service. It's just, as a reference point, we can all relate to that, because most of us have probably used it. They're extremely successful, they serve about a third of the traffic in the United States for many years at their peak levels. So that's huge, huge, huge scale. And I want to compare that to something that I've been working on, which is a little close to that level, you know, and it's called Commons Host. And what that is actually is World's Smallest CDN. Okay, this is an open source project that I've been working on. It's based on Node.js. Last year I worked on the Node.js core implementation of HTTB2. That's all open source. And since then I've been working on like a little web server built on top of that, some front-end tooling to basically be able to deploy stuff. And I'm now at the point where I'm actually even deploying these little servers, you know, sort of around the place, starting to, and that's all running this open source stack, which I think is kind of a unique thing. The reason I'm doing this all with open source is I hope that other people can also deploy these things. And together we can create a network that is sort of the largest CDN, right? I think it would be kind of hilarious to use these tiny little servers, these open source scrappy little things made by some random bunch of people, and we become like a really huge network of CDN servers. So that's sort of what I want to do in a really short amount of time, because I can't really afford to work on this for like ever. So we need to make this happen really quickly in the next couple of months or a year or whatever it is. You know, if you look at other CDNs and how long they took, I want to accelerate that. And so I have this thing. So, and just to prove that I have this thing, I've brought it along. And this is a CDN server in this box. So I'm not hiding. It's all open hardware as well. It's a company called Odroid. It's a Korean company. And they make these little sort of NAS servers, like these little network-attached hard drive things that you put on your network for backups. Sorry, could you help me? Okay, thank you. Okay, so I'm very proud of my baby here. Okay, so this is a server, a networked ARM-based tiny little server. And it contains an SSD. It has a gigabit port and a power connection. And that's really just about it. And this is what I've been using to host the commons.host website. So if you go to that, you're hitting something like this that's currently at my house. So anyway, you can pass this around and show it around. Please don't break it. It actually belongs to this gentleman right now, because I'm selling the name, but I'm not that that's the part of this talk. But you know, just he's a cool dude. He'll find you. So anyway, so if you look at the specs of these things, completely different ballgame. What Netflix has is called the open connect appliance. This is an initiative that they took because they used to be running on a third party CDN. And I guess at some point, when to hit certain scale, the fees of these things become really expensive. And you start thinking, what, what, what couldn't we do this ourselves? So they came up with this thing called open connect appliance. Now, while it's called open, it's kind of hard for me to find information about it. So caveat, this is really just sort of this is based on, you know, all kinds of articles that may or may not be technical or publicity or whatever. But, you know, so again, not exactly picking on Netflix, because I don't really know what they do. But you can sort of imagine an eight core Xeon, whereas this thing contains an eight core, you know, Samsung X and S and X and S is what's in their phones, what's in the Samsung Galaxy phones. In fact, I think this is literally from, from a Galaxy S5 from a couple of years ago, or Note 4 or something like that. So this is not like state-of-the-art, but it gets the job done in my experience. Netflix, they talk to you once you do more than five gigabits. Who's, who's it that they talk to? They talk to ISPs who have a lot of customers watching Netflix that cost a lot of money for both of the ISP as well as Netflix. So at a certain level, they go on like knocking your door and say, you should put this little pop, we'll pay for it, don't worry. And just connect it to your network and give us a certain amount of bandwidth and about, you know, five to 20 gigabits I think is roughly what they serve on these servers. And they consume about this much power because that's what you have to provide them. They have a bunch of, you know, hard drives in them, like 30 or 40 hard drives, offering about 100 to 200 terabytes of storage as far as I know. And I'm guessing that they spend about 10 to 20,000 dollars on the hardware, the build materials. This guy, Ordroid, it's the Korean thing that I mentioned, right? It's got a little tiny 8-core CPU from, from basically a phone. It's got a 1 gigabit port. That's the big field because people might be like, oh, what about Raspberry Pi? When you just use that, well, Raspberry Pi doesn't really have a good network connection. This thing is slightly better in that, in that respect. 20 watts, compared to 750 watts, that's a big difference. But this is, this is essentially not using much more than an actual phone. So very, very powerful, very cheap to win. Contains one hard drive, not 30 or 40, one hard drive. Extremely cheap. And this is not a sales pitch. I just want to, you know, compare the numbers. This is ridiculous. There's orders of nine to two's difference. And you can tell just, but the vast majority of the money goes into an actual storage and the size of what you want to host. So sort of the numbers that, you know, that I want to use. Oh, and by the way, they're very cute. You can actually stack them, which I think is really cool. And if you let your imagination play, oh, yeah. So when you stack them, you should use something like ARP, which is a low-level protocol, which means that you connect like a bunch of them to your, like, router, and they all have like the same IP address. And so they kind of load balance automatically. So it's quite scalable, if you think about that. But if you really think about that, then how far can you go? And this is too slow. Okay. Okay. What if you did this, so if somebody has done this, somebody with about 200 of them, these are slightly modified, you can also get them without the hard drive support. So this is purely for compute cluster, like say you want to run like a joint Kubernetes, but I figure about $30,000 I can get 200 of them to look together and make like a giant 200 gigabit per second server. I'm not sure what I'd put that even into, because I don't have an inter-exchange at my house. I'm like, maybe a car for people could help me with that, probably. But I think that that's maybe not the point. Like I don't want to build a giant mega server in one place when a CDN is all about serving lots of places, right? And by making them really cheap, I can split them up more easily, I think, right? And the way I look at it is you should track where people live with a CDN. So the closer you can get to the most people, the better it should be. That's sort of what I'm hoping for. So if I got 200 of them and put them everywhere, how does that compare to everyone else? So I kind of looked around. And again, these are not exact numbers. And I love all these companies. I'm just trying to sort of get a reference for validating or invalidating my own ideas and hypotheses. Now, most of the CDNs have like a couple hundred, dozens to a couple hundred, Occamara has a really pretty huge number. Alibaba, like somewhere in China, they have ridiculous numbers for these servers. I don't know what that means. Maybe they just mean like, that's how many computers they have, not how many actual locations. I don't know. So some of these numbers are a bit weird. But mostly they have a couple hundred. And I don't think that's nearly enough because there's about, you know, 560 million people and, you know, or 500 something cities, sort of cities that have over a million people, over a thousand, you know, urban areas like, sort of like cities, agglomerations, whatever you call, over half a million. And BGP active ASNs, you know, that might be a little jargon-y. That means that an ASN is sort of a unique number. It's been an autonomous system number. That means every single network on the internet has its own unique number that they've registered. And BGP active means that they're actually announcing routes to their network. So that's probably the number of maximum, the highest number of ISPs in the world. Now, maybe it's a lot less than that. Maybe I think some estimates I've seen also like around 10, 20,000, but there's like 10, 20,000 ISPs or computer networks or universities or whatever in the world. And if you have a couple of hundred pups, you're not really close to all these networks, I feel. So my idea, I was like, what if these, you know, data server racks, you know, each one has like about 10 gigabit, right? And storage all the data. What if I'm more like that? This is literally my house from a few months ago. What if they're, you know, they look kind of similar to me and they all have a lot of same connectivity, they have power, they have air conditioning and all that kind of stuff too. So there's this thing called fiber to the, right? Fiber to the, you know, neighborhood or to the, you know, to the curb, to the building, to the home. In Singapore, most of us have fiber to the home, FTTH. And this is sort of a really interesting game changer, I feel, in Asia. And I wasn't even aware of this. This is sort of like, you know, my random idea, like when we turn our HDBs into data centers. But then I started looking into this and it turns out that the rest of Asia also lives in apartments. And it turns out that, actually, this is really cheap to deploy today. So if you look at, look around Asia and you see all those cities that are being developed right now, they're all building fiber. Nobody's putting in DSL and cable anymore. They're putting fiber because you go to China and you buy them bulk and you just deploy, it's dirt cheap. And you get infinite capacity essentially, right? Because you can upgrade this for decades. And it's pretty much everywhere now in, you know, China, Japan, Korea, or Southeast Asia, India. You look at all those cities and you can like just search for like a fiber ISP and you'll find the memos everywhere. Doesn't mean that everyone's using it. But sort of this, I feel like this is maybe like a San Francisco sort of image of Asia as everyone's like struggling on 2G feature phones. That's not really the case. That might be the case on average in comparison. But there's a lot of places that are really, really developed in terms of infrastructure. And I think we are living one of those as an example. And everyone's copying that obviously. So you can get fiber in almost every city in Asia, right? Most of the world's fiber sub fibers are in Asia. That was a quote from about a year and a half ago. It's only getting better. It's pretty impressive. Again, this is new to me. Let's look around the world a little bit. There's this really cool map that I just wanted to show for no reason. If you look at Africa, things are a little bit weird there, I think. I don't have them really visited much. But the reports that I'm reading sort of marketing analysis stuff, they still talk about heat and dust being problems for days. And they're like, whoa, OK. This is not the same. So I think these kind of places are what you formistically describe as requiring a more pioneering mindset. Sure, yeah. Now, that's going to be hard. But I think the way to solve that is again with even lower cost stuff. Not saying we're going to deploy hundreds of billions of dollars of infrastructure overnight. You start somewhere, right? And I feel like the best place to start is what they've already done, which is put hand phones in everyone's pockets. There's more hand phones in the world, especially in those places, than there are toothbrushes and toilets combined. Right? That's insane. OK, so I think that this sort of solution would kind of work in this kind of environment as well. And I looked into this a little bit further and found that the number of data centers per capita this might not be 100% reliable, accurate data. But I think it's sort of all of it is inaccurate enough to compare. I don't know. If you look at per capita data centers, India versus Africa, similar population, but four times more data centers than India than all of Africa. South Africa is doing pretty well, though. Good job to them. Singapore, though, is just completely insane. We have way too many data centers. I mean, yeah, more data centers, right? But the members are just kind of amazing. So we want everyone to be at Singapore level, which is probably what most cities in developed countries approximate. I don't know. We might be like number one. Seriously, this is crazy. But you want them to be somewhere here. And if you have this few, this is a ridiculous number of people per data center. This doesn't work, right? You need to put servers not just in a data center, but everywhere else. And then slowly you can develop towards having sophisticated infrastructure. And they quickly went through, again, our comparison. So right now, a typical CDN that provides this cloud where you just send the order traffic to and they magically handle everything, they go out and negotiate peering or transit. That means that they figure out how to send bits around the internet. And they might pay for it or they might not pay for it depending on how they negotiate that. You have no clue what's going on with your data. They take care of it. It's a nice abstraction. It's very comfortable. I use it. But I'm trying to build something that you could maybe deploy your own CDN like I'm doing right now. I'm trying to host my sites on my own computers and share computers with my friends. And we can agree with each other which ones we host. The technology to actually coordinate that traffic with a traditional model is called Anycast IP. And there's only so many registries that give you an IP address. You get it from your ISP, but your ISP buys it from a registry. And there's a few of them and they're very regional. So if you're in Asia, you get it from this one. If you're in America, you get it from that one. And they use this protocol, again, BGP, this word that came back from earlier, to manage that. And again, you have no insight in that. Because that's just announcing, hey, here's my IP address. All my peers that I'm connected to, you can send traffic to me. Whereas what I'm doing is called Geo DNS. And that's available from anyone. There's a lot of vendors out there. On my GitLab, you'll find a list of dozens of them that you can play with. They're very cheap because of that, right? There's so much competition. And you kind of have total control over that. You can say, this domain should point to this IP address in that country or on that ISP. It should point to that IP address. And you can sort of manage that yourself. Like I said, everything I do is open source, whereas typically most CDNs are going to be proprietary. And that's potentially for security reasons. So I have to come up with actual stuff that is proven as well. But I think it's nice that you can play around with the code. And other people could modify stuff. And in terms of, I don't want to talk too much. This is not a commercial pitch. I'm sorry. Now, one thing that scares me a little bit about CDN, bringing stuff on a CDN, is that sure you get a medium benefit for DDoS protection because of the huge scale that they provide. But it's sort of a managed and middle service that I'm always a little bit uncomfortable with. And I think some people have those reservations when, for instance, your cloud provider gives you a certificate. That's cool. But it also means that they can spoof whatever you do. Now, of course, their business depends on not doing that. But what if other people snuck in there? They become a target. So there's all these things that you want to think about. I don't want to sound too fuddy about it, but it doesn't need to be that way. In fact, I wanted to share how another network, probably one of the largest networks in the world, works. And it's this thing here. It's RIPE. There's one of the registries. I think they handle Europe and the Middle East. RIPE Atlas is a project that they have where they're non-profit. And they just send out these little rebranded TP-link routers with their custom firmware on it. If you ask for them politely, they'll send you one for free. You just plug this into your house and you join their network and you can then run sensors, or what do you call it? You can run tests, performance tests. You can ping people from around the world, basically. And because you contribute one to the network, they let you use other people's probes from around the world as well. And everyone contributes something. It's a total non-profit situation, but it works pretty well because they have over 10,000 of these things around the world. I've got many in my house. And I'm actually, you know, anyone here who wants to run one of these things? Okay. Do you live in a cool place that's not Singapore? Like, there's loads of these in Singapore. And I really want someone who puts this into like, some kampung in whatever, some place. Yeah, we could talk, we could talk afterwards. Let's, let's, let's. I promised it to someone else before, but I forgot their name. So if you are here, you have dibs, right? So sorry. If you're not here, too bad. Anyway, so I think, I think that this is, this sort of, my point is that this proves that it can be done. People are genuinely nice and put these things out everywhere. And it works, it really scales. It's a really cool thing. You should check it out. So again, I go create the largest CDN closest to the most people in the shortest amount of time. And all because of these tiny little things. And I wanted to ask for some help because I've been doing this for about a year and a half now on my own. And it's, it's, you know, it's pretty cool, but it's also hard. I need just people to use this. Okay, so it's free. Okay, you don't have to worry about the charge of anything. You, you just publish a website just like you would it on GitHub pages or Netlify or Surge or whatever, all these other things. You just post your static sites there. That's it. Do that, please. And as you do that, you'll probably find that there's things broken. So please tell me about those things and I'll record them as bugs and we can, you know, have a discussion and fix those. If you have any suggestions for features, you know, I'll open to that. In fact, there's a, there's a GitLab page. If you just look me up, you'll probably find my GitLab repository where you'll find out there's already a bunch of stuff planned and being discussed right now. Also, if you're interested in hosting one of these things, again, if you live in Singapore or other places, Vietnam, right? You want to host one of these things. Actually, we have another gentleman here who wants to run one in Vietnam, but the cool thing is we can put lots of these everywhere because there's a small area so probably need lots of them. I want to put these everywhere in the world and I've already talked to a lot of friends in various cities around Southeast Asia from mostly other developers that just want to provide hosting to their friends in their city because let's face it, these are not the places with the best infrastructure and this way they can contribute slightly to alleviate those performance issues. You can have faster websites in every city by just putting one of these pops there. So I am also not fabulously wealthy, unfortunately. So like I pointed out, they're pretty cheap and I feel like if you're looking for a company that has visitors from any of these other countries and you're finding it hard to actually reach them with your websites and you want to just fix those performance issues really cheaply, I'll help you set them up everywhere you want. I'll go and send them out or I'll go in person if I have to. So I think that we can sort of match hosts and funding people that are really nice and solve this problem. We'll figure it out. And lastly, just tell people about this. I'm quite proud of this project and I would love people to actually find out about it and use it. And that's what I wanted to talk about and I think we should have these huge ideas but really just start small and stay focused and keep working on it. Thank you very much. Okay, thanks, Sebastian, for an interesting talk. Does anybody have any questions? Yeah, go ahead again. So it's actually just last week that we spoke to Akamai. So we were talking about a partnership for Indonesia and they have 1,437 POPs points of presence in Indonesia, 137,000 globally. So yeah. Yeah, we need like- I'll update my slides. Yeah, so it's either like join them or like really everyone has to join to kick their ass. So it's not exactly very, but yeah. So. Thank you. Anyone else? Okay, I have a question if everybody's still thinking. So how do I get started? I went to the page right now, commerce.host slash signup and it loads a blank page. So I think this is a bug report, actually. Yeah, but so I go and buy one, I sign up and then NPM install, I saw instructions and then what happens? That's right. So actually right now to just deploy your site, you just use an NPM tool. So this is very easy to put it into like your Travis config. So just like you deploy to any other web server, it's a one-line command thing. You use a signup on the website that I'll check why it's not loading. It could be a Firefox issue as well because I'm using web components. And maybe this is more for the front end of, but Firefox and web components don't get along. It actually works really well in Safari and Chrome. And I'm doing this to sort of make a point that they should just implement it. I'm actually using a Firefox beta browser on the mobile phone. So that might be exactly the reason. I'll double check. Okay. Okay. So just to finish that, I actually last night, so I had a really nice friend here, a really good friend, Kenny, thank you very much for sponsoring the first server. And that's possible. It's still around, nice. In one piece? Good, good. So, like I was saying, I can't find all these things myself out of pocket. So I actually put it up on Carousel now. Turns out people from the rest of the world can't use Carousel. So anyway, I might change that. But if you want one of these things, just check out Carousel or tweet me and we'll make that happen. You can put these in your house and have the world's fastest self-hosted website on your own CDN. So the idea is actually going forward. It won't just be like your own website. You can actually trade that with other people because I think we actually have a fundamental business issue here that needs to be solved where it's, I don't think a single company like myself, I don't see myself being like anything better than a Cloudflare or Nakamaya, for instance, at managing this sort of level of network. I think actually this problem goes beyond the capability of a single company or organization. I don't think it can be done by one hierarchy. I think this is something like the internet that is just the entire humanity needs to do this. And we just need some tools. And I'm just trying to build those tools. That's why it's open source. You can use these yourself. You can spin them, fork them off. And if, you know, like, for instance, everyone else mentions the China case, right? You want to run a CDN in China. Like, you can't really do that if you're outside of China. There's all kinds of legal issues and restrictions. But if you're based in China and you got these tools, you could set it up probably better than I could. And I didn't even need to know that person. They could just run and do that. Sort of like how we set up NGNX or Apache, right? It's just a tool that's available and we just use it. The author of NGNX, Igor Cisoyev, you know, Saint, they don't really know what we're doing with their server. It doesn't matter. It's fine. And so I'm just trying to contribute something. And this whole, you know, if it looks like I'm trying to monetize and all that, that's really just sort of I can stay alive. I have a very humble lifestyle, but it does cost a little bit of money. And that's really all that is. It's not supposed to be a huge company. I'm not looking for funding and all that kind of stuff, right? I just literally want to connect people so that they can deploy it as they want. Yeah, so one more question? Yeah, so. Okay, so it came a bit wrong. Okay, so just to clarify, Akamai does have 99% of the servers, but serves 1% of the companies globally. So it's kind of a little bit paradox because they only serve like the one percenters because nobody else can afford them. That's why a lot of companies are like talking to us and we're partnering like with the likes of Cloudflare, Cloudfront actually. Minimize the gap between what Cloudflare can do to companies as compared to what Akamai can do because currently it's really difficult to compete with them and they're not really talking to small companies. You can try it, we've tried it. Like it took us really a long time to actually get into negotiations of our partnership with them because they simply do not talk to small companies. They ask you about your bandwidth and we tell them actually that we serve clients, you know, across all our clients, like we serve a lot of terabytes and still it's like, oh, okay, then it takes like a month for them to get back. So it doesn't beat the purpose. I actually disagree with your opinion. One point is that you said you don't want to monetize it. I definitely think you should because I think there should be a lot more people who are actually trying to do this who try to connect people so it's like easier for example for companies like us that actually want to optimize assets. But let's take for example, Indonesia. We have absolutely no choice in Indonesia. We're talking now, for example, the likes of CV networks that have one European Jakarta and they're thinking of opening two more and they're all really interesting for us because at least they can serve the traffic for 10 companies, right? So now we're like picking, oh, okay, who are gonna be the 10 companies that we offer this to? It's like being like you are in a real situation, right? And you're throwing out bread and it's like people are jumping out for the bread, right? So it's really bad. So I think what you're doing is really good and I hope there are a lot more like that. Thank you, thank you very much. All right, let's go, Sebastian. By the way, is there anyone from Akamai? We shouldn't be talking bad about anyone. This is all, I love that we're sharing these stories. Like this is really what the meet up is for. Like to meet people who are working on other web performance problems and we can sort of get feedback and there might be someone from Akamai here who's probably not going to raise their hand. It's okay, this is a safe environment. You're welcome to come out. If you are, we should totally host it at your, because you are the inspiration for all other CDNs in the world. You guys started at the first one and I mean that's incredible with the amount of good that has done for the world. I think that's really commendable. No, it sounds like we should all applaud now but there's nobody here probably. I'll finish here then. Okay, thank you very much, Sebastian.