 Bonjour. I'm very sorry for my French accent. Je suis désolé. So I'm so happy to be here. I'm so grateful for the chance to come and talk about the early days of Kubernetes. Kubernetes has had a profound effect on my career, a profound effect on the careers of a bunch of people I know. It's been a few years since I was here at KubeCon because of COVID. And it has been so wonderful to be here, reconnect with old friends, meet some new ones. So my talk here, I was thinking about it's easy to believe that from all the people here and the success and all the amazing things that have happened with Kubernetes and all of the ecosystem of projects around it, that success was guaranteed. But it wasn't. And there were definitely some moments. And I wanted to talk about some of those moments. This is maybe a little bit too dramatic of a title for this talk, but this really is stories I like to tell at dinner about Kubernetes. And after this, I really am going to have to find some new stories to tell because I will have actually told everyone. So very good. All right, so 2014, I was working at a small carrier. And we were doing edge deployments. We were building a thing that we weren't even really calling cloud at that point. And this t-shirt, which I'm sure many of you see, this was our lives. It worked on my machine. It worked in development. It didn't work in production. We were having lots of operational issues. We built a whole orchestration system. We weren't using containers. So this t-shirt was definitely our lives. So I had been invited by Google to come and give them a talk. This was before GCE launched about what it was like to build and run applications on the public cloud. And as a result of going and doing that little talk, I met Joe. And then Joe was telling me about this thing they were working on, introduced me to Brendan. I was involved. I'm still involved in this really amazing organization in Seattle, Seattle CTO club. So we asked Brendan to come give us talk. And I don't know. There was maybe a dozen of us in the room. It's a pretty small group. And so he's talking about it. And the light bulb went off. We had had this big orchestration problem. And at the same time, there was this guy on the team there at this little carrier working with Aaron Krickenberger. He eventually was on the steering committee. And he's talking in the office about there's this new docker thing. Wouldn't this be amazing? So the light bulb went off of like, oh, there's this thing coming. And it could really help us. And so the first time I ever installed Kubernetes was we had a little set of Intel knocks hanging around the office. I installed it. And then I really knew that we were on something. So after that, I went to work for Samsung. We didn't call it this to begin with. But eventually, we figured out that what we should do is call this team the Cloud Native Computing Team. I was hired to be what we called, ended up calling the Cloud Native CTO. This is a picture of the initial team. There's Aaron again. And the gentleman standing there in the back, Sean Kay, who's the president CEO of Samsung SDS Americas. And so he came out to kick us off. We built this little team, this little office near Fremont so that we could be near the Google office. And that had two advantages. We wanted to make sure we could go and talk to Kubernetes folks there at Google anytime we wanted. And we also loved scheduling meetings with them at lunchtime or breakfast. So we could go score off the amazing free food at Google whenever possible. So at any rate, we did a full evaluation. We looked at everything in the very early days. We knew that what we were doing at Samsung was charting the course for the next five or 10 years back then. And it was really important that we make a choice for the future. And we thought Kubernetes was going to be a thing. But there were a couple of obstacles. And this is where maybe things could have gone a little bit differently from those days. So two things, one of these a little bit more technical and one of these a little bit more organizational. So the technical thing that we had was we knew that to use Kubernetes at Samsung scale or at any big scale for a big enterprise, that we were going to need to run some really big clusters, that we were going to need to run a lot of Kubernetes. And the project at that time was testing to 100 nodes. Like 100 nodes was considered to be the limit. And I cannot tell you how many times I had conversations with people about, oh, Kubernetes is only going to ever work for 100 nodes. It was designed for 100 nodes. And so I'd say good open source behavior here is we decided that we were going to engage. And so we had actually hired Joe as a consultant just before he did the Heptio thing. And he and I started the first SIG in Kubernetes, SIG Scalability. And it was because we knew how important it was that we'd be able to really run Kubernetes at scale over time. And the couple of slides I have here were from the days when I was SIG chair and I would put together the slides and we'd go through and discuss. And these are, I found these in my Google Drive. And you can see here, maybe the font's a little small. The main thing we were worried about was Kubernetes is only good for 100 nodes. Meme, which was everywhere. So we put a lot of effort into that. I think the Samsung team, we decided as one of our goals we were going to start running 1,000 node clusters. This is in the very early days, even like three 1.0 territory. And so I think we were the first team to actually run a 1,000 node Kubernetes cluster. And that was a compliment. We were very proud of it in those days. Okay, so second issue. This was a bigger one and a more important one, which is we were not willing to make a bet on Kubernetes unless it was community governed. So, Kubernetes was open source. I think this is true to this day. Open source is not enough. Open governance really, really matters. And Kubernetes was always a multi-org project, a multi-company project. So many companies involved in it. One of the folks I really wanna make a shout out to here is Clayton Coleman, who really built the confidence with us that you could be a leader in the project and he's influential to this day. Works for Google now. But Clayton was really the one that convinced us that we could not only use Kubernetes, be part of the community, but we could be a leader there if we needed to as well. So we pushed Google pretty hard to take what was I'm sure a hard step for them, which was not just to have open sourced it, but to actually release it to the community. And the foundation, the CNCF Foundation, which Chris was just talking about was part of that. So thank you to Google and all the folks that were involved in that. That was a big moment, that's what. And I've got one more moment that derived from that decision as well. And so as a result of that, we were able to make this big bet and Kubernetes was very excited to be asked to be part of the Kubernetes 1.0 launch. This T-shirt I'm wearing is the T-shirt I was wearing on stage at the Kubernetes 1.0 launch. So I've worn this T-shirt twice. Once then, and I found it in my special swag drawer, so I thought it would be fun to bring it out here. And then standing on stage there with me is Richard Kaufman, who is also part of the cloud native team there. So I dragged a bunch of knocks out on stage. I was so nervous about wifi at the conference that we dragged a whole pile of cabling out and I demoed Kubernetes live and it worked. So very good. All right, so another story here. So there was in those days some tension going on between the Kubernetes and Docker communities and I was definitely a participant in that. I wrote this piece which kind of went viral talking about should we fork Docker? Obviously that would have been a bad thing to have happened and I think this was one of the contributions. Dan made many contributions but one of the big contributions he made was he pulled together a nice collaboration in Seattle that included me and a bunch of other folks from Docker, Kubernetes, Mesosphere, the AWS ECS folks were there, pulled together a nice collaboration and we were able to work through all those issues. So I think thanks to Dan, I think this is one of the really important things that the CNCF, CNCF leadership, folks like Dan, folks like Priyanka really bring to the community to help us all work together. So thanks to Dan and thanks to CNCF for doing this. This is also another really interesting moment because I think this was the first real specification driven thing that we did here in the Kubernetes community. More broadly, upstream is great, forking is not a good idea, but you also sometimes need specifications that ever we can run to and the OCI was really a critical specification. Okay, so open governance meant that AWS was willing to launch Kubernetes service and so they had hired me to be the GM of EKS and we built it and launched it. I have here some pictures that were behind the scenes of the launch. We launched EKS not at reinvent, which is where AWS loves to launch things, but we couldn't wait. And so we did what I think was the first ever live streaming on Twitch product, major product launch at AWS and so you're seeing the behind the scenes here and the little studio. And I think a couple of things to point out here on this picture. One is there's Arun Gupta, who's now the chairman of the governing board, CNCF. So he was there and part of the team. And the other person I would really like to call out here is Yusor Bala, who ran the initial engineering team runs. EKS engineering to this day did just and that's such an amazing, fabulous job of leading that team and had such a huge impact on Kubernetes, on EKS. And I think he deserves a lot more recognition than he gets. So that was like, this was a great moment. And again, without open governance, we could have never done this. And EKS was the last of the big three cloud providers to launch a Kubernetes service. And I think that was really the moment where everyone realized yes, Kubernetes is really the standard. That was the final moment in that journey. So I wanted to use my moment on stage here since I have the mic to give some recognition to the BuildPacks team. I see Terrence sitting down there. So thanks Terrence for all the leadership, open source leadership you bring to the Heroku team. So, and by the way, there's both Heroku contributors and Heroku is part of Salesforce. They're both Heroku contributors and Salesforce contributors to BuildPacks. So we're very excited to support this. And if you have been tracking the work that we've done here, been doing here, you may have been wondering if we were going to do this. And so I'm letting you all know today here that we are in fact rebasing Heroku on Kubernetes. This is a project that's in progress. Heroku was one of the OGs of cloud native. And for Heroku at least, I would say the circle is now complete. Heroku had a lot of influence, I think, on Kubernetes and some of the other systems that came along. And now Heroku is rebasing on Kubernetes and we're very excited about this project. So I just wanted to say thank you again CNCF for this chance to come up and tell some of my stories. But I think you all need to thank yourselves as well because there has been so much toil that has been saved, so many pagers that have not gone off, so many systems that have scaled up and down. I think that everyone here needs to take a moment and celebrate for yourselves all of the good things that you do, all of the innovation that's happened here around this. So give yourselves a hand, please, please. Thank you, thank you all. Awesome.