 My DeafNation friends, I hope you are doing well. It's not green screen that is real behind you, so I'm trying something completely new today. There's a 50 second delay, so let me wait a bit and make for sure that you can see me, hear me, let me go here in the back up, okay, looks not so bad, by the way. So let me know in the chat, James, hey, James is there, awesome to see you, James. I realized that my, because some of you know that probably, but my connection at home is really bad. I rely on a 4G connection. And today I was in my garden and I did a speed test and I saw that the upload was really high compared to my office, okay, because I'm outside. And yeah, you know, it's light, it's not so cold and I say, yeah, you know, why not? So I'm here in my garden, the cats are here, they don't really understand what is happening. And well, hopefully let me know if it's better than last week, last week it was really, really shoppy. Let me go to the chat and I can see you and you can hear me, yeah, cool, okay. Maybe it's not the best, I think, but at least it looks like it's smooth, okay. So a few minutes ago it was really nice weather, okay. Now it's getting cloudy, but yeah, you know, we're going to talk about the cloud. Let me be honest also on something else, my guest is still not there, okay. I send him some messages, he didn't show up yet. So I have to talk while he's arriving, but his name is Conta and he's working for a company called Claver Cloud. So yeah, no, we have clouds and anything. So anyway, I'm really happy to be here. It was a bit the last minute I tried to set up my microphone. I can show you how I'm holding my microphone. Let me show you that. You see that? How do you call that? A palm, because I could not use the stand. It was not working. Oh, and you can see a cat behind me, or two cats, even two cats, okay. They're wondering what is happening. And there is my pool, but my pool is empty right now because I need to fix some stuff, okay. Anyway, let me ping Conta and I don't have his private number. If someone is on the chat and that he has his private number, please tell him to check his Twitter private messages. There's the link to go on the stream, okay. Anyway, how are you today? Let me see. What do we have here? We have a Tony. Oh, Tony. Yeah. Yeah. Abdel Malek. Hi, Abdel Malek. Hi, John. And yeah. Don't be shy. Tell me where you come from and let me drink some water. I'm a bit stressed that my guest is not there. If he's not there, you know, we can do other stuff. I can show you some live coding. I've been doing some really exciting stuff around Qwarkis lately, Qwarkis and operators in Kubernetes. How can you make a Qwarkis operator and Kubernetes operator using Qwarkis? So if you want to see what I've been doing around this, well, you can just say it in the chat. Let me check the restream. If Contain is there, no, let me send him another message. Okay. I'm thinking in French, oh, now come on, see, okay. So hopefully he's late with his previous meeting. I know he had to move some meetings to make time available for me today. So maybe, yeah, he's the boss, you know, so maybe he had a crucial business opportunity or something went wrong and yeah, he could not delay it. Anyway, I hope you're doing fine today. Usually I do a presentation of my weapons, you know, my cup of your weapons. Unfortunately, I forgot to bring them in the garden. I could have show you some of the weapons. But anyway, he's still not there. You know what, let's, let's, oh yeah, we need, hello everyone, Winnipeg, Winnipeg, okay, yeah. So, and John is joining from Frankfurt and we have Code Zombie there. Okay, hi, Code Zombie, I hope you are doing well, doing well. Okay, cool. Anyway, it looks like my connection is better. I could even maybe try to enhance my camera definition. Is that dangerous? Let's try it. Let's try it. Okay, maybe the definition will be a bit better and I will see in 15 seconds what happens. So my guess is not there. Let me see. Your garden is beautiful. Thank you. Yeah. Show us something, please. That would be great. Okay. Awesome. Can you talk us about ConnectAPI of Reddit? Alec is asking me, ConnectAPI, could you tell me more? I'm really sorry. I don't know what it is. There are so many things at Reddit to be honest. I don't know exactly what code ConnectAPI is, but anyway, you know what, let me share my screen and let me show you a bit, let's do some live coding, okay? Let me share my screen and there we go and I share and oh, there is content. There's content. Oh, you know what? Let me wait once again. I bring content on screen. I stop my screen share. Awesome. He's there. Hey, content. Hello. Hello. How are you? Fine. Really fine. A great day, a little bit late, like usual, you know? Yeah. Yeah. I was telling the people that I was not too worried. I was about to start a live coding demo while we were waiting for you, but you're here. It's great. It's awesome. Super cool. So let me see. Let's start right now and let me see. We can hear you well. We can see you well. That's awesome. I have even a banner to interview. So I always started this way, content, please introduce yourself because maybe not everyone knows who you are. So do a quick intro of yourself and then we will take it from there. Sure. So my name is Krem and I am the CEO of Clever Cloud. So Clever Cloud is a platform service slash cloud computing platform, global cloud computing platform, who basically automated everything. And at the beginning I'm a developer and now I'm like I'm a CEO, so I'm doing a lot of stuff. Yeah. So that's what I did. That's what you did. Okay. Second question that I always ask, and for you I know the answer will be different because I know the last year it was not the normal years, the weird stuff happening and usually I ask people how do they handle this situation? And I know you did something incredible last year at the beginning of the pandemic. Crazy story. Maybe you can share what you did, what you decided to do with a lot of other people because yeah, I think it's pretty unusual. Yeah. It was a really, really good idea. So at the beginning of the pandemic, what we see with some friends, it's depending the pandemic is affecting a lot of people in the lung and like a small part of the population affected by the coronavirus will be affected into the lung and will need to have a mechanical ventilator assistance to breathe. And I think it's the part of the people going to mechanical ventilation is very small, but the thing is the coronavirus is spreading very quickly. It varies with a lot of capability of contagion. And so the small part becoming a large amount of people, which is not usual to manage as a health system. So that's why a lot of health systems have a need of a few ventilators more inside their stuff. And the point is, we decided to say, OK, maybe what we need to do is created a mechanical ventilator. And we started to think about creating a ventilator from nothing, a full open source and a reproducible and everything. And the good thing is the project I've been kickstarting in two weeks from just five people to 300 people working together, coming from hardware, software, marketing, certification, legal certification and legal on some people to the certification also on legal and industry stuff. And with all these guys, we have been able to create a ventilator in just three weeks. In fact, we're beginning at the 17th March on the 4th of April. I plug the first light being, which was a pig on top of the ventilator. And in the summer, we have the first human plug. And it was the first made ever. It was a digital first open source 3D printed ventilator with clinical trial with human people. And over all the open source project all over the world, including Tesla, including NASA, including everyone, just one of clinical trial with human people. It's a Mac Air. And it's a very cool project, which we were very proud to work on. And it was really awesome. Yeah. But it's still going. And it's still going. Yeah. How is it going, by the way? You are, what are the next steps of this project? So right now, we are going to do the next step of clinical trial, which is very complicated because of bureaucracy. Basically, health is a fucking hell of bureaucracy. And there is just a few people pragmatic in this area. So I'm very sad to say that we have been able to save lots of people without no problem, which have not been possible for paper reasons, for bureaucratic reasons. But we need to smash that. So today, we know that there is people using the Mac Air. We have an industry partnership with Seb, which is a Roventa. And all these brands is the Seb group, which is one of the biggest industrial group in French. So we're working with the Army, with the CEA, with a lot of very large groups in France. And basically, today, Seb is building the Mac Air. And we have Mac Air used in Senegal, in the island, Madagascar, this kind of place, and in clinical trial in France. And what we want to do is like we have open source, the new Mac Air version 5, which address way more than just the COVID-19 pandemic, which is a device with total cost of ownership, but total cost of building out of the factory, the Mac Air cost around 4,000 to 5,000 euros, which compared to device which is 37,000 euros. OK? And it's very precise. As a ventilator, it's a very precise range of pressure and everything. It's a very interesting project in the open source side for various very clever ideas that have been made there. And what we want to build around that is to say, OK, what we will do now is creating a nonprofit. We will own a company which buildings the Mac Air and build something around open source health system. Because today, all the health system are proprietary, closed, not able to communicate with each other, and not able to be improved by all those people in the medical life or those people around the world. So that's why we want to build something around open source. And what I want to do is creating something like the red dot of open source medical device, because it's very important to build something around that. So it's a very enthusiastic product we have here, and we try to build something great. Yeah, and you can be really proud of everyone that helps you. And it's awesome. And yeah, open source health is even something that we discussed internally at Red Hat with friends. When the vaccine came out, the NRA messenger vaccine, there was a really nice blog post explaining it's more or less like a code, you know? And it's code that's a bit hacked. And there's a header. And we were saying, wow, it would be awesome if that could be open source, you know? And people can contribute to send a poll request to the vaccine to a new version. And so yeah, I don't know if that's anything realistic around that. Yeah, if you think about what you can do is RNM is kind of a super nice technology to build a protein you need in the human body. So with RNM, you've got virtually built things that what is awesomely made to cure people, to transform people for people who want to have augmented you. And in fact, if you look at the big numbers, it's cheap. Building an RNM molecule formula is very cheap. And I really think that this is the way that else can flip to this other world where the cost of produce becomes so cheap that we can be in the same way than IT with open source. And the formulas in RNM can become something open source hackable, which is order with company taking the responsibility of putting a formula at some point at the market. And when you buy something to read that, you're not buying the code, the code is open source. What you're buying is the support, the certification that this version is working, the certitude that someone will look at the problem if you have a problem, and the confidence that every problem will be handled. If you think about that, what you're buying is a certification, is an insurance, but you're not buying the tech by itself. And using the RNM, I think in a few years, it can become very interesting to have open source formulas sponsored by a company which is basically just under the certification process and be totally fabulous because some company will have a huge amount of RNM fabrication systems. So totally, the RNM is for me one of the key to have open source chemicals in the arts industry. Yeah, well, nice. OK, so our idea that we had was not that stupid. But I don't know if you read that article that's compare that RNM messenger to code. It's really well explained. We've written for developers like us. And wow, it was mind blowing for me. So anyway, let's talk about you introduce quickly CleverCloud. Could you explain again, basically, what do you do at CleverCloud? What's for company? In fact, if you look at the thing is today, what people want is to ship application more quickly. And to ship application more quickly, what we can do is to reduce the numbers of meetings. People are tied with meetings. So how to reduce the numbers of meeting? Automate stuff, like make no brainer stuff, no decision to make. So CleverCloud was for 10 years something that we can call DevOps improved by high. DevOps forward by high. What's means that a developer can code, push, and will deploy, and process, monitor, update, like everything to guarantee that in four years you can come back, the code will be still in production, which means we will update everything around it. We will relaunch when it fails. We will manage the automatic scalability system and everything. So it's a platform as a service, container as a service, data as a service, and more and more service coming up, which is no brainer technology for developer with huge performance and a very huge improvement on the security model. So that's what we do at CleverCloud. OK. When you say with an improved security model, what do you mean exactly by that? We're kind of people which is taught antagonist to all the market. What I think is we really think that the best way to isolate application from each other to come at the core of the CPU. So basically, we're using virtual machine where those people are using containers, for example, and which means that we have better performances on IO, which is way more interesting than the CPU. And we have better preferences on security, which means also you can access it inside your machine. You can have a very great layer on monitoring on top because monitoring is way more easy inside of the turn machine than a container. And you can also use container at CleverCloud. The isolation model would be there. So we've right our only provision layer, which is able to propels way more quickly. For example, when you start a return machine at CleverCloud with Linux inside it, it costs you six to seven seconds at least. OK, yeah, that's true. OK, so I know already the answer, and I'm trolling you. You are not using Kubernetes. No, no, we do some things that works. I know that's not the point, but yeah. Using Kubernetes is very interesting if what you want to solve is consulting. Because the thing is so much complicated that if you want to solve consulting, it's awesome. But what we do is something that actually works. The team have only three people to maintain the old infrastructure, and we're managing hundreds of thousands of applications live up to late all the time. And there is no prediction incident every time. So I don't know any time with any size of SRE with a lot of application like that with the same results in Kubernetes. Because Kubernetes is very complex and there is several flows in my humble opinion on the model, which drives to that. But at some point, we will offer Kubernetes as a service for the customer. I really think that Kubernetes model is not a good model for application, but I think some people want to do it. So it's like everyone. Sometimes people don't take your technology, but you need to support it, and we will support it in a good way. Like we do for Docker, for example, where we were the first to be called ever to support Docker. I really think this is maybe the worst technology happening to the internet since maybe Pearl. But people want to use it. No problem. We're here to help. Yeah, yeah, yeah. OK, OK. You have really smart people inside your company. I know some of them. And what you also do, you create new products, new libraries that you open source. I don't remember the name. You have this reverse proxy one. What is the name of that? Ah, Sosu. Sosu, Sosu, Sosu, so. Sosu, it's the name of the Japanese fontane where there is water enterrated into a bamboo and sometimes a bamboo does stuff. Oh, oh yeah, exactly. This is a Sosu. And that's why we think it was accurate for a reverse proxy. OK, yeah. And I think I believe it's written in Rust. Is that correct? Yeah. We are very strong Rust believers since years and years now. So we are the people behind NUM, for example, the Parcel Library, Parcel Combination Library in Rust used by nearly everyone because it's a transitive dependency for, I don't know, most of dependencies in Rust. So we are very strong contributors to the Rust ecosystem. And years ago, we've write something called Sosu. So Sosu is a reverse proxy which is able to reconfigure why it is running. When I say that, I've not seen things like a tri-fig or stuff like, what is the name of the CNTF stuff? You know, it's a thing from the oldest reverse proxy. What they do is basically creating a new process inside the process. Yeah, like Envoy. They're basically creating a new context inside the context and letting old connection relaying on the old one and making a process split. The problem with this kind of strategy is when you restart your reverse proxy very often, like really often, you will stack a lot of stuff. And in the modern web with HTTP 2, 3, web socket and everything, you keep the connection open, which means that it becomes more and more complicated to kill the old version of the system. I don't know if you know, in Azure Proxy, the people who've write the Azure Proxy system developer which are able to reload Azure Proxy in the system it was us who write that. So we really know this kind of strategy and we know where it's going, when or when it will be done. So we decided to rerun everything from the scratch with Rust, with Sazoo. And Sazoo is taking something like 30% more request than Azure Proxy on the same workload and same hardware on what we do. So you can, what is great with Sazoo is you can change the configuration without restarting it. You can change the configuration from front's website per websites without changing it. And what we can do is same, for example, this certificate and website will use open SSL and this website and certificate will use Russell. So you can switch the SSL library at live and why it is working. And you can also change the binary and change the restart of Sazoo, all the vision and all the binary of Sazoo, without losing any TCP connection. So it's something which is already made to you know, you build something on Sazoo, it will work forever. Yeah, yeah, yeah. Okay. It's full open source. It's full open source, yeah, like everything, almost everything you do is open source. Yeah, we do a lot of open source, yeah. Open source, yeah. And then you were saying you were the first to provide Docker support. I think you were the first one to provide Chiklokas as a service. I think I'm not. Yeah, sure. Yeah, I remember Laurent thinking me. And you were also the first one providing support for Quarkus, I believe so. Yes, we also the first one to provide support for Quarkus. I'm a really huge fan of Quarkus personally. Yeah, okay, including native. I think you- Including native support. From the first day I think, when you provided support, you say, okay, you can do a JVM or you can do it to native. So, yeah, because it's one of the best interests of Quarkus is this way to clean up a lot of old libraries to be able to use it on multiple version. So for me, Quarkus is kind of, I don't know if you remember it, I don't know how it was international now, but in France, we have something called Play 1.0. And basically Play 1.0 was a guy, very clever one, who say, okay, we will build Java on the raise. This kind of productivity framework, like Ruby on raise offered to the web developer, but on the Java platform, because Java is awesome, you can do stuff and you have all the ecosystem. If you want to talk with an Azure cluster in Java, it's, you will have the driver because all the open, the big data platform and everything. And Play 1.0 was very usable and very developer friendly. So Play 2.0 switched to Scala and it was another story and it was really great. I was a huge fan of Play 2.0, but it was not the same story, you know, for the Java developer ecosystem. And for me, Quarkus is clearly the Java on raise, a productivity framework, really happy to use and very clean and smart and used to the developer. And one of the part was the idea of saying, okay, we will try to know what is running inside it because when all the GE is tagged, a little bit lost sometime to the dependency, the complex dependency tie on, you know, of thread and stuff. And what bring Quarkus for me is a clarity to the ecosystem. So that's why at the beginning we're starting with the native support also because I think it's very important. And what we can do with that is going also to the web assembly one, which is really important for us also. So I really like all the going on Quarkus and I really think it's a fresh air for all the Java ecosystem. Yeah, yeah, yeah. And there's clearly an influence from Play 1.0 because part of the Quarkus team, there's Stefan Napaldo that you know, and he has always been a huge fan of Play. So he put some spirit of play in Quarkus. Yeah, yeah, yeah. And he also was the guy who write red pipes. Red pipes was very interesting also as a framework. And I really think that everything in Quarkus profit for all this experiment and story. And I really think it's a great platform and great framework for everyone. I really think it. Yeah, yeah, yeah. I'm not saying that because it's a Reddit but when the first version of Quarkus came out two years ago, I was feeling like when I was a junior that you could spend the whole weekend trying everything out and it was just magic and it was just working and oh, let's try native. That looks complicated, but oh no, it's just working. Oh, it's less and yeah, a lot of excitement. So I can, well, you see now with Spring Boot is also starting with native. Everyone is moving as though it's only good for the competition. Is there is a bit of competition around the Java ecosystem? I believe I think it's, it will benefit everyone. You know, an ecosystem is living well when there is some competition, so it's some new ideas or it's people pushing new stuff around. And when the competition is still friendly, you know on the Java ecosystem, it's a competition is friendly. Like everyone will have a meal at the end of the day on the table and on the rest ecosystem, there is various, you know, framework going around and you have, you know you have this kind of good spirit where the framework on the presentation page say thing, okay, we take inspiration from there and there and we're looking a lot what does the team of this framework because basically this is very interesting stuff. Go check this project. Maybe we will like it more than what we do here. And you know, this kind of fresh impression which is great. And I think the Java ecosystem is really nice about that because there is really people, you know, paying for that. It's something I don't like for it's some and for example, cloud native foundation, you have a lot of project which is basically all just funded by DC, you know, and everyone need to hardly fight, you know to win over the battle of the number of star, GitHub, number of pool on the Docker app. And we are the most important there. And someday we will make money. And for that, we will be buy by this big company for whatever we know or is on, you know, I don't like when there is too much, you know, perspective money on the market, but on the Java ecosystem, there is a lot of money like everyone is using Java. And I spend a lot of money on Java every year for the ecosystem. And you spend a lot of Java money on the ecosystem. Uber does it, Yahoo does it like everyone is spending money on the Java ecosystem. So you have all the libraries, you have all stuff and you have project, which is really built by people who have the same interest that the project needs to be doing well and growing on the good way. And you have several foundation able to take the perspective of the future of the ecosystem like the Apache Foundation, the Eclipse Foundation, which building a great community around everything to building stuff of great quality. Yeah, yeah, yeah, yeah. It's quite perspective, the fact that, yeah, there's a lot of money of Java, which makes it easier. I never took it through this scope, but you're right, you're totally right. Let's see, is the chat is doing, yeah, the chat is agreeing with you. Serverless, oh, there was one question. Yeah. Where is the chat? On YouTube. Well, again, I'm missing, I'm missing the people. We are on YouTube, we are also on Twitch. Sorry for the people on Twitch. I always forgot you, but hello. So there's maybe chats on Twitch. I need to set up the aggregation. And there's also on Facebook. Usually on Facebook, there are not that many people, but maybe if there are people on Facebook, hello. Oh, and the sun is coming back here. Yeah, there was a question about Serverless. I think it was when you were explaining what you are doing at Clever Cloud. Do you have any solution around Serverless or maybe an opinion about Serverless? I think you have an opinion about Serverless. I have strong opinion about Serverless. Yeah, I'm not surprised. I'm just tweeting the URL of the YouTube. So about Serverless, what we do with Serverless. Basically, at Clever Cloud, we have decided that it was really important that we can bring to people a Serverless possibility. By essence, Clever Cloud is Serverless. I know you get pushed and we made everything like monitoring of data and everything. It's what we can say Serverless. But when I say Serverless, that I mean about function as a service, a fast. So we say, okay, how to do a fast would make sense in our ecosystem. And we're bringing about some things thinking, oh yeah, what we can do, it was great. It was launching a virtual machine for all function execution because it will be secured. Yeah, but a virtual machine, even if we launch it at seven seconds, it's a little bit long for just one execution. Yeah, but do we really need POSIX to do that? No, no, no. And we basically build a kernel less function as a service, which is basically a small system which able to start a virtual machine in less than one microsecond, and one millisecond, I mean. And we have used a special already something we built on top of the virtualization stack, which able to use the virtual machine as a process and fork it. So in each execution of the function, we're forking it and we're forking it. And it costs us like six microseconds to fork a virtual machine. So for each function execution, we will give you a virtual machine, which is a really small virtual machine with just a few megabytes and few stuff like that. But you will have the isolation. And what we built is we built something you can put wasm, don't say web assembly inside and HTTP function and die. And we are able to reply to HTTP or to Pulsar event broker to the system. And we built something very fun about it, if you want, I can show you. Yeah. The fast demo, does the fast demo is working well? Oh, the fast demo is not working well. Okay. I miss the URL of the fast demo. And you are putting that in the chat, in which chat? I ask for the URL of the fast demo for Clever Cloud and I put it on the YouTube when it's done. And the point is what's very interesting here is what you can do with that is saying, okay, so we give you a micro-determination to execute each function. And we use it right now, if you go on Clever Cloud, you have an interactive map where each HTTP request will get a dot on the map or from where it's executed. Each access log is actually processed by one function inside of, in our live serverless system. So it's working a lot on this. We currently on Alpha inside Clever Cloud and we're bringing all the new API on top of that. And we will be able to push this for the summer thing on an Alpha release, like something a little bit, you know, raw. Yeah, yeah, yeah. That sounds exciting. Really, yeah, yeah, really interesting. Yeah, a kernel, kernel less, you mentioned that. Yeah. And it's really nice. Yeah, yeah, go on and go on. It's a really nice way to rethink about how we can build around all the platform a new way to have a good experience on functional service around something which is bringing you the ability to get all the old libraries, like you on WebAssembly. So if you want to encode a video, use this FFMP, go for it, you can encode WebAssembly FFMP. So what we're building, it's some great demo. For example, we have the odd dog, not odd dog demo where a full neural network is actually encoded in some wasm system. So it's really nice stuff. If you want to see weird inception about how it is, you can follow on Twitter, someone called Jofua Kupri, which is Jay Kupri, a clever mind of, one of the clever minds of Clever Cloud. Okay, yeah, yeah. And I love the reference to hot dog, not hot dog. One of the best service ever, Silicon Valley for people who don't know. You have to watch it. We're running almost at the end already, but I want to ask you just one last question because you mentioned Apache Pulsar. And I know we had a discussion one year ago, I think, quickly together. And you say you were a big fan of that. Is it still the case? And do you still think it's the next generation after Kafka? I think it will be next to Kafka. Like Kafka is growing and it's a very good platform. I'm not here to shame Kafka, but Pulsar is a very enthusiastic platform because you can build a lot of great stuff about that. So what is great about Pulsar is you have this model where all topics are virtual. So you don't pay for a topic creation. There is no administrative cost for that. Which means that you can create a 50,000 topics, no problem, it will not be a problem like you can have on Kafka, which is not the same model there. You can listen pattern like a regex pattern of the number of topics. And you have the basic idea of tenants inside Pulsar, which is really great. So you have, you have few stuff which is very interesting inside Pulsar ecosystem where I really like. And we built a lot of stuff on top of Pulsar at Klaverklau because it really fits to our models and needs. Which don't, don't say, I don't say that at some point Kafka will disappear. I don't think so. It's a great technology with great people maintaining it with a lot of great use case. And I think that at some point, Pulsar and Kafka have different life and the progression of the project. But really I think what Kafka bring to the old ecosystem is the basic idea of event source system or event oriented system or maybe secure if you want to have a really, you know, event source system. Putting the event and the event bus in the middle, you know, of your infrastructure. And I think with that Kafka have done a great stuff to all architecture everywhere. And a lot of company can have the profit. And I think it's really great for all the ecosystem that many possibilities are offered on the software industry. And for that, I think it's really great that you have sometime Kafka, sometime Pulsar or using RabbitMQ, whatever. And if you want to try with Pulsar, there is a beta of Pulsar as a service of Clare Claude. And so you can play with it and use it and be happy with it. Okay, just last thing on that. I think it was you that, wasn't you that had the idea to replace DN, well, to use Pulsar, Pulsar or Kafka to you for DNS, something like that? Yeah. That was you. It says, yeah. I call this stream DNS. Right now, we have right on open source, a first version we are not confident with. So we will launch a second version, which is totally rebranded. And yeah, oh, found the, I found the URL of the demo. So I put it on the chat. Yes, the point with stream is to explain that you have to give me, I just realized your messages don't appear because YouTube removed the links. So we don't see your messages. But you know what you can do? You have a private chat here in the restream, put them there and I can, I can pay them. I say hello here. Sorry, I just realized that. Look, yeah, okay, it's there. So yeah, what you can do is, so we decide to propagate DNS using a topic inside our Pulsar and private stream. So basically what we think is DNS is basically a very simple security system where all the data will be contained in a topic, a compressed topic, compressed topic on Pulsar and Kafka are basically the same thing. You just compress on the key, you know? And what is great is you will get an archive topic with all the change so you can see it. And the compressed topic will be just a few messages. Like even if you have like a few domains inside it, like if you have like a 500,000 message, it's not that you will consume it in like three minutes. It's not very big deal. And we're leaving all the database of the DNS on the RAM, in fact. So you're just on the memory. Oh yeah, memory. And basically you just have a consumption of the topic and everything is on the memory. So it's called string DNS, but the first version we put on GitHub is not used in prediction at home. We see that there is some implementation problem inside it and we will push an observation in a few weeks, I think. It's internally in test, the new version. Okay, awesome. Because when we discussed that, I thought that maybe it was still yet an idea. I didn't knew that you were actually really trying that out. I found the idea really nice, but you actually did it. And that's what I love in your company. You are so, you're all innovators, you try it, you experiment stuff. I think it's really important. Okay, I need to keep some time because I promise the people here that we will play a game. You can stick around in the YouTube channel if you want. It was an awesome discussion. Really, yeah, really different from the other. Yeah, I like it because you have some strong opinions, but also openings and people appreciated that. I can see that in the chat. So thank you so much. I can't wait to meet again in real. I don't know when that will happen. Yeah, me too. Yeah, I don't know. Maybe Devox friends, I don't know, but I know it will happen. But if it happens. We are all hoping for Devox friends, I think. Exactly, exactly. Yeah, yeah, yeah. With us, maybe up here anyway, you know. Exactly. And oh, even the cat come here to say goodbye. So yeah, I'm in Zergarden. Okay, thank you so much and speak to you soon. Okay. Thank you very much. Bye. See you soon. Bye. Oh, I did the opposite one. So there we go. Hey, that was an awesome interview. I don't know what you think about it. Let me know in the chat. I see a lot of comments, okay? So we had a longer interview than usual, okay? And that's not a problem because it was really, really interesting. Anyway, in my opinion, we have 14 minutes left, okay? And that's enough to, well, let's play a game again. And again, it will be correct free. But this time, I just want to show you, I did a small change to the game, okay? So remember, let me go here and share my screen. That is how I do it. Okay, so we should be in inception mode. And just to remind you, Quake Kubernetes. Okay, so it's Kubernetes, sorry. I'm using this version of Quake 3 that runs, that you can deploy on Kubernetes just by applying a resource, okay? And that's awesome. That's really fun. The maps, the bots, it's all put in config maps so you can change your config map and that will be applied, okay? I got it running here. But I did one change last week. Let me go to the developer view. So that is my cluster here running in Amazon. That is my Quake pod, okay? And if I take a look at the logs, let me go to the content server. I think it is the content server. Maybe we don't see it here. No, let me go back here to the server. I realized that all the logs, doesn't be running for a long time, but anyway, all the logs of Quake 3, when you kill someone, when you get fragged, when you pick up an item, it's put in the log of the pod. And I say, hey, you know what? Let's do something. Let's write a small Quarkus application that will grab the logs from the pod. And yeah, because it's fun, let's turn that into a Kafka record and put that on a Kafka topic, okay? And then I have another application. It's way over-engineered, but it's just me having fun. If we go here to the topology. So that is Quakekism. Maybe people know Dibisium, which is how to put something in front of your database to capture events and push that to Kafka. It's an awesome project by Gunnar called Dibisium. I make Quakekism, okay? And then I have Quake aggregator, which is a Kafka stream that will do some computation, basically calculating how much kills you have make, okay? And put that on an other topic and I push that to a UI. I can open the UI, it should be empty. Yeah, it is empty. Oh no, I haven't, so I haven't cleaned up the UI from yesterday because I ruined that yesterday for a Brazilian conference. Anyways, not a big problem, but it's working. I just want to show you a little bit of code before we play, because it's a really nice piece of code in my opinion. Look at this, this is Quarkus, of course, and I'm using the Kubernetes client, which is from Fabricate, which it's really easy to speak with your Kubernetes cluster, okay? And basically what I do here, I retrieve the pod Quake, okay? And then I can create an input stream by watching the logs. Basically I make a preferred stream of that and then I call lines, which creates for me a stream, okay? So I have my streams of logs. And then I'm using mutiny, which is a reactive library in Quarkus, okay? And I start watching the logs, okay? So that is this method here. And this annotation is all I need to do to send a record anything I want on a Kafka topic, okay? And here I'm using a multi, a multi, that is a construct from mutiny, a reactive construct. I say create a multi from my log stream, do some filtering because I only want the kill messages. So here I'm just doing some basic string manipulation and I map each message, I create a Kafka record of it, okay? That's all I do. And I can quickly show you, do I have the Quake aggregator, which is my other application. This is a bit more complicated, but it's Kafka stream. Kafka stream is, I think, always a bit complicated, okay? And basically what I do, I construct my stream here. I consume my Quake event topic that is here. I consume it and for every value I create an score event for that so I can manipulate an object is better for me. I group that by key. The key is the name of the player and then I do an aggregation and look at this really complicated aggregation. Look, kill plus plus, that's all I do, okay? But it's just to make the point. And that's enough to create an aggregation and to see how many tracks, how many kills you did, okay? So I package that as a pod, deploy it, and it's running, okay? That's a nice thing. It's developer Joe with Perkis. Anyway, we can try it quickly. Let me see, let me share the link with you. If you want to play, of course, we have almost 10 minutes. So I hope some of you will join. There we go. If the first time, I hope you have a good connection because you had to download the game, but it's pretty fast. You have to agree also, there's an agreement and you should maybe hear the police there, but it's always okay. Let's play Quake from my garden and I realize I don't have a mouse. It will be on the trackpad, so I will be really bad. Okay, anyway. And let me see, yeah, it should happen. Yeah, it's running. Here, welcome to the developer comp. That was yesterday. Oh, let me remove the sound. Okay, a new party will start. And I have huge ping here. Let me see. Oh, okay, let's try to play. It's really hard with the pad. It's almost impossible. Okay, let me go here. Let me try to roll some people. Only bots out there, so I think people, let me try that. I've lost the lead. Oh, but yeah, I cannot, oh, and I've got killed. Okay, let me try it again. Oh, yes, I frag some people, but you know what? Let me show you them. Oh, it should get updated here. Sabi12, oh, looks like my, hmm. Looks like my stream is blocked, okay? Let me see. Let me check the logs. Maybe something went wrong. No, my aggregator is there. Quake aggregator, okay. Why is it not aggregating any stuff here? See this, render some. I don't know what is happening, but we still can play. Let me see. I lost the lead. No one is joining, so, or it's not working for you. Let me see. Very nice, Sebastian's streaming example, consume and deploy producer. Yeah, so you know what? You can play the game. I will leave it open for a while. If I can exit it, to, to, to, to, to. So producer, yeah. In fact, my log reader here, that is my producer, okay? It's an outgoing. And again, all I need to do in Quarkus is two things. Well, I need to add the right dependency, okay? That is one. Quarkus, a small way reactive messaging Kafka, okay? And then I need to add some configuration, pretty easy to do, application configuration. Here, I say, well, my Kafka service running. So I can do that also locally with Docker compose. I say, I want to use a smaller Kafka. I give the name of my Kafka topic. So it's Quake event. I say it's an outgoing. Then I have a serializer. Here I'm using just a string serializer. We keep it simple. And yeah, that's it. Once I have that, I can become a producer of events just by using this annotation. Log reader, is that, yeah. Just by using that, okay? I could also create an, an emitter. Well, I have that here, yeah, emitter. Emitter, that is in the case, I want to be in an imperative style. Let's say I'm in a REST endpoint method. And so I need to have an imperative way to send a Kafka request. I can do that with the emitter. But if I'm in a fully streaming, async stuff like here, because here I'm relying on a stream, my logs, I can just use this annotation, okay? And I could just do also, in going, incoming, okay? And I could here for aggregated. And here I could write my methods and consume a multi-reactive stream from another topic. Okay, so it's, it's really, really nice. I really enjoy combining Quarkus, Quarkus Quake and Kafka in this case, okay? And I should use, try Apache Pulsar, okay? Let me go to the chat, see if there are any, any, do, do, do, do, do. If there are any questions, there are no questions and we are almost out of time. So we can end it here. I think, let me, let me, let me, do, do, do, do, let me stop my screen share. There we go. So I'm happy because I looks like that today, looks like the connection was pretty good, okay? So I have to find a way. I won't be able to stream always outside. Well, summer's coming, so it will be easier. And yeah, I see you again in 15 days. I still don't know who my guest will be. I still had to go to fishing to find a guest. And yeah, thank you everyone. And oh, did you marry? Yeah, I'm sorry. You know what? We are stick to the US time for, for these events. And in the US, they're still on, they changed their time two weeks ago. We are changing this weekend. So we are in this weird period where the US and the rest of the world is not in sync. Well, I'm sorry for that. You can watch the replay. Watch the replay, DJ Maddy. Cantin did an awesome interview. So when it's, it's all I had for you today. Thank you so much. Stay safe, enjoy the rest of the day, evening, depending where you are. And well, see you on the show in 15 days, but I will be there on other streams, deep dives and et cetera. Okay. Thank you so much. I wish you a nice.