 from Las Vegas. It's theCUBE, covering AWS re-invent 2018. Brought to you by Amazon Web Services, Intel, and their ecosystem partners. Welcome back to Las Vegas everybody. I'm Dave Vellante with my co-host, David Floyer. You're watching theCUBE, the leader in live tech coverage. This is our third day of coverage at AWS re-invent 2018, our sixth year covering this event that keeps getting bigger and bigger, David. 53,000 people, places still jammed. We still barely have our voices. Adrian Cockroft is here. He's the vice president of cloud architecture and strategy, very well known in the industry. Cube alum, thanks so much for coming back on. Thank you. Yeah, it's a, I've been to all of the re-invents. First for as a customer, and then that was it. We've been to all but one, but we watched remotely and hung on every word. You know, back when there wasn't a lot of information about AWS, now it's like too much information to process. It's going to take us months to sort through it all, but at any rate, it's a phenomenal opportunity for us to learn, to share, to inspire folks, and you're doing some great work. Talk a little bit about some of the fun stuff you're working on in your current role. Yeah, I have a few different things I do. One is, one part of my role as I go around the world giving keynotes to AWS summits, but mostly I call it doing one of Vogel's impressions, because it's his deck, and I get to present it around the world. So we have to digest all of this stuff into a 90 minute deck that we can take to around the world. That's a, you know, what do you leave out? There's some, it's harder and harder every year. So that's a lot of fun. But the team that I run for AWS, I mean, recruiting and running is around open source. Right. And we do, we sponsor various events. We are members of various foundations. We make contributions to projects and have been helping that by hiring people from the open source communities into AWS to help some of the AWS service teams with their launches of open source related projects. So what I've got, what's been happening this year is I have like a hundred blog posts related to open source, lots of tweets, lots of activity, lots of events like OSCAN, All Things Open, and KubeCon. So we'll be there in a couple of weeks, chatting to you guys probably again. But this week there were a few of the launches where we got quite deeply involved. We did a blog post on the open source blog, launched at the same time as Jeff Farr. So okay, here's the service. And here's the open source part of it. And this is how you contribute. And this is what's going on. So we've had some fun with that. So was it two years ago when we first met? You've just been on a job for about a month, I think. A month or two. At that particular time. And you laid out what you wanted to do in terms of from your previous experience about how you wanted to turn AWS into an open source contributor. How would you rate yourself in the two years? I think we've made some good progress, really. I mean, AWS was making contributions to open source, but had nobody talking about it, and nobody, it was nobody's job to go out and explain what we were doing. So part of the problem two years ago, it was actually more happening than most people knew about. But we were just not telling the story. And it sort of wasn't coming across well. And the culture and the culture. I mean, it was spotty. Like some parts of AWS were doing a lot of open source. Other parts were kind of not really seeing it as a priority. So by talking a lot more about it, we kind of get a more uniform acceptance across AWS. It's a huge organization, not just there, but Amazon as a whole. We are actually telling the story, a much broader story than just AWS. And we're able to bring that and get everyone go, oh, I see everyone doing it, so I should be doing that. So it helps create the leadership for more teams to follow. And what we've seen, spent really the first year building the team, the last year kind of getting the content flowing and getting the processes kind of working to get all of the different events and blog posts and out to the outbound part. We're just getting increasing number of contributions and launches. So, now Coretto was a few weeks ago, it's not actually an AWS launch, but that was an example. That was, it's a lot happened from my team, from Arun Gupta on my team. He's a Java champion. He used to be at Sun. He was worked at Red Hat on J-Bar. So he's like, he knows everybody in Java, has great credibility across the Java community. And he said, we should launch this product in Belgium at like midnight or so West Coast time. And let's fly in James Gosling and like do a secret, like get him on stage without anyone knowing he's going to do it and do the introduction. So it was like, this totally crazy idea. And it came off beautifully. And we haven't had the, you know, the Oracle Java people saying nice things about the contributions to OpenJDK. So just a really nice example of figuring it out, all that get everybody on board, get everything done right and then say, here's something that matters to the community that we can contribute. Like the Beatles showing up on the rooftop. People are like, oh my God, James Gosling here. So everyone's got complete, like the star power thing. But convincing James to do it. Those are, are in a lot of credit for that, that particular launch. But you know, this is the kind of people I have on my team and we're like, we're pulling them in and pointing them at, okay, can you help this team figure out how to take this open source project to market? No, I mean, that was a major contribution to the open source community. And it was just in time, wasn't it? But another slight view might be that you and Oracle should have been working this out not leaving it until the last minute. Well, no, I mean, we were doing this work anyway. Right, okay. We were effectively self-supporting our own version of Java internally. We were getting better performance and better sooner bug fixtures on OpenJDK. So we'd made a decision to just move to the OpenJDK stream. And we were just unhooking our internal use of the other options. We had a whole mix, you know, a very large organization along with, you acquire lots of different versions and flavors of Java, even though it's just one language. So we're like, clean it up. Let's get JDK eight and 10. We're self-supporting it. And then we announced, okay, we'll support our Amazon Linux version. Right. And the final step was like, the customers were saying, please just like support it on my laptop and anywhere else I need it. And the thing we didn't announce then, we didn't make a big thing over there, and ARM support. We did. We kind of, it was in there by default. We didn't talk about it because the ARM chips came out this week. So yeah. And part of it is also have exactly the same version of Java now on all of the Amazon Linuxes, even the Intel, AMD and ARM. So that helps the compatibility for people kind of going, well, it's a different processor architect. You know, it all just ties together. So that was all part of the thinking there. You didn't want to tip your hand on the announcement this week is why, okay. So I think sometimes AWS is misunderstood, partly from its own doing. I mean, you just mentioned, you contribute a lot to open source, but you never talked about it. Generally when AWS doesn't have something to say, they don't say a lot about it. So others are left to, you know, make the narrative. You come on, you've now got an open source agenda. Can you just sort of summarize what that motivation is and what the objectives are? Well, we have, you know, lots of different pieces of this, but you have service teams saying, I'm going to launch this product and there's an open source component to it. Can you help? Sometimes that means I hire someone in my team to specialize in that area. Sometimes it's just our consulting with the team. We may be, you know, connecting them to the open source community. So that's one piece of it is having that. If you think about CNCF, in particular, Cloud Native Computing Foundation, that's got lots of projects. If you think about the AWS service teams, no one team really owns the scope of CNCF. But my team has that ownership for CNCF as a whole. We have the board seat position and we say, okay, we have the serverless people over here. We've got some insane things over here. There's some Linux kernel virtualization bits here. We can reach out to lots of different teams across AWS, but act as a central point where if you have something about open source you want to talk about with AWS or Amazon even as a whole, you can come to us and we'll find the right people and we'll help you make those connections. So part of it is acting as an on-ramp for the sort of buffer between the internal, the external concerns of the communities as somewhere to go. And partly just getting contributions out there. And we can criticize for not making enough contributions. Well, we've been making more and we're making more. We'll just keep making more contributions until people give credit for it. And that's the, if you're like, what's the strategy? Contribute more. And then point at it and hope that people like what we did and take the input. It's the customer-driven thing, right? We're going to do what our customers ask us to do and we're focused on the things we want to do. And we've been contributing to Spinnaker, the Netflix OSS project. We made some serious contributions to that in this year. Firecracker, as we talk about that a bit. And the RoboMaker. Those are all areas where we've been working with. Firecracker is particularly interesting, isn't it? I mean, that's a major contribution of improving the performance and capability of those micro-VMs. Can you talk about that a little bit? That seems to be the... It's interesting because it's a piece of software pretty much no one will ever see or use. It's the thing you run on the bare metal that lets you run your container D, that lets you run your container on top, right? It's deep down in the guts of the system. There's this piece of code. But we kind of... There's a few reasons. We're using it particularly in production now. It's supporting some of our production use of Fargate and Lambda. You're in the middle, it's not 100% rollout, but there's a good chunk of the capacity running on it. And that's where it turns out to be useful. And just how long we have to get into this. But if you think about a customer running a Lambda function, we would create a VM and we'd put that Lambda function in it. If they wanted a second Lambda function, we'd put it alongside that one. Another customer comes and we start a new VM for them and we start a Lambda function in that. VMs take a while to start up. So we have kind of a pre-made sum sitting there waiting. But these are big VMs and we're putting lots of little functions in them. What Firecracker lets you do is start a separate micro-VM for every function and safely put all of the customers on one machine. So you start packing them in. It's a much more efficient way to run your capacity. Our utilization of those machines supporting Lambda is vastly higher than having a machine with a bunch of empty space in it that we're trying to running for the customer. So it's that efficiency is the thing. And then the speed of starting a VM. It's a very cut down VM. So it's 125 milliseconds just to start the VM, which is incredibly fast. When you think, hey, give me a VM on EC2. It's different from like 30 seconds to a few minutes. It's like, I get 12 terabyte VM. Takes a little while to boot up. But you don't have to pay for it until it finished booting. That's always one of my good things about these huge machines, right? How about RoboMaker? Can you talk a little bit more about that and its importance? So RoboMaker is interesting. On the open source blog, which we posted on late on Sunday night, early on Monday morning, I did an interview with Brian Gerke, who's the founder of the Open Robotics Foundation. And what we've done there is, it's kind of an extension of SageMaker, if you think about that being AI. If you've got these eight, I can deploy an AI model. What does that AI model want to do? It wants to read something from the real world and then modify the real world. So it wants to read from a camera or at some other sensor and then control motors and servos. And that's what RoboMaker does. It wraps the intelligence you can build with SageMaker with the robotic operating system that has a library of actuators and a library of algorithms, control algorithms. You've got a little brain in the middle and you've got a new robot that does something. And we have the RoboRacer little racing car. So all of these things come together to make a little toy race car that we can drive around tracks with, which is a whole other topic we'll get into. But I interviewed Brian on what is the history of RAZ, the robotic operating system? Where did it come from? What is the hard thing about running in? It turns out the hard thing with RAZ wasn't building the robots, it was simulating the robots. And the simulator is quite a CPU intensive job. It's graphics intensive. You've got this virtual world you're running and VR worlds are quite intensive. And getting that installed and running was the hard part. So what RoboMaker is, is that as a service. It's this simulator. It's called Gazebo, which is a funny name. So Gazebo as a service is the actual service that effectively we're charging for with a free tier so you can play with it. And then we charge you for the simulation units, like how much computing time you're using. And the rest of it is all cloud nine for the front end and deployment of fleets of data to fleets of robots and updating them and managing them. But the interesting thing is this is getting into like the people that are the first robotic thing. It's high schools, high school robotic competitions. They're interested. Universities are interested in a university service. So we kind of, it's not just for commercial production robots. It's the whole training thing. We get into STEM education. Kids like playing with robots. It's like, and we're pulling all this in. So you now you can go home and take these like the latest most advanced AI algorithms that used to have to be doing a PhD at Stanford to be playing with and play with your kid over Christmas and see what you can come up with. Really simplifying the whole software development side of that. And when you look at the Dean Cayman competitions which are just awesome. You know all the kids, they could have gravitated to the hardware because they can touch the software was really hard. And this is going to, I think take a new level is particularly in this competition. And it's all open source. Like you can go, oh, you've got this robot. No, no, I pointed to them. Somebody who's complaining that we've done it with some proprietary robot thingy with the toy cars. And I pointed them at the get hub. You know, it's there, it's all there. You can go build this thing. It's all open source. You can put anything you want on it. But the little robot has rods on it. The robotic operating system, stage maker, robot maker all combined together. And they're off running races and having fun. Now you guys are both Formula One fans. And you guys have been having some profile of Formula One folks here. You've got the little mini vehicle. Riff on that for a little bit. It's not really open source, but... It's not open source, but I have another thing I'm doing on the side. It turns out that over the last year or so we started looking for opportunities to do sports sponsorship with a particular focus on Europe and the rest of the world. We had a few U.S. sports where they, I don't know, something with balls, I forget, baseball, basketball. I don't know. I'm English, I don't know. Fake football. Something like that. But I like sports with wheels, rather than sports with balls. That's my kind of thing. And Lewis Hamilton has done very well. Yeah, Lewis has done very well this year. So about the middle of last year, like last June we announced the deal with Formula One, which is a multi-part deal. Part of the deal was just take them to the cloud. They have some data center stuff. They were running out of space in their data center. It's like, no, they wanted to do a technology refresh. So for all the reasons that everyone else is moving to cloud, we moved the sports core infrastructure to cloud over some number of years. So that's a process we're starting. And part of that is the archive of all Formula One races. It's a treasure trove, like 67 years of archive of everything that got all the videos. We're digitizing it. We're going to figure out what to do. We've got to process it to label everything. Anyway, so that was one thing. And then we all turned up at Silverstone in the UK that race, it was the week after the announcement. And that race, we have AWS logos turning up on the screen because another piece was sponsorship. So we start sponsoring the core video feed that Formula One uses to the world. And that's 500 million fans watch Formula One. So now 500 million fans for the next few years, you're going to see AWS logos on screen around the analytical insights of what is going on in the sport. Like, your rear tires are overheating. You went around a corner this fast. Here's the pit stop strategy. So we get brand advertising associated with a high technology sport and analytical insights. And that's why we did that deal. And they get all of our technology, AI, a lot of help helping them migrate. And then the third thing we did that I got involved with was I'd already done a few CIO summits at Formula One races along the way. So I was kind of like trying to poke my way into this thing that was happening. I'm not involved in sponsorship setup, right? So hang on, have you done that thing yet? So we decided to do some executive events around Formula One. So we'll pick a few races. We'll have some corporate hospitality like things. But when you put a bunch of senior executives together for a few days, they share, they solve each other's problems and you just get out of the way. And the people that have solved one problem will share it with the other. So it's a really, it's like a tiny reinvent. Like here, everyone is sharing. If you sit next to someone, what problem have you solved? You can find stuff out. This is a concentrated version of that. And we tiled it in Monza earlier this year. Went great. Amazing. I mean, it's fun. And it, you know, next to the business. Finally it was like, can we get someone on? Yeah. Can we get a car on stage at re-invent? Okay, so the last race was in Abu Dhabi on Sunday. On Sunday. So it said, can we get them on Sunday night for the launch, for the robots launch? No. They're in it, they're in Abu Dhabi. So Rose Braun, this is like the top guy in Formula One got here from Abu Dhabi by Wednesday morning. I was like, I'm just happy that they got here. Yeah. That was a huge ask. Well, Adrian, on behalf of the entire Cube team, we've watched your career. You've been somebody who shares his knowledge and done some great work. So thank you so much for coming back on theCUBE. Thank you. Congratulations on all your great work. Andy Jassy's coming up next. We're very excited about that. Keep it right there, everybody. We'll be back with our next guest, Andy Jassy, CEO of AWS, right to this short break. Thank you.