 Good morning. How's everyone doing? Good. Good. Awesome. All right. So we've had four years of Mesa's con so far. Four years in North America. We were in Chicago. Anybody here happen to be in any of the North American conferences? All right, there's a couple. Yeah, anybody in Chicago? Were you at the original one, Rich? That's amazing. That's awesome. And Vinod, awesome. Okay. We had one the year after that, we were in Seattle. The year after that in Denver. And just this year we had one in Los Angeles. So we had four years in North America. We've had three years in Europe. First year, 2015 Dublin. Anybody in Dublin? All right, got some folks there. Last year, Amsterdam, and this year Prague. So it's been a really, really fun run. We really enjoyed it. I also wanted to mention that this year we had our second Mesa's con Asia. Our first Mesa's con Asia, we did about a year ago around now. And then we did kind of a fast follow earlier this year in Beijing. And it was really fun to see the conference go from about two users presenting the work that they were doing, using Apache Mesa to over 14 different companies that presented the work that they're doing. So it was really, really, really fun to see the growth and adoption there. Okay. So we have a few additions to the conference this year. We decided to change a little bit based on feedback that we'd gotten from folks. The first is we introduced tracks to the conference. So basically, this is you can go into specific topics that you're interested in. The ones that we have this year are basically smack and machine learning. So the smack stack, which I'll talk about a little bit later. A DevOps plus ops track, which is just going to be today. A Mesa's internals track, which is just going to be today. And then a Mesa's frameworks track, which is tomorrow. And a users plus ops track, which is also tomorrow. So if you're interested, you can kind of just sit in the single room if that's just the topics you're interested in. Otherwise, you can bounce around. A couple other additions. We introduced something that we're calling Mesa's con university. So this is basically longer sessions where you bring your laptop, go and actually work on things, hack on things, go even deeper in the talks. Again, we heard from a bunch of folks that this kind of feedback is the second time we did it. We did it in Los Angeles as well. And it was hugely valuable. People really, really enjoyed it. So if you're interested in that, check it out. We've got three across the two days today at 1045, building production applications for maximum. Debug ability tomorrow, a smack stack one on one and also an intro to GPU isolation and TensorFlow. So if you're interested in that stuff, go check it out. There'll be longer sessions. Bring your laptops. All right. And then the last addition that we also added is this concept of town halls. So we felt like oftentimes we were having the conferences. We weren't spending enough time just talking together. And so we wanted to change that. So we added what we call our town halls there, 7 p.m. tonight after the booth crawl. So go have some beers to check out the sponsors and then come. We're going to have four. We're going to have an Apache Mesos one. So internals of Apache Mesos. That's going to be in Congress Hall one. We're going to have one on DCOS. That's going to be in Congress Hall three. One on Marathon and Chronos. That's going to be in Congress Hall two. And one on Kubernetes, which is going to be in Rome of Vienna, which I think is just around the corner here. So please come check those out. All right. You know, big thanks to all of our sponsors, but also a huge thanks to all our speakers. We have an amazing group of individuals across numerous different organizations. So it's really, really fun to see such a diverse and eclectic group. I'm really excited. I almost didn't believe that I could fit everything on the slide here. So really, really glad to have you all here. Okay. Getting into some of the good stuff. The state of the project. So since the last MesosCon Europe, we've had four minor releases and nine patch releases. We just done a 1.0 going into MesosCon Europe last year. You know, and since then, we've also added 13 new committers and PMC members for our Apache project. Everyone that we bring in as a committer, we also make a PMC member. So it's great to see all of those new committers and PMC members. And it's also great to see all the releases that we're doing. One of the other things that when I was asking some folks in the community what they'd like to me to share about the community, one of the things they said was they've been really, really excited to see the growth in Slack activity, more and more people there asking more questions, getting more answers. So I wanted to share that as well. Okay. When I also asked folks, hey, what should I share? What's new and noteworthy? They gave me a huge list. Obviously, I can't cover everything from that list in a few minutes. But I did want to call out a couple of the things that I was personally pretty excited about. And I think others in the community were as well. The first one I really wanted to call out is all the work we've been doing around containerization. So many of you know that in the Apache Mesos project, we started doing containerization a long time ago, back in about 2010. In fact, the very first version of Apache Mesos use Solaris zones and projects if you want to run on Solaris to actually get containerization. Over the years, we've gone through many different iterations of how we wanted to do containers. We've done C groups and namespaces ourselves. We had a Docker support in 2014. And in 2016 and 2017, one of the things we really focused on was making that initial containerization work that we'd done even more robust and to be able to support even more kinds of containers you wanted to run. So we introduced something that historically we'd always called the Mesos containerizer. We called it the universal container runtime. And it lets you run containers with a Docker image, with an app-c image, with an OCI image almost. I think that might be committed now. Or without any image, just with a zip file or a tar ball or a jar file. So it's the core components of containerization that we've had in Mesos for a very, very long time that now you can use to also run your Docker images or any other images that you want to run. So one of the really cool things about the containerization is it was really easy for us to add some more sophisticated support for doing other things. So one of the things that we added is this concept of nested containerization, which is, again, a feature that we've talked with a bunch of people about that they said they were really interested in. So the idea is when you run containers, you could then run containers within containers and containers within containers. And I think there is probably a kernel limit. There's only one place left in software where you can write limits. And I feel like that's the Linux kernel where you can write constants like 32 or 64. And so I'm pretty sure there is a limit on the number of containers you can have. But you can do really amazing things where you can launch your very first top level container as a zip file, have that itself launch a nested container from a Docker image, have that itself launch another container, which is an OCI image. And there's lots of really great practical applications for this. One of the big ones that we're excited about is things like Jenkins. When you go to deploy Jenkins, Jenkins itself wants to run a build, which is a Docker image. And maybe that Docker image itself needs to create yet another Docker image. In the past, what would end up happening is these Docker containers would all potentially be top level containers and could become orphans. Now we can contain them all really, really nicely. This is also how we run Kubernetes on top of Mesos. So you'll hear a little bit about that later. So I won't go into too much detail. Okay. And the last really cool thing about this is the top level container doesn't actually need to be privileged, which is really cool. You don't have to worry about any sort of security issues there. We also added the concepts of container, attach and container exec. So it was really, really easy for you to get information out of the containers. To hear more about all the stuff that we're actually doing in containers, you can check out a talk, a containerization in Mesos embracing the standards. It's later today. One last thing I want to mention is we're also out in support for standalone containers, which is going to make it even easier for running something like Mesos on top of Windows. So I'm really excited to see that work in progress right now. Okay. A big part of containerization is networking and storage. And one of the things that we've been really, really excited to see the community embrace is a lot of these more emerging standard-like things for doing stuff like networking and storage. The two big ones for us is CNI, the container networking interface, and CSI, the container storage interface. And I'm especially really excited that we've had folks from our community, G, U, and James, Dave Felice, that have really been helping to drive the container storage interface forward along with various other open source communities, including Docker, Cloud Foundry, Kubernetes, and others. So that's been really, really exciting to see. There's two talks where you can check out the work that's going on around networking and storage. Both are today, I believe they're in the Mesos internals track. Okay. One of the really cool things we're getting out of the storage work is this concept, this future concept of being able to extend the resource management capabilities inside of Mesos. So we're adding this concept of local and external resource providers. We're going to use them initially for all the storage work, but I think the sky's the limit in terms of the different things we might be able to apply here, how we can use this. We can make things like an external resource provider be a cloud, for example, where resources can be dynamically added in and removed. All right. I mean, we did a survey last year, a patching Mesos survey where we gathered a bunch of information about how Mesos is being used. And one of the things we heard is it's being run both on-prem as well as in the cloud. So there's a lot more work that we're doing around the project when it comes to fault domains, hybrid support, multi-cloud, and there is one talk you can check out later today, fault domains in Mesos where we talk a little bit about that and you get a glimpse into what the future looks like when it comes to hybrid and multi-cloud for Mesos. You know, one of the other big areas that we've talked about in the survey was how people are actually, what things people are actually running on top of Mesos. Of course, many of you know the Mesos architecture is based on this concept of two-level scheduling and multiple frameworks and many people run many, many different frameworks on top, which is really driven for us this whole notion of the future world of the smack stack. So many of you, I'm sure, have heard of smack, Spark, Mesos, Acca, Cassandra, and Acca. There's various different frameworks that people are actually running on top, but these are really the frameworks that we're seeing the most and for a lot of organizations that are adopting this, they're seeing the same kind of wins out of adopting something like the smack stack as they got when they adopted something like LAMP. So there's a bunch of different talks today from organizations that are talking about running things like Kafka and Spark and Cassandra on top of Mesos. The one thing I wanted to mention that we're really excited about is when you're running these very disparate frameworks, there's a lot more multi-tenancy concerns that really come up and there's a bunch of work that we're doing in Mesos itself to address that. You know, the core for us is things like hierarchical fair shares, hierarchical quotas, and hierarchical reservations, where you can get much finer control around how many resources you want these various frameworks or these various users to actually be consuming within your cluster. So it's really cool to see all this work happening as well. The hierarchical fair share work and the hierarchical reservations and reservation refinement is all things you can use in your Mesos clusters today, and the hierarchical quota work continues to be work in progress. All right. So a bunch of great presentations. I think you can check out over the course of the next two days, a bunch of really, really interesting work. Go check out the town halls. Please try out the MesosCon University. I thought I'd end with just a couple of fun predictions. These are personal predictions, things I would love to see in the project. Along with all the other multi-tenancy work that we're doing, it'd be great to see the concept of revocable resources allocated by default. We're looking forward to potentially being able to remove the zookeeper dependency, so it's even easier to deploy a Mesos cluster. And who knows, maybe in the future we won't just be launching containers, but we'll also make it even easier to launch VMs. So that's all I had as a welcome. I hope everyone has a great MesosCon, and we'll see you again next year.