 from Seattle, Washington. It's theCUBE, covering KubeCon and CloudNativeCon North America 2018. Brought to you by Red Hat, the CloudNative Computing Foundation and its ecosystem partners. Okay, welcome back everyone. We're live here in Seattle for theCUBE's exclusive coverage of KubeCon and CloudNativeCon 2018. I'm John Furrier, Stu Miniman. Our next guest is Stefan Fable who's the Director of Product Management at Canonical. Kube alumni, welcome back. Good to see you. Thank you. Good to see you too. Thanks for having me. You guys are always in the middle of all the actions. It's fun to talk to you guys. You have a pulse on the developers. You have a pulse on the ecosystem. You've been deep in it for many, many years. Great value. What's hot here? What's the announcement? What's the hard news? Let's get the hard news out of the way. What's happening here at the show for you guys? We've had a great number of announcements. Great number of threads of work that came and took fruition over the last couple of months. Now just last week where we announced hardware reference architectures with our hardware partners Dell and Supermicro, we announced ARM support, ARM64 support for Kubernetes. We released our version 113 of our charm distribution of Kubernetes last week. And we also released, very proud to release, MicroKates. Kubernetes is a single snap for your workstation in the latest release 113. Yeah, let me explain that. Because we often talk about scale, but there's big scale. And then we were talking about edge, we're talking about so many of these things. That's right. That small scale is super important. It really is, it really is. So MicroKates came out of this idea that we want to enable a developer to just quickly stand up Kubernetes cluster on their workstation. And it really came out of this idea to really enable, for example, AML workloads locally from development on the workstation all the way to on-prem and into the public cloud. So that's kind of where this whole thing started. And it ended up being quite obvious to us that if we do this in a snap, then we actually can also tie this into appliances and devices at the edge. So now we're looking at interesting new use cases for Kubernetes at the edge as an actual API endpoint. So it's quite nice. Steph, I'll talk about, I want to take a step back. There's kind of dynamics going on in this Kubernetes wave, which by the way, it's phenomenal. 8,000 people here at Kube gone up from 4,000. It's got that hockey stick growth, almost like a Moore's law, if you will, for the events. You guys have been around. So you have a lot of existing big players that have been in the space for a while, doing a lot of work around cloud, multi-cloud, you know, whatever, what else, the new word. But again, you guys have been there. You got like the Cisco's of the world. You guys, big players actively involved. A lot of new entrants coming in. What's your perspective of what's happening here? A lot of people are looking at this, scratching their heads saying, okay, I get Kubernetes. I get the magic Kubernetes enables a lot of things. What's the impact to me? What's in it for me as an enterprise or a developer? How do you guys see this marketplace developing? What's really going on here? Well, I think that the draw to this conference and to technology and all the different vendors, et cetera, is ultimately the multi-cloud experience, right? It is about enabling workload portability and enabling the operator to operate Kubernetes independently of where that is being deployed. And that's actually also the core value proposition of our charmed Kubernetes. The idea that a single operational paradigm allows you to experience, deploy lifecycle manage and administer Kubernetes on-prem, as well as any of the public clouds, as well as on other virtual substrates, such as VMware. So ultimately, I think the consolidation of application delivery into a single container format, such as Docker and other compatible formats, OCI formats, right? That was ultimately a really good thing because it enabled that portability. Now I think the question is, I know how to deploy my applications in multiple ways because it's always the same API, right? But how do I actually manage a lot of Kubernetes clusters and a lot of Kubernetes API endpoints all over the place? So break down the hype and reality because again, a lot of stuff looks good on paper, love the sound bites of people saying, hey, Kubernetes, all this. But there's people are admitting there's some things that need to be done, work areas, security is a big concern. I'm working on people are working on that. Where's the reality? Where does the rubber meet the road when it comes down to, okay, I'm an enterprise. What am I buying into with Kubernetes? How do I get there? We heard Lyft take an approach to saying, look, I solve one problem, get a beach head and kind of incremental, take the incremental approach. Where's the hype? Where's the reality? Separate that for us. So I think that there's certainly a lot of hype around the technology aspect of Kubernetes, right? Obviously containerization is invoked. This is how developers choose to engage in application development. We have microservices architectures. All of those things were very well aware of and have been around for quite some time and in the conversation. Now looking at container management, container orchestration at scale, right? It was a natural fit for something like Kubernetes to become quite popular in this space. And so from a technology perspective, I'm not surprised. I think the rubber meets the road as always in two things, in economics and in operations. So if I can roll out more Kubernetes clusters per day or more containers per day than my competitor, I gain a competitive advantage, right? That the cost of per container is ultimately what's going to be the deciding factor here. Yeah, Stefan, when I think about developers, how do I start with something and then how do I scale it out in the economics of that? Boy, I think Canonical has a lot of experience with that to share. What's the same what's different about this ecosystem, cloud native versus when we were just talking about the Linux or previous waves of infrastructure? Well, I think that ultimately, Kubernetes in and of itself is a mechanism to enable developers, right? So it plays one part in the whole software development lifecycle and accelerates a certain part. Now it's on us, distributors of Kubernetes, to ensure that all the other portions of this whole lifecycle and ecosystem around Kubernetes, where do I deploy it, how do I lifecycle manage it? If there's a security breach like last Monday, what happens to my existing stack and how does that go down? That acceleration is not solved by Kubernetes, it's solved for Kubernetes. Yeah, you know, your software lives in lots and lots of environments. Maybe you can help clarify for people trying to understand how Kubernetes fits and when you're playing with the public clouds, you know, your Kubernetes versus their Kubernetes and the distinction I think is there's a lot of nuance there that people need help with. It's true, yeah. So I think that first of all, I mean, we always distanced ourselves from the notion of having our Kubernetes. I think we have a distribution of Kubernetes. I think there's conformance tests that are in place, that they're in place for a reason. I think it is the right approach and we won't install a forked version of Kubernetes anytime soon. Certainly, that is one of the principles we adhere to. What is different about our distribution of Kubernetes is the operational tooling and the ability to really cookie cutter out Kubernetes clusters that feel identical even though they're distributed and spread across multiple different substrates. So I think that is really the fundamental difference of our Kubernetes distribution versus others that are out there in the market. Some of the role of developers now, because obviously you're seeing a lot of different personas of emerging in this world. You're just going to lay them out there and then we'll get your reaction. The classic application developer, the ones who are sitting there writing code inside a company, could be a consumer company like Lyft or an enterprise company that needs their rebuilding inside. So it's clear that CIOs or enterprises, CXOs or whatever the title is, they're bringing more software in-house, bringing that competitive advantage under application development. You have the IT Pro expert practitioner kind of role, classic IT, and then you've got the open source community vibe, this show. So you get these things interplaying with each other. This show to me feels a lot like an open source show which it is, but it also feels a lot like an IT show, which it also is. It also is, and it feels like an app development show, which it also is. So opportunity, challenge, is this a marketplace condition? What's your thoughts on these kind of personas? Well, I think it's really a question of how far are you willing to go in your implementation of DevOps cultural change? If you look at that notion of DevOps and that movement that has really taken a hold in people's minds and hearts over the last couple of years, we're still far off in a lot of ways, in a lot of places. Even in the places who are saying they're doing DevOps, they're still quite early, if at all, on that adoption curve. I think bringing operators, developers, and IT professionals together in a single show is a great way for the community and for the markets to actually engage in a larger DevOps conversation without the constraint of the individual enterprise that those teams find themselves in. If you can just talk about how you should do something better and how would that work? And there's other kinds of personas and roles at the same table. It is much better to have the conversation without the constraint of like a deadline or milestone or some outage somewhere, right? Like something is always going on. Being able to just have that conversation around the technology and really say, hey, this is going to be the vehicle that we use to solve this problem and further that conversation, I think it's extremely powerful. And we always talk about who's winning and who's losing, it's what media companies do. We do it on the queue, we debate it. At the end of the day, we always like, there's no magic quadrant for this kind of market. But the scoreboard can be customers. Now Amazon's got over 5,000 representable customers. I don't know how many CNCF has. It's probably a handful, not 5,000. The customer implications is really where this is going, right? Multi-cloud equals choice. What's your conversations like with customers? What do you see on the customer landscape in terms of appetite, IQ or progress for DevOps? I mean, we were talking, not everyone's on serverless yet. And that's so obvious, that's going to be a big thing. So enterprises are hot right now. They want the tech, seeing the cloud growth. Where is your customer base? What are those conversations like? Where are they in the adoption of cloud native? It's an extremely interesting question actually, because it really depends on whether they started with pass or not. If they ever had a pass strategy, then they're mostly disillusioned. They came out, they thought it was going to solve a huge problem for them and save them a lot of money. And it turns out that developers want more flexibility than any pass approach really was able to offer them, right? So ultimately they're saying, you know what? Let's go back to basics. I just give you a Kubernetes API endpoint. You already know how to deal with everything else beyond that. And actually, you're not cookie-cuttering out PostgreSQL. So Kubernetes is a reset to pass. It really does. It kind of disrupted that whole space, right? And then took a step back. Stefan, how about serverless? So a lot of discussion about Knative here. We've been teasing out where that fits compared to functions from AWS and Azure. What was the canonical take on this? What are you hearing from your customers? So serverless is one of those, well, it's certainly a hot technology and a talking technology of interest to our customers. But we have longstanding partnerships with Galactic Fog and others in place around serverless. I haven't seen real production deployments of that yet. And frankly, it's probably going to take a little bit longer before that materializes. I do think that there's a lot of efforts right now in containerization. Lots of folks are at that point where they are ready to and are already running containerized workloads. I think they're busy now implementing Kubernetes. Once they have done that, I think they'll think a little bit more about serverless. One of the things that interests me about this ecosystem is the rise of Kubernetes, the rise of choice, the rise of a lot of tools, a lot of services, trying to fend off the tsunami wave that's hit the beachhead of Amazon. And I've always said in theCUBE they're going to take as much inland territory on this tsunami and let someone put up a sea wall. And that's, I think this is this community here. And I think the question is, is that I want to get your expert opinion on this because the behemoths, the big guys are getting richer. The innovations coming from them, they have scale. You mentioned that as a key point in the value of Kubernetes, it's a scale. As one of those players, and I would consider in the big size, not like a behemoth like an Amazon, you've got a unique position. How can the industry move forward with disruption and innovation with the big guys dominating? What has to happen? Is there going to change the size of certain TAMs? Is it going to be new service providers emerging? Something's got to give. Either the big guys get richer at the expense of the little guys or market expense of new categories. How do you guys look at that? Actually, developers are out there so it's a promising look to new categories, but your thoughts? I think it's, so a technology perspective certainly would be there could be a disruptive technology that comes in and just eats their lunch, which I don't believe is going to happen, but I think it might actually be more of a market functionality actually. If it goes down to the economics and as they start to compete, there will be a limit to the race to the bottom. So if I go in on an economical advantage point as a public cloud, then I can only take that so far. Now I can still take it a lot further, but there's going to be a limit to that ultimately. So I would say that all of the public clouds, we see that increasingly happening, are starting to differentiate. So they're saying, come to me for AIML, come to me for a rich service catalog, come to me for workload portability or something like that, right? And we'll see more differentiation as time goes on. And I think that'll develop in a little bit of a bubble to the point where actually other players who are not watching, for example, Chinese clouds, right? Very large, very influential, very rich in services. They can come in and disrupt their marketing in a totally different way that a technology ever could. So key point you mentioned earlier, I want to pivot on that, get to the AI conversation, but scale is a competitive advantage. We've seen that, we've said it on theCUBE, we see it in the marketplace. Kubernetes by itself is great, but at scale it gets better. It's got knobs and policy. AI is a great example of where a dormant computer science concept that has not yet been unleashed, it's unleashed by cloud. Now that's proliferating. AI, what else is out there? How do you see this trend around just large scale Kubernetes, AI and machine learning coming on around the corner? That's going to be unique and it's new. So you mentioned the Chinese clouds could be a developer here. So it's a lever. So absolutely, I mean, we've been involved with Kubeflow since the early days. Early days, I mean, it's barely a year, right? So what are early days? It's a year old, so a year ago, we started working with Kubeflow and we published one of the first tutorials of how to actually get that up and running and started on Ubuntu and with our charm distribution of Kubernetes. And it has since been a focal point of our distribution. We do a couple of things with Kubeflow. So the first thing, something that we can bring as a unique value proposition is because we're the operating system for almost all GKE, all AKS, all EKS, such a strong standing as an operating system and our strong partnerships with folks like NVIDIA, it was kind of a, you know, one of the big milestones that we tried to achieve and we've since completed actually as another announcement since last week is the full automatic deployment of GPU enablement on Kubernetes clusters and have that identical experience happen across the public clouds. So GPGPU enablement on Kubernetes as one of the key enablers for projects like Kubeflow which give you machine learning stacks on demand, right? And then in parallel, we've been working with Kubeflow in the community but very active form the steering committee to really get the industry perspective into the needs of, you know, of Kubeflow as a community and worked with everybody else in that community to make sure that Kubeflow, well, releases on time and hopefully soon and at 1.0 which is due this summer but right now we're there, we're focused on the open form. That's a key area of innovation though, opportunity. Oh, absolutely, absolutely. And we see Amazon certainly promoting that. What else is new? We've got one last question for you. What's next for you guys? Get a quick plug in for canonical. What's coming around the corner? What's up? Well, we're definitely happy to continue to work on the GPGPU enablement. I think that is one of the key aspects that needs to stay moving to stay on top of. We're looking at Kubernetes across many different use cases now especially with our IoT open to core operating system which we'll release shortly and here actually having new use cases for AI ML inference, for example, out at the edge looking at drones, robots, self-driving cars, et cetera who were working with a bunch of different industry partners as well. So increased focus on the devices side of the house can be expected in the 2019. And that's key to use the data in a way that's really relevant. Absolutely. All right, Stefan, thanks for coming on theCUBE. Appreciate it. Tenauticals, great insight here. Bringing more commentary to the conversation here at KubeCon, CloudNativeCon. Large-scale deployments as a competitive advantage. Kubernetes really does well there. Data, machine learning, AI, all part of the value partnership is in above and below Kubernetes. We're seeing a lot of great advances. Kube coverage here in Seattle. We'll be back with more after this short break.