 Live from Barcelona, Spain, it's theCUBE, covering KubeCon CloudNativeCon Europe 2019, brought to you by Red Hat, the CloudNative Computing Foundation, and ecosystem partners. Welcome back to Barcelona, Spain. This is theCUBE's live coverage of KubeCon, CloudNativeCon 2019, 7,700 people in attendance, including myself, Stu Miniman, and my co-host, Corey Quinn, and returning to the program, Abby Fuller, who is the principal container, Zarina, at Amazon Web Services. Yeah, Abby, I could say it without laughing, but I don't think you can. Yeah, so, you know, Zarina, how does one become a Zarina in their career, Abby? Let's start with that. You ask Deepak really nicely, and he'll change your title for you. Longer answer, I think I'm doing a similar version of what I've always done for Amazon, which is, how can I get what customers are asking for and their feedback and what they're struggling with, they're working on or enjoying, making, taking that back to our internal product development process, and then doing the same thing back the other way. So if we're building something, how can I help educate customers on how to work with it and how to use it, how to build with it? So same thing, just funny your title. All right, well, Abby, you know, it's a big Cloud show, so of course we know Amazon will be here, lots of developers here at the show, lots of activity. Yesterday, AWS held a kind of pre-show workshop. Maybe start there, tell us a little bit about that. Yeah, so we had AWS Container Day. Maybe five, five or 600 people, we did it at the hotel that is allegedly across the street, but it's really like 25 minute walk away. We did some workshops, we did, we did a Birds of a Feather session at night. We had a little mini product preview announcement, so that was pretty fun, something called Container Insights from the CloudWatch team. I think my favorite thing about KubeCon is my favorite thing about the Kubernetes community, right? Which is that everyone is so happy to be here. They're all so enthusiastic. I've never had that many questions at a Birds of a Feather session before. We sent a ton of Amazon people here to kind of talk about EKS and Kubernetes and community work and the energy at the KubeCon is always a question. Give us a little sampling. You know, there's passion, there's other questions. Are they trying to understand the various pieces? Are they excited about some of the new features? What's some of the energy you're capturing? Yeah, I think it's both. I think on the EKS side, there's always the balance, right, in the Kubernetes community between how can I have more power and flexibility and then how can you carry the pager for more of this? So I think it's always an interesting balance between the folks that are like, do you think you could manage that for me as well? And the folks that are like, I want to be able to pass in control plane flags. So there's always an interesting balance. A lot of questions about version upgrades. I think that one is always seems to be top of mind because the Kubernetes community moves so fast. So compared to a lot of other products and how quickly they can release new versions, Kubernetes moves so fast. So if you don't have a good upgrade strategy, you're in trouble, so. Well, to that point, yesterday during the talk, there was a slide that went up that listed over the trailing 12 months that there were 1,900 in change, major service and feature releases. And that's very much a two-edge sword sitting in the audience. Because on the one hand, yay, the pace of innovation continues to increase and services are getting better all the time. On the other, it's one of those, at least four of those would have been critically important but I may not know about them. And to that end, something that the container group seems to have done that almost no one else has has been to put up a public roadmap of what's coming down the pike, which has been tremendously helpful for customers as far as being able to plan things out. How did that come to be? A lot of talking. I think ultimately, right, all teams at AWS work the same way, which is backwards from what the customer is asking for. So we have a lot of customer meetings, we have a lot of customer conversations, we talk to a lot of people, I do a lot with that on social media or at conferences or with blogs or with live streaming but ultimately, at the root of it, we all follow the same process. And I think the roadmap is really an extension of that. It's how could we get both what we're working on to customers a little bit faster but also, how can you have a voice that we hear so much more loudly so that you can be the smallest startup or the largest enterprise and you can open a GitHub issue just the same and say, hey, you know, I'd really like to see you do that. And I think the other piece of it is that everyone has an AWS story where they built something custom to work around something or to add a feature and then six weeks later, we're like, we shipped it. And that's awesome, it's a good problem to have and being able to delete code is one of everyone's favorite problems, I think. It's one of life's true joys. It is one of life's true joys. But what I think is even better than that is a little bit of heads up and I think that that really builds trust between us and the community is how can we let you know we're working on so that you can plan around it or if you don't see something, let us know that we're not thinking about the things that you value. So Abby, you know, we've been at the Amazon shows for a number of years and that customer feedback loop is something that we hear a lot. Are there any dynamics about just being in a big open source community here is just listening in feedback loops as part of that? So how does that impact how you work on things? Yeah, so when we do events like this, I try to talk to as many people as possible. I try to listen into the conversations when I can. People come by the booth, they come by the meeting rooms and I think it's about taking that back all the different sources that were at the conference, the reviews online, the blog posts that people were right after this, coverage like theCUBE, taking that all back and then let's go through it and how many of these things do we know about? Because I have a lot of people ask us for this. Is this something new? If it is new, how can we go find other people to talk to to see who else is having that problem that maybe we just didn't know to ask about before? So it's all part of that same working backwards process but feedback comes from so many different places and I think that ultimately is what makes it cool, right? It's because you get different feedback at a KubeCon than you will at a re-invent, than you will on a Twitter or that you will at a customer meeting so you need all of those sources to kind of figure out what's more important and who is it important to? Yeah, one of the things that I find fascinating about the entire AWS container story is you almost get to decide your own level of involvement. You can run it all yourself on top of EC2, you can wind up doing one of the managed services with ECS or EKS and then there's Fargate which I'm very bullish on for the future if for no other reason that if that takes over suddenly we will never have to hear someone from Amazon mispronounce AMI ever again which I'll take my victories where I can find them. But what are you seeing customers doing with Fargate? How what's the paradigm look like that's different than you might have expected at launch? Yeah, so the way that I ultimately think about Fargate is it's a capacity provider for EC2. So when you think about kind of the levels of control you start at maybe the orchestrator level so an ECS or an EKS and if you're using ECS through Fargate you're not interacting directly with EC2 so it's about how can I control and define everything at just the container level just at the task definition level without having to think about the underlying EC2 instances and they're still there before someone tells me that server lists still have servers but you're not the one that's actively managing them. We're managing them on your behalf. All you care about is your workload itself and then you can go a step deeper than that and say, you know what? I want control over those EC2 instances. I want to manage them myself. Maybe I want to do something in user data or I want to be able to run demon sets myself on the underlying infrastructure and that's fine. So I think it's ultimately about the level of control that you want and Fargate to me is interesting because it's like Lambda in the sense that people have seemed very joyful about not having to manage EC2 because ultimately that's not what's providing the business value. That's not what lets them differentiate. I think the way that Werner puts it is you want everything that you write to be business logic and I think with things like Lambda and Fargate it gets you one step closer to that that instead of having to manage infrastructure to then manage your code it's just manage my codes please figure out the rest of it for me. This is borderline heresy in some circles so don't at me. But what I'm wondering is are things like containers and functions as a service aligned longer term on the same axis at some point where it just becomes an implementation detail and not a battle that needs to be fought? Yeah, the way that we think about it right is that and I think the way that customers see it is that serverless is ultimately a spectrum. There are many different flavors of it depends on how you kind of want to work with it but ultimately I think even longer term maybe this is even more heretical right but I want to not care. I don't want to have to care about the primitive that you're using. I don't want you to have to choose and right now I think you have to choose regardless of the tool that you're using you must choose very early and to take advantage of a new tool to go from containers to Lambda or whatever else you want to use you have to rewrite or you have to rebuild or you have to rewrap what you're doing and I want to get to a point where you don't care that I can use whatever combination of the below that I want to use and that AWS will provide tools around that that just says you run this however you want you mix and match whatever flavors you like and we'll take care of it. Yeah, it's interesting. When I cut you know almost every time we've done one of these Kubernetes shows we've had somebody from Amazon on and even if we haven't had an AWS employee almost every customer we have on is doing some if not a lot of Amazon. There are some out there that look and they're like well Amazon doesn't have the biggest booth and Amazon has all of these different choices out there so they must not be fully committed to capital K Kubernetes and things like that. How can you help us understand what's going on? Yeah, so I think Bob Wise and his team spend a ton of time working on the community and the whole team does, right? We're one of the biggest contributors at CD we're hosting Birds of the Feather. We've contributed back to a fair amount of community projects and I think a lot of them are in fact around how to just make Kubernetes work better on AWS and that might be something that we built because EKS or it might be something like Cluster Auto Scaler, right? Which ultimately people would like to work better with auto scaling groups. So I think we have the community involvement but I think it's about having a quiet community involvement that it's about chopping wood and carrying water and being present and committing and showing up and having experts and answering questions and being present in things like CIG groups than it is necessarily having the biggest booth. Yeah, I mean from my perspective at conferences across the board community involvement can never be measured by who spends enough money on the conference to have a booth large enough to play ice hockey in. That doesn't really seem to be as good of a barometer. Things like the roadmap tend to be a spectacular I guess, expression of how that engagement has started to look and I'm really enthusiastic to see what's been done so far and I'm looking forward to seeing more of it. Well thank you, I'm really proud of the roadmap. It's been so interesting to see customers taking kind of a new level of transparency for us product roadmap wise and then I love seeing people go through and start adding more. So I feel like the roadmap started to feel successful to me when customers started opening a ton of issues and saying hey have you thought about this? Or our new thing is we've been posting requests for comments or design docs on there and saying we're thinking about building this and here's what we were thinking about building. Did the way that we built this solve the problem that you're trying to solve? Because ultimately you can build the coolest thing in the world and if it doesn't solve problems for your customers, what's the point? And Abby I'll reiterate that the roadmap was something that the ecosystem, the community was very excited about. What are the things that you want to share before we wrap things at the show or related to the container space that you're hearing your customers talking and asking a lot about? Yeah so I've heard great things about all the sessions. I think that I'm a little biased because I was on the program community so obviously the selection was universally excellent. Yeah I think what I like the most I think about events like this is that everyone seems to have a different way of solving things. They're all asking for something new. They're all talking about a different project. They're all in different SIG groups. They're all making different feature requests. They're all using different tools. I think that that's really powerful. I think that what's made Kubernetes so amazing is that the whole community feels like this. This is a huge turnout for a conference and everyone feels very like actively engaged and I like seeing us kind of push the boundaries right between how much can I pass off to something like EKS and then how much can I keep customizing but on only the things that matter to me. I guess as you're talking about roadmap and plans for the future, if I were to build an environment on AWS going back let's say a decade-ish I would have built something in a single AWS account using EC2 classic and maybe simple DB as a data store which generally is in no way aligned with best practices today and migrating off of those types of architectures for some customers has been painful. Is there any way to I guess loosen the abstraction for lack of a better term of what the things we can do in a build in a forward-looking way today that will make migrating to whatever best practices emerge from customer learnings or the rest in the future not be the equivalent of an entire migration? Yeah, so I think what you're asking right is how can I make kind of adopting new technologies or migrating a little bit easier? Yeah, or even adopting new patterns. Yeah, I think where I see this space kind of going and where I think it gets interesting to me is things like AppMesh. So I can have many different kinds of compute inside of a mesh through AppMesh, right? So I can have an application running on EC2. I can have a container running with EKS or ECS. You can have Kubernetes on EC2. In the fullness of time I'd love to see things like Lambda functions inside an AppMesh. What I like about that is that how that can make the migration process easier because if I can have many types of primitives in the same mesh, I can mix and match or I can drain traffic off from one to the other and I can experiment a little bit more without having to rewrite because I can try it out. It can be part of the same mesh and if I want to move, I can just move more stuff over. I think that's interesting and as for kind of the best practices and stuff like that, we evolve hand in hand with our customers as our customers are figuring out new technologies that they want to use or new ways of building things. We want to be right there with them and I think that the AWS way is about how can we help customers build whatever way they want to do but help them be secure, reliable, and scalable. Yeah, what I'm hearing from that as a takeaway is if I'm not playing around with service meshes or App meshes now, it's probably time to fix that and learn how they work. I think it's a new technology. I think it's an interesting one. I'm excited to see where it goes but watching it kind of grow along with Kubernetes has been really interesting. All right, well, Abby Fuller, thanks so much for joining me again on theCUBE for Corey Quinn, I'm Stu Miniman. You're watching KubeCon, CloudNativeCon 2019 in Barcelona, Spain. Thanks for watching theCUBE.