Ólafur Guðmundsson's path: From Iceland's shores to Internet's core

Hello everyone and welcome to the show We Are Cloudflare and this one is full of stories and history. Today we have a special guest with more experience than most and he's also a real guru network protocols and in terms of DNS. Hello Ólafur, how are you? Hello João, I am doing good. It's a nice December day here and just very happy to be talking to you today. So just for clarification, your name is pronounced Ólafur Guðmundsson, is that right? Almost. It's Ólafur Guðmundsson. What's the right pronunciation in Icelandic? Ólafur Guðmundsson. It's a little bit more mixed together, like the language does that, right? Yes, and I speak fast. Where does this show find you? I am in my home in Chevy Chase, Maryland. Chevy Chase, that's a popular name. This is my office that I have been working from for many years. And many years at Cloudflare, it's going to be 10 years next year, right? It's nine and a half in February. So it's a good long run in terms of Cloudflare. Of course, you have a lot of experience before Cloudflare existed even, so we will discuss a little bit on that. More specifically, and why not start there, we want to have two parts of segments. One is about your journey, the other one is more about Internet standardization over the years, something that you know really well. To start your journey part, why not just start, where did you grow up and when did you start being interested in technology and things like that? I grew up in Reykjavik, Iceland, and I was interested in science from very early on. And I had decided when I was a teenager that I was going to become a chemist. And when I was finishing high school, I decided to do a little bit of due diligence. And I discovered that most of the jobs that chemists did were boring. So I switched to computers. And to situate ourselves, first, when was this in terms of time period? Like you were growing up in Reykjavik? I was a teenager in the 1970s. 1970s, as a teenager. And how was the science perspective, the technology perspective in that part of the globe, specifically Reykjavik, because of course it's more to the north and countries that are more developed in terms of technology and universities. How was the setting there in terms of learning curve, in terms of what did technology look like at the time there? That's a good question, but I'm not sure I can answer it. I was, technology was going through a very rapid evolution at that point. We had just started getting calculators and programmable calculators. So we, me and my high school buddies, we would compete in who could write a function on each different kind of calculator faster than the others. So this is pre, for those who are really young and don't know about the time period, this was pre-personal computers, before the personal computers age, let's call it like that. Yes. I remember seeing an Apple II and I bought a Sinclair CX81 for myself as soon as it became available. And yes, the IBM PC came out much later when I started in university, that was the first year they did not have punch cards because they had converted it to a new wax system from digital equipment. So I missed on punch card. And that's a funny thing. When I went to graduate school a few years later, that was also the first year they didn't use punch cards. So wherever I went, punch cards got paced out first. That's really interesting because, again, you missed on a specific technology that when you started, it was a new one already in place. So you didn't miss much in a sense, right? Well, I had a totally different experience. And of course, some old timers kept telling me, oh, you're not learning it right because everybody says the way they learned it, it was only a real way that real people did it. Of course. And you were mentioning the moment where you decided going from biology and chemistry to computers. In terms of choosing the path, in terms of college, education, what was the process there? What did you choose specifically? Where did you study, really? Well, at that time, I picked the only university in Iceland. And as it was a small country, we couldn't afford much more than that. I had a choice between electrical engineering and computer science when I was picking a new field instead of chemistry. And electrical engineering had a drawing class and I didn't want to go through that one. So I picked computer science. And this was already the 80s or late 70s? Yeah, this was in 1980. 1980. So that personal computer time, actually, even the data science and data computing part was starting in the early days, right? Yes. The biggest things in data at that time were the fights over databases, database theories. SQL had just been invented. People thought it was good, thought it was bad. And then the mess of what we call data science today was then called information retrieval. That's a very different name. Yes. And it was basically, yes, that was the foundation for search engines and other things like that. But I'm curious also in terms of how exciting it was at the time. If you're growing up in the 90s, Internet was coming, there was a big shift because of that. But in that time, it was the personal computers that were coming, in a sense, right? Not still. At that time, it was mainframes. Mainframes, before that. Yes. And then there was this, yeah, this new thing, PCs coming. That's not useful. It's nothing but... It wasn't seen as it is today, like a revolutionary thing, right? It was not sure. People were not sure. Well, they were underpowered, all of that. Time -shared systems was what everybody used to get the capacity to do any data crunching. You had to be on a machine that had disks, et cetera, et cetera. In those days, personal computers, the programs that people shared, were shared in a printed format. There were whole magazines that published lists of programs. So people would submit their games, they would submit spreadsheet-like things, various things. And then you would sit there, typing it in, making sure you didn't have any typos. With a typo, it wouldn't work, right? Yeah. Then you had to find what line the typo was on, and everybody on the personal computers was programming in BASIC. And then, of course, fancy academics looked down on BASIC. In that regard, when did you start, like, okay, I'm ending college, this learning curve, going to the job market and seeing what's there? What was that process like? Well, I had decided I was going to go to grad school. So I was trying to figure out what did I want to study in grad school. So I looked at what I thought was going to be the future, and I was interested in performance, I was interested in networking. So I ended up applying to universities that were good, in my opinion, and had those two fields taught there. And I ended up at the University of Maryland, a few kilometers away from here. You went there in what year? 1984. And you stayed there for a while, like, in the university? Yeah, I was there for a while, then I became a research faculty there. And then in 92, I switched to industry. Exactly, to the tech sector directly, in a sense. But how was that period? And this is 80s, mid 80s, maybe? Yeah, so I switched in, sorry, it was in 93, that I switched from to a private industry. And I started working for a small startup called the Trusted Information Systems, which was a computer security startup. And they had multiple different projects, and I was hired to work on a DARPA project to enhance the DNS system to add security to it. Before going to the technology part, in the university, how was it like? Like that period of research in that time, late 80s, how was it like? What was the perspectives there in terms of the learnings that were and research that was going on? I became a disillusioned academic very quickly. I got tired of dealing with prima donnas, and the professors teaching what they thought was right. And yeah, so there was interesting work to be done. There was interesting work that we participated in, but overall, I was not enamored or impressed with what I saw in the computer science field. So that's why you also changed, right? In the 90s, 93, you were saying. And that adaptation to an industry that was booming at the time, right? The World Wide Web was becoming a thing a few years after that? It had just been announced, and there were people talking about it, but it was enabled by these underlying technologies of TCP IP that was obviously at that point winning the war of the different transport protocols, and I was very interested in that. In what way do you thought, hey, this is really interesting and will possibly change a lot of things, how we use the Internet, in a sense, but also, hey, I can contribute with my background to this, all of this. Yeah, that I just wanted to participate more than I thought I could participate. But you did participate in a sense, right? I did participate. And that was a relevant project from the get-go, the ARPA one you were mentioning, right? Yes, it was. And I worked at TIS and its successors, because it went IPO, then it got purchased, and worked on multiple DARPA projects, projects funded by DARPA while there. And I also worked a little bit on product development. For example, we were trying to create the first specialized appliance to deal with DOS attacks. That was really the first get-go in terms of those types of attacks, right? Yes. The company had a firewall product, and we realized it was not going to work well against DOS attacks, so we started a project to do that. The project got killed later on for internal politics reason. So politics are not just in academia. But it's kind of amazing in terms of background and experience to be at the starting point of some of those security elements, thinking about that on an early stage. Looking back, what did you think of that time, of that experience that you had? I thought it was great. I had established contacts, I had established reputations, so I was in a good place. But I decided, for family reasons, to step back from industry and spend my time on taking care of my children, while my wife provided for us. And I did that, and then I started doing consulting on the site. And I kept doing that for the next few years. And when did you come back to more specific roles? I joined a specialized consulting company here in DC in 2009. So that was when you came back more to the full-time type of work in 2009. And in that process, because you had a view of what it was before the personal computer, before the World Wide Web and all that, what are the main takes from your experience that you take from this experience? For example, in the early 90s, could you see what was coming, or was it not that evident for most people at the time, even in the industry, really? Interesting you mention the 2000s, or the 1990s. In the 1990s, in the US, the dial -up in DSL connectivity boomed. And the biggest player here was a company called AOL. And I nicknamed America offline, because when you used it, you didn't get anything done except see advertisements. Oh, yeah, that's true. And they were huge, right? Everyone had an email from AOL, so they were huge. Yes, they were humongous. They were also headquartered here in the area. And there was another big Internet company, UUNet, also here. And there were a whole bunch of others. So many of the things that we see that make the web today not fun are an artifact of the thinking of that era. The advertising part you're mentioning, things like that. Sites are sponsored by advertising. Advertising is all about getting as much details on the end users or the eyeballs, etc., etc. That addresses have a meaning. Yeah. Makes sense. But I was more interested in making things secure and better for everybody. And closed ecosystems like AOL was never going to deliver that. We're going to discuss more on the Internet standardization in a while. But in the return to industry, like full-time job, you're mentioning in 2009, around that time. What was that process? In the early 2000s, I was basically acting as a consultant to various TLD registries around the world doing software development for them. So you were participating, right? Yes, I was. And I was a chair of the DNS working group in the ITF in my spare time. That's how I kept myself sane. Got the break from dealing with the kids. And this was constant in terms of DNS was always around. Yes. Yes, it was always around. And also during this time, I got to do things that were not technical that I enjoyed a lot. So for about 10 years, I was coaching soccer teams or football teams for my kids and others. And that was lots of fun. And I think that was the best management training anybody can get. Makes sense. Kids can be a full hand, right? In those competitions. Yes, it's all about psychology and figuring out what they want to do, when they want to do it. And then giving them the right challenges to learn new skills. One of my biggest aha moments in that was I was coaching my daughter's team. The girls are like 10, 11. And I had been trying to get them all to learn how to dribble the ball around without looking at the ball. And look around the field. So here in this area on the east coast of the US, we have a bug that comes out every 17 years or so called cicadas. Oh. Yes, very rare, strange beasts. They can barely fly. And at the practice, the field was covered within a minute. So I said, OK, go dribble the ball out, grab a cicada and bring it to me. In one minute, every single girl learned how to dribble without looking at the ball. They had the right motivation. They had something they wanted to do, yes. That makes a difference for sure. That's a good managing lesson for sure. Yeah, so it's find out what people want to do. Makes sense. And you joined Colfer, we were mentioning before, in 2014, right? What drove you to Colfer? Why did you choose? The company was very small at the time. Well, when I realized that the job I'd been doing for five years, I didn't want to do it anymore. And the company was probably going to go away anyway. So I started looking around. And I could have gone back into the industry that I knew well, the DNS industry. Or I could find something different. So I looked around, Colfer came up. I read some blogs of theirs and said, these people need some clue. And DNS is important to Colfer, right? Yes. So I talked to somebody who had worked at Colfer earlier, and he introduced me to Matthew. And Matthew had scheduled two interviews within 15 minutes of getting that email. That was super fast. But you had the experience, right? To contribute there. Well, that was the only fast part of my hiring process. Oh, the initial part. Yeah, the first two. And how was it like joining the company at the time? It was fabulous. And what were you doing initially at the company? So, yes, I walked in and I was told I was going to be the team lead for the DNS product. Oh, good fit. Basically, it was two people in the beginning. And we were both just hired. But we're given this product because the person who had been maintaining the DNS server, he was given a new task that was more fun. And he went on to create the original DOS protection system for Colfer Gatebot. That was important at the time, right? And Dilip and I did take on doing some of the work that the company wanted us to do, and also work on making the system more stable than it was before. Colfer was a very immature company at the time. And small, right? It was just about 100 people. Very quickly, you got to know everybody. We had two offices, one in San Francisco, one in London. And I was remote. The only one. Except for a couple of salespeople. Although John Graham-Cumming, our CTO, also started in London as remote, in a sense. As the only one in London. Yes, and then he hired some people there, and an office was created. And I spent a lot of time in London over the years. For that reason, to be a part of an office, everyday type of work, in a sense? Yes, and information sharing, learning from others, hiring. Because interviews were either done on Skype or in person. And one of the things I was tasked to do was to bring on more staff. And build the team up. And we worked on that. And yes, there were some parts of the system that were inherited. And they were not working well. So we had to replace a whole system with a new one. First time the company really undertook that without shutting down and starting up an hour later. And so we did that on the fly. And nobody believed, or a few people believed, that we could pull it off without customers noticing. But they didn't notice. So those were the first missions, right? Setting new standards, actually, in terms of how things work and are not disrupted. But also hiring, building the teams. Those were the first missions. And along came challenges, new positions, possibly, also, in terms of your role at the company? Yes. So one of the things that I really like dealing with is what I call scaling challenges. When you hit a certain, when you build a system, it works fine for a while. Cloudflare was growing fast, or is growing fast. So systems, once in a while, you discover that what you've been doing doesn't work anymore. Or it's not going to work much longer. Because there's just too much data, there's too many requests, there is not too well written, whatever. Not that there was anything wrong with the system that was developed. It was perfectly fine for when it was written. And so I took that on with the DNS team. And after we had made the DNS team system quite stable, I could go on to the second mission they had given me when I was hired, which was to bring a public resolver to Cloudflare. And that was a big mark, right? A big step. Yes. So that was a very interesting process. We wanted good addresses. And there were thrown many ideas about. And Martin Levy, another old timer at Cloudflare and I, were able to strike an agreement for the company with APNIC to borrow addresses that had been deemed unusable. Because they got too much junk traffic. We pointed out that what was considered too much junk traffic was not a problem for us as a company. And so we got it. We set up a resource agreement with APNIC. And we got this service. Rolled out. And that's today. It's a service used by millions of people everywhere in the world. Well, it is the second largest public resolver in the world. After Google's, right? Yes. And it's growing. It's in a tendency of growth, which is amazing. It is growing amazingly. Given all of the... It is... Eyeballs are frequently under the hostess of their ISPs of where their traffic goes. And it's only technically savvy customers that really change the traffic. And then companies and others. So I was very proud last summer. When I discovered that some hotel I was staying in. Had actually configured quad one as their default resolver. Along with quad eight. Google. But I did not have. So that was interesting. This is the blog post you wrote in the 1st of April of 2018. Which was a good day. A fool's day in a sense. And not a joke. So it was a good mention to say this is not a joke. But this was the announcement of that public resolver. Quad one. That you were mentioning. Yes. And it was a very intense effort to get it done. So this was actually the project itself. It started a little bit by chance. Is that right? No, it didn't start by chance. The decision of making a public resolver. No. That is actually something that Matthew and Dane. Both had asked me. During my interview process. What it would take to set up a public resolver. So they were thinking ahead. They were thinking ahead. But the challenges of putting it together were big, right? Yeah. So first when we once I had made. We had made an assessment of what the status was. Of our authoritative DNS. Which is essential for the business. I said, let's fix that first. And then go with it here. To more consumer side in the sense. Yes. And why the 1.1.1.1 perspective? Well, it was like I said before. It was a dirty address. There is a little company that you may have heard of. Called Cisco. All of their manuals use that as an example address. In every single one of their manuals. There is a recommendation for every ISP at the time. To say you should block. 1.1.1 slash 24. Because that's just garbage. That's an example traffic. So we had to go out. As a part of the rollout. To demonstrate that a large portion of the Internet. Could actually get to us. And we discovered a whole set of networks. That couldn't reach us. And then we had to reach out to them and say. Hey guys. Don't block this. This is a good address. It's a good service. It's a real service. Yes, it's real. And then I started announcing the addresses. To be able to test. Without anybody telling anybody what's going on. So that a rumor mill of the Internet. Noticed this. So there was a lot of rumors. What was going on. And then we started seeing. All kinds of other garbage traffic coming to us. For example. We discovered that. Some. People had. Databases. That they were. Sending logs from. And they were sending it to the wrong place. They were sending it to that. Because they just want to throw them away. They couldn't figure out the disable the log. So they sent them to quad one. We discovered the great viral of China. You gave out this address. To users in China. For various sites. So yes. We discovered an awful lot of these. Dirty underbelly of the Internet. Makes sense. Again. A lot of work after that to unblock. To make it available. Also right. And about two months later. In end of June. A. Cloudflare engineer. Makes a mistake. And. And Gatebot decides to drop all traffic. Or resolver. But that meant putting it offline in a sense. Yes. It was offline for about 14 minutes. That was a turning point for me at Cloudflare. Why? It was decided by higher ups. That. The way we were running. Or at all systems. Was not scalable. And we were looking at. The techniques we had developed. They had hit the scaling point. That we're not going to take us forward. We would have to go to a new tech. New. So they gave me. The task of creating the next generation. Of DOS system for Cloudflare. And at the same time. Also implement a layer for load balancer. What became Unimog. So. Those really. Those two are really important of course. For security reasons. Also. Without them. Cloudflare today would not be functioning. Exactly. And how was those tasks? To follow through on those tasks. A lot of hiring too, I believe. We. Had to do an awful lot of hiring. Mainly in London. Where I put most of this work. And. We're able to hire. Really, really good people. Both internally and externally. And we. On our DOS systems. We were able to hit the ground running. By the end of the year. And there is an awful lot of legacy. That had to be done. And the team. We deployed the level four drop. Which is the new technology. For dropping traffic. Instead of gate bot. Then we started working on fixing up. Or systems in core. That were. Struggling with the load. Because they were written in Python. And the load we were doing. Yeah. And so. And the Unibug. It was very important to get it done. Well. And. There was expectations from higher management. It would be done quickly. And. I think that was. A challenge. The sometimes. Make it very quick. Is not making very good. Right. So there's a balance there. That has to be reached. Yes. And. Layer four load balancers. Were quite new at that time. So we did a fair amount of research. Of figuring out what technology. We want to use. Which way we wanted. And we realized that Cloudflare. Is very different. From the companies. That had done it before. And we had constraints. That they didn't have. They had things they wanted to do. And we didn't care about. So. We had to build our own system. Largely from ground up. We have way too many. Make it from the starts. In a sense. Yes. We had way too many services. We had way too many addresses. That we have to manage. And yeah. In that regard. But. It came to be. So it all. The pieces were there. And it was brought to. To success. Let's call it like that. Right. Yes. Because. Of the wonderful people. That we hired. To work on these problems. We have some amazing people. Working on these teams. You already said. Like very different moments. At the company. Any. In terms of management. Engineering. Things like that. Managing products. Making changes. Needed changes. From what was before. And any lessons learned. A type of. Observation you have there. In terms of. Those processes. It is. Every person is different. And the best thing. That a manager can do. Is to keep people happy. And figure out what motivates them. And dealing with the different personalities. I have a relatively simple. Philosophy. That I use in management. I say. I'm in the rent-a-brain business. If the brain is happy. It stays. If the brain is unhappy. It leaves. That's true. Motivation again. It's really. Really important. For a person. Right. But at the same time. You have to tell them. Your employees. That you are going to get some. Good things to do. You're going to get some. Shitty things to do. There needs to be a balance. Also there. Somebody has to do the shitty things. And it's between you. And your managers. To get the right balance of that. And. And. You have to make sure. That everybody knows. That is the general concept. That everybody gets to do. The shitty things. No matter how important. They think of themselves. Or whether it is worth their while. To do it. So basically. You have to break the ego. Of the biggest. Prima donnas. Once in a while. Makes sense. Just to break up the morale of the others. Of course. Of course. Makes sense. More on the. High level perspective. After. So. You've been. VP of engineering. For a while now. For a number of years. After that. Job. For two years. After the. Quad one. Perspective. The security. DDoS. That you were mentioning. What came next? What was the next steps there? Well. I had been a director. Since the early. I think. 18. Yeah. 18. After we did the resolver. And I was working. Really hard. On the DDoS systems. And. We signed up a new partner. In China. And we. Wanted to deploy. A system server. The way we had dealt. With the old partner was. In my mind. Not scalable. Everybody hated it. Et cetera. So. My boss was trying to figure out. How we were going to run. This project. And I went to him. And I said. Okay. I will run it. But you give me a total. Dictatorial. Control. To be able to do. What you want. Yes. And after a while. He agreed. And we started our project. And. Adrian. There were. Yeah. There were people who didn't see the light. In the beginning. But. Eventually. They all came to realize. Yes. We're not going to do it the old way. We're going to do it the right way. How important is. To sometimes. Go a little bit. Back. Start things. Over. To make it right. To. To make it. Scalable. Like you were. You were saying. You had. A lot of. That experience. Yes. Before. Right. Yes. And. So. Like I said. One of the. First things. We did. I did. That cloud. Was to replace. A system. That was not working. Well. That is the system. That copied. The data. That customers. Put in the DNS. And. We copy to the ads. That was. The. Exporting function. Was not working. Well. From the database. So. We rewrote it. We have rewritten. That system. Three times. Since that. It's the. Iteration. Iteration. Iteration. Yeah. The basic principle. We put in. The database. In the first place. Worked. Then. We have discovered. The number. Of hurdles. That. Where it didn't scale. We had to. Work. Talk differently. To the database. We had. When the databases. Get replicated. Things changed. We moved. Yes. So. We have modified. That system. A lot. Over the years. So. Yes. In a growing company. Like. Cloudflare. You have to. Assume. Every system. Is going to get replaced. Eventually. Or. Regularly. That is true. That. Keeps you. Honest. Right. Yes. Which. Is a good thing. In this case. Yes. So. Yeah. And. Yeah. So. If I didn't have something. To replace. All the time. I would be bored. So. So. Always. New challenges. Ahead. You already. Spoken. With the next scaling problem. Makes sense. Makes sense. You already. Spoke. A lot of different. Challenging times. Sex. Sex moments. What would. You say. Were your major. Achievements. During your time. Almost 10 years now. Well. The thing. That has surprised. Me most. Is. That I have actually. Been able to thrive. Here. Why. That's surprising. You never know. That's true. No. There are different personalities. There are different scales. When you make mistakes. What happens to you. Then. Would. Yeah. I never expected. I would be a vice president. When I started. I figured. I would get. To hang in my little corner. And be my. Or run my little team. But. As the company. Has grown. I say. I have floated. With the growth. Hmm. So. So. My distance. To the CEO. Has basically remained. Constant. Even though new layers. Are introduced. Makes sense. Makes sense. And. Of course. The importance. Of what you were doing. Was very big. Right. So. That's. That's why you had. That. Those types of roles. Of managing things. But you were put. To the test. Because there is new tasks. And you were put. To the test. Constantly. Right. Yes. So. The company. Has given me. The opportunities. And. I haven't messed up. Too badly. Not at all. Not at all. Any favorite moments. You want. To highlight. Specifically. One of the things. We. Decided. Was. To see. If. We could help. The. Some root server. Operators. To make. The DNS. Root system. Better. We could. We have. Lots of. Any kind of instances. Everywhere around. We have. Lots of. Capacity. We have. Lots of bandwidth. We really can. Absorb. Massive. Amounts. Of. Data. So. And at that time. People were very. Worried about it. So we approached. A number of them said. Hey. We will do this for you. And. We got. To. To sign up. And going through. The process. Of. Configuring. Our systems. With new. Addresses. And everything. Of course. Made. To tying. Into a new service. And then. Arguing. Sorry. Discussing. What was the safe way. To turn up. This critical infrastructure. So nobody would notice. And we had. A big debate. About. Okay. Are we going to do this. In the big locations. First. Are we going to do. In the small locations. First. Et cetera. Totally. Unknown. How to do it. We're picking up. A traffic. That. Would come to us. If we did it wrong. We would. Get. Flooded. And. When we turn it on. And. Just started. The lines. Start going up. Of traffic. No complaints. Nothing. No problems. That was. One of my. Prouder. Moments. And. What was. The year. 2007. I want to say. 2006. Late. 2006. Is when we. Started working. On it. And we. Did this early. In 2007. We have a blog. About it too. But. That. Everything. We. All the pain. That we went. Through. In that project. Led me. To. Later on. Sponsor. A new team. That. Is called the. Addressing team. Which is. About. Automating. How we operate. Or addressing infrastructure. At cloud. Without. All of the work. That they have done. That we wouldn't be able to do. Magic transit. For example. And. All of those. Are good. Bring your own. IP addresses. And. They have done. Such a good job. That. Yeah. Now. I can actually have them. Do the work. That I really wanted. Them to do. Before the team. Was formed. Makes sense. Makes sense. The. I have a. Two more questions. One. What's the main thing about. Qualifier. That most people. Don't realize. But they should. It's a fun place. To work at. Hmm. It is. Unusable. For many. Of the large. Internet companies. It is not. In the business. Of mining. The eyeballs. It is. Protected. By those. Who pay us. That they only get. Good traffic. So. We are trying. To be good. And. I believe. In that mission. To make the Internet. Better. For everybody. And make security. Also available. To anyone. That is part of it. The part. Yeah. The. I could. When I started. A class. Where I had. A very hard time. Describing. What I was doing. So I came up. With a metaphor. My job. Is to beat. The speed of light. Faster. For everybody. Around the world. And. As I have gained. More and more things. We are. Basically. Becoming. The filter. Of. Try to filter. Away. The. Badness. That comes. At many sites. For whatever reason. And. My daughter. When she was. A teenager. Used to say. Oh. So you are the quantum. For the Internet. That's a good one. Yes. A gatekeeper. Also. In a sense. I don't like. That word. Gatekeeper. Because it's. It's too much. A gate. And. No. We have. A system called gatekeeper. As a part of the old. System. Oh. So. When you say gatekeeper. I don't know. Which one you're talking about. The function. The. Or the old service. Makes sense. Also. What is the. For you. In terms of DNS. What's the main thing. People don't realize. That most people don't realize. About DNS. The domain name system. DNS. Is designed. As a loose. Coherency. Distributed database. And. The distributed. Means that. You can publish. Information. In one place. And eventually. It will find its way. All over the Internet. The. Contract. Convergence. Is controlled. By a field. In the protocol. That's called TTL. People think. The TTL. Is for caching. Purposes. No. It's for coherency. Reasons. So. When you set. The value. Of your. DNS entries. You're giving out. High. You are slowing. Down convergence. Around the world. If it is low. Convergence. Is very fast. So. Yes. So. People misunderstand. The TTL. All the time. Because they don't understand. The basic. Database theory. Behind. The DNS. And because DNS. Is a loose. Coherency. Database. It is. Successful. It is. Very hard. To do. A strict. Coherency. In a distributed. Protocol. Like that. So. That is the. Most. Important. Decision. That. Paul. MacArthur. Has made. When he designed. The system. Is. To keep it. This. Self. Converging. Consistence. Makes sense. And it was. One of those. Decisions. Right. That make things. Work. For future. Endeavors. In a sense. Yes. We will. Always have a fun. Fact. Usually at. Cloudflare. When you join. There wasn't. A fun fact. In place. A formal. Fun fact. In place. But you have. There was. Right. There was. But. I didn't have to give it. In San Francisco. Oh. Because. It was birthday week. When I joined. And the beer meeting. That I attended. Was. All about. The. Universal. SSL. Oh. It was. Getting around. At the time. But I had to give one. In London. Oh. And which one. Was it. I said. People were looking. At a. Former. Part-time. Owner. Of a. Premier League. League. Club. Oh. You. So you had like. A small. Take there. Yes. I. Owned. Stock in a company. That. Got West Ham. When the owner. Of West Ham. Went bankrupt. Oh. Back. Back. Back in the day. Right. Yes. And then. They sold it. Very quickly. Yes. Any. You know. Icelandic. Still. Any. Word. In Icelandic. You really. Enjoy. And like. And want to share. With the audience. Oh. That's a good question. Of course. The favorite word. Of the day. Is. I really like. Watching volcanic. Eruptions. And there is one. Going on. Right now. In Iceland. That's true. Beautiful. It's. Recurrent. In Iceland. Right. But. This one. Is a big one. Right. Yes. But. This one. Is not starting. In a mountain. So it hasn't been. Named yet. So. People don't have to. Worry about. Pronouncing. A. Good. Again. That was. A. Very easy word. To say. Yes. Yeah. I remember. Soon. After the. F. Erupted. That all these. Airline travel. Was. Interrupted. You're. Mostly right. I think. Yes. I arrived. In Holland. And. The immigration. Officer. Wouldn't let me in. Because. Because. Oh. That. Was. Very. Nice. Of them. Yeah. Because he had been stranded. On vacation. And. He also. Was complaining. That some. Icelandic. Bank. That. Branches. In Holland. Have fallen over. So I said. And he was. So I said. Well. We have. Misunderstood you. We thought. You said. Ask. Whether you said. Send us. Cash. Make. Sense. At the end. He was mad. Yeah. I pointed out to him. It was. It was not. The Icelanders. Fault. It was the banking. Really. Regulators. Fault. Before we go. To the Internet. Standardization. Part. Do you have. T -shirts. You want. To share. And show. Well. This is the T-shirt. I got. When I joined. Cloudflare. True. Any new. Product. A new T-shirt. Right. Whatever. They were in the offices. Well. I showed this one. Before. For my. Most famous. Project. This. Is another. Project. I was very proud. Of the. The new China. Part. Of the 80. Cloud. Everybody. Who worked. On that one. Got this one. So if you see. Them. Then you know. They did something. Of that. I used. To carry. All kinds. Of T -shirts. Around. In the company. But this. Is the one. That people. Love the most. Whoa. That's. A Superman. Like. One. Super geek. Trust me. I'm a super geek. I like that one. That's really. Great. Really. Yeah. Makes sense. That people. Would like that. Yeah. Really. Original. Yes. There was. For a while. We had. We had. A cloud. Operator. Really. Complicated. Systems. We have. Good people. In support. We have. Good people. On various. Teams. But sometimes. People. Don't understand. What's going on. And. When they run. Out of ideas. Of what to do. They start. Asking. For help. And. One other thing. That. Various people. Did was. Any problem. They couldn't figure out. They assigned it. To the DNS. Team. And. Because. The. Debugging. Problem. So. We used to call. The team. Debug. Nasty. Systems. And. One. Of my. Product. Managers. Gave me. That. That's a good one. Because. Usually. Actually. Now. The. It's. BGP. That has. The bad rap. Of. And. Things like that. Yes. So. This is the first part. Of my conversation. With Oliver. The second part. Is about DNS. Standardization. And you can find it. In our. We are called. Third. The show page. Oliver is retiring. In February. 2024. After decades. In the industry. And this is also. A big thank you. For all of his. Contributions. To Cloudflare. And the Internet. At the end. Of that other segment. There's also. This lessons learned. Bit. That we want to share. Here. Any. Final thoughts. Before we go. On. Your. Experience. Your amazing experience. Over the years. Like. The biggest lesson learned. You want to share. Possibly. With the audience. It's all about the humans. It's all about that. Everybody has a voice. But I tell. My teams. Whenever you're building. Working on a problem. You'd never know. Where the good idea. Is going to come from. So. It is the process. Of talking things through. Listening to each other. That makes the difference. Allows us to build. The best possible thing. Today. We are. I am. Fearful. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am. I am.