 Live from Copenhagen, Denmark, it's theCUBE covering KubeCon and CloudNativeCon Europe 2018, brought to you by the CloudNative Computing Foundation and its ecosystem partners. Hello, welcome back everyone. We're live here at theCUBE in Copenhagen, Denmark for KubeCon 2018 Europe, part of the CNCF CloudNative Compute Foundation, part of the Linux Foundation. I'm John Furrier, my co-host, Laura Cooney here this week. And our next two guests are Yvonne Huron and Javi, founder and CTO of Iguazio and Doug Davis, who is the co-chair of the serverless working group in the CNCF as well as a developer advocate for IBM, IBM Cloud. Great to see you, welcome to theCUBE. Thank you. Thanks for coming in. So I love the serverless working group, I want to dig into that with a bunch of questions. So super important trend, obviously we're seeing that success, functions and all of the good stuff's going on, programmable infrastructure. So I want to dig into that. But first, Yvonne, I want to get into what's going on with the business. What's new with you, Iguazio? I saw you're on the sponsorship list here, you're doing a lot of work. You have some news as well. What's going on at KubeCon Europe for you? Yeah, so we're expanding on the big side very nicely, taking more momentum on this trend towards edge analytics, edge cloud. People starting to understand that central cloud is not the only way to build clouds. We're also progressing nicely on our serverless framework called Nucleo. It just was published maybe eight months ago already, about 2,000 stars in GitHub users. We've got some quotes and PRs around the production version of that, including partnership, strong partnership with Azure on being able to run the same functions in Azure and the cloud and the joint development effort as well as customers actually using it to build real-time analytics use case in development in the cloud and deployment in different locations. Our audience knows you well. We've been on the Kube many times. You're also right for us as well as other blogs. With your opinion, pieces, and commentary. So it's always edgy and strong and right on the money. I want to ask you your thoughts on serverless because you were there from day one. I remember the conversations. It wasn't called serverless. We're talking about resource pools and looking at cloud computing, simplifying about potentially what Kubernetes and orchestration was going to look like. It's happening. So are you happy with the progress of the industry performance of the tech stack? What's your thoughts on serverless today, this data, the union? What's your opinion? I think it's progressing nicely. I think many people call everything almost serverless now. You have serverless databases. You have serverless everything. I think serverless will become more and more a feature of a platform, not necessarily a thing, but Salesforce will have serverless functions, Wix will have serverless functions for their own stuff. Obviously cloud platforms and analytics platforms, et cetera. So there'll be maybe a family of generic ones and a family of sort of platform-specific that are more use case-oriented. Does that connect with your business plan for Aguazio? Are you evolving with it? How are you navigating those waters on the adoption side? So, you know, I'm sort of trying to be inclusive. I think there's room for more than one serverless framework. There's also OpenWisks and OpenFaz and a few of those. Our focus is mainly real-time analytics and high performance and data processing. Yes, we can also do other things, but maybe we won't invest too much in some features that are more front-end oriented or something like that. So you're staying focused on the core? Yes, on the other end, other people that deal with front-end will focus on HTTP and glue logic and things like that. Most of the frameworks don't have the same capabilities of nuclear like real-time stream distribution, real-time low latencies, all that stuff. So I think there's room for multiple frameworks and that's also part of the relationship with Azure. Azure have their own product, which is very good in integration with Azure Stack and, you know, Azure components. On the other end, in areas of real-time analytics and IoT, nuclear is stronger. So their interest is rather than saying, no, we'll choose just one horse. Why won't we enable the market and allow the people the choice in solution? That's great. On IBM's side, I want to get your thoughts on the working group as well as IBM. You guys have done a lot of open source IBM, well-known in the Linux history books, as we know. And now I'm very active, again, continuing that mission. Congratulations and thanks for doing that. But the serverless working group, this is a broader scope now. Can you just give some color on the commentary around how that's evolving? Because you guys have a lot of blue chip customers, you know, Cloud Foundry just did a survey, I was talking to Abby Kearns yesterday about the results came back. Mainstream tech, not middle of the country, but they heard about Kubernetes like, what's Kubernetes, you have people going, okay, I got a job to do. But now Kubernetes has arrived. This is a key part of a micro services focus. Right, yeah. And so the way the serverless group kind of got started was about a year ago, the CNCF TOC, Technical Oversight Committee, decided serverless is kind of a new technology. We want to figure out what's going on in that space. And so they started up a working group and our job wasn't to really decide what to do about it yet. It was sort of give us the landscape, what's going on out there, what are people doing, what does serverless even mean relative to function of the service or even the other as is and stuff like that. What does a serverless frameworks generally look like? What do people use it for, use case, stuff like that? And then at the end of that, we produced a white paper with our results as well as a landscape sort of spread you to say all the various technologies out there in that space, who's doing what, without trying to pick winners to say what's there. And then we ended with a set of recommendations in terms of what are possible next steps the CNCF could do in this space with an eye towards interoffability more than anything else, because that's what really we care about. We don't want vendor lock in all the good stuff. And so we had set of recommendations and one of the main ones was two main things. One was function signatures is a very popular one, but we decided to focus on eventing first because we thought that might be an easier fruit to pick off the tree first. And so we're going to focus on the format or metadata of an event as it transfers between systems. And so from the service working group, we created a cloud events sort of little subgroup within our working group to focus on creating a specification around what the metadata around an event would look like just so we can get some commonality. That way, at least the infrastructure between the two systems can transfer the events back and forth much in the same way HTTP layer doesn't have to understand the body of the message but can look at common headers and know how to route it properly. Same kind of things with eventing. And again, this is all about trying to get interoffability and portability for applications or end users more than anybody else. And so that's kind of where our focus has been on how can we help the end user not get locked into one platform, not get locked into one solution and make their life easier overall. Where are you now with that? Is it running? Is it? Oh, we're all done. No. Where have we done that last week? No, actually as of last week though, we just released our first version at 0.1. It's a very, very basic thing that people might look at and say, what's the big deal? But even with that simple little thing, we've been able to get some level of interoffability between the various platforms. And if people actually join, when is it, Friday, 11 o'clock? We have a session where someone's going to demonstrate interoffability between, oh gosh, IBM, you guys, Microsoft, DMware, Google, all the various companies involved in this thing. That's great. Yeah, they're all going to be receiving, or sending, either sending or receiving events using the common event format, I'm sorry, cloud event format, to prove interoffability around the specification. So we're just at 0.1, we have some way to go, but that first step was huge just to get agreement and everybody to the table to agree. So it's been really, it was really fun. And it wasn't easy. It wasn't easy. And he's the peacemaker in the group. We have a lot of, I'm the troublemaker. We have a lot of vocal people in the group. Yes. We're not pointing at anyone. More on that. Important first step, obviously commonality and having some sort of standardization kind of thinking. Yes. How is this? Yeah, you don't use the standard words. There are people allergic to that. Well, yeah, I mean, this is standard bodies and whatnot, but in terms of the community work going on, this is super important. And what's the impact of that? Obviously it's small step, but a big step, right? So what's it going to impact? What's next? What's coming next? Now that you've got the metadata and you've got the interoperability, what's next? Well, obviously we need to finish it up because that's the 0.1 is obviously the first step. As I said, I think beyond that, people are really itching to do function signatures. Because I think if you can get the event format coming in to be somewhat similar, and then you can get portability of moving your function from one platform to another with hopefully minimal changes from a function signature point of view, you're a long way there towards getting portability for people. And I think that's probably the next step we're going to be looking at. And what's the technical case from a commercial entity like yourself, who's in business to make money? I'll see you in business to run. As you build out your architecture, where is this going to be applied for you? What's the impact of this project to your product? So beyond my strong religion around open APIs, and you've seen the blogs I've written about, you know, our interest is twofold. First, we're not the market leader, you know, Amazon is the market leader, et cetera. So if you have a better technology and things are standard, it's easier for customers to move. Second is we believe in this portability across closer to the data, closer to where the processing, especially when 5G is going to evolve and we're going to see bottlenecks between metro locations. So our sense is go develop in the cloud and then push it, you know, the digital twin model. This is exactly what we're demonstrating with Azure. You know, you could develop it Azure, on nuclear functions and deploy in a factory. So having, you know, it may not be the same platform. It may not be the same serverless framework. So having the ability to run the same sort of code in different frameworks or different platforms is very important. And IBM, you're doing a lot of work. OpenWISC has been something that's gotten a lot of press and notoriety. What's up with you guys at OpenSource? Obviously, we see you guys out there doing a lot of stuff. You're doing a lot of content, doing a lot of coding. What's new over on the IBM side of the house with the serverless? From my point of view, I think probably the biggest thing is we're leading the charge importing OpenWISC to run on top of Kubernetes. And I think what's interesting about that is we're going to see probably some changes to Kubernetes needed to be made to get the better performance that we need. Because when OpenWISC runs, vanilla on top of, say, Run-C, or the Docker stuff, we have a lot more freedom there, pausing containers, stuff like that, stuff you can't do in Kubernetes. We're probably going to see some more pressure on Kubernetes to add some more features to get the kind of performance numbers we need going forward. And the scale too is important to understand. I was just talking about the keynotes earlier with another guest and CERN is up there. They have 1,000 nodes, not massive numbers yet. That's scale. Amazon on the big clouds, you guys have clouds, like, got a lot of nodes. So there's a lot more scale going on in the cloud as Kubernetes starts to get its footing. How do you explain Kubernetes, how do both of you guys explain Kubernetes to the IT transformation group out there that's going cloud operations? So what we've seen, because we're also selling sort of an appliance, a fully integrated solution, people don't necessarily, in the enterprise, they don't necessarily want to understand low level of Kubernetes. And actually, serverless is a nice way for doing that. If you look at the new, like, Nucleo dashboard, you just go, you write some code, you click, deploy. It auto-scales, you don't need to think about the underlying, you know, kubectl, underlying networking. It's all done there for you. And I think what you see in the trend in the industry, some people call it serverless, some people call it other things, is more and more abstractions where users will deploy code, will deploy containers, and some frameworks underneath will deal with the high availability, elasticity, all that. That's what enterprise customers are looking for. Another one is eBay and Google, you know, and Netflix. You have thoughts? What I think is interesting, I agree with what you said, but I think it's interesting, is you actually have a wider range of people, right? You have some people who think Kubernetes, as you said, nice abstraction layer, you don't have to get into the need grid if you don't need to. But Kubernetes does allow you to get under the covers and twiddle those lower level bits if you actually need to. And I think that's one of the things that, you know, people start out with Docker. They like it, it's so simple to use, and it's wonderful, and I love it. But they found it a little bit limiting, because it was too opinionated or didn't give you access to things under the covers. Kubernetes, I think, is trying to find that right balance between the two, and I think for the most part, they kind of hit it. There's a little bit more of a learning curve, because it's not quite as user-friendly as Docker is. But once you get over that learning hump, all the flexibility it gives you, people seem to really, really like that. What are some of the things that people do under the covers? You mentioned some tweaks here and there. Is it policy-based stuff? What's happening under the covers that Kubernetes is getting a groove swing on now? There is something called custom resource definition. So for example, when we deploy a nuclear, maybe OpenWizk or others as well, is essentially, nuclear becomes another resource that you can actually view when you're running the Kubernetes CLI or all the other things that manage its liveliness, et cetera. So those are services that you get from free as a platform, but if you want your function to keep being alive, you need to code your functions into the liveliness API. The thing that monitors it is staying alive. So you're getting a generic service, but you need to work with it. Yeah, actually, I'd go one step further with that and abstract it a little. Because obviously, Kubernetes has a lot of knobs you can turn. A lot more than other platforms like Docker has. But I think for me, the biggest benefit of Kubernetes is the plugability. Custom resource definitions, one of them, ripping out schedulers or whatever controllers you want to replace it with your own. That kind of flexibility to say, I don't have to leave the entire Kubernetes world just to write my own scheduler or write the infrastructure around it. I can plug in my own. That's the kind of flexibility people seem to really, really like. That way they don't feel locked in, they can still play a part of the ecosystem, but get the flexibility and customization they need. Awesome, great commentary there. I want to get your thoughts on KubeCon 2018 Europe for CNCF, continuing to see growth in CNCF, fantastic to see. But as you get, as the boat gets full with people, you got to be the peacemaker if you're a co-chair. As people want to start getting their cause into projects, there's some balance on the community side. Are you guys happy with the direction, obviously the success and the visibility has increased? What's your take on the show here? What are you guys doing? Just what's going on around the event for you guys? So it only started today, but my impression comparing it with the previous show in the US, there are a lot more decision makers here. I don't know if it's the European culture of not funding every sort of student to every show, or just the maturity of the ecosystem. But that's something I've noticed that discussions I had with decision makers and they're also not everyone like in the US, I felt everyone wants to build it their own way. People hear things about operationalizing solutions. So sometimes you need to take something that someone else already built and tested. And what's the conversations like that you're having? Is it architecture? Is it deploying production workloads? So for us, it's a lot about use cases because we're doing things in a very different way. We're doing some nice demos on how we're running real-time analytics with a sample sort of database as the core. And we're showing how it's equivalent to another solution that they may build. And that immediately clicks, you know. We need that. The other aspect is really the need to, there's so much technology, but we need someone to wrap it up for us as a packet solution. And that's another. Doug, your thoughts, first of all, I love your shirts as code with all the words in the community. Yeah, they're my favorite shirts, I like it. Love that shirt. I'm just looking at all these questions that pop into my head. What's your plan that's shown here? What's your goal? What are you guys doing? What conversations are you hearing in the hallways? Well, obviously, even from IBM, we just promote IBM as much as we can, obviously. Now, but beyond that, really talk about the interoperability around what we're doing here and make sure people understand that we're not here necessarily to sell our products, which we all obviously want to do. We want to make sure that we do it in a way that give people a choice. And that's why we had to have the service working group, the cloud event spec. It's all about giving everybody the choice to move from one platform to another to get their job done. As much as we want people to buy our stuff, if the customer isn't happy in getting what they need, then we're all going to lose. Yeah, and these projects are super important to get the solidarity around these both standards. Just to follow on your previous question about the conference itself, what we like. Obviously, it's great that it's growing so much. But what I really like about this conference, beyond some other ones that I've seen, is a lot of the other ones tend to have more marketing flair to them. And obviously, there's a little bit of that here if you are promoting their stuff. But I love the fact that most of the stuff that I'm doing here aren't in the sessions, because the sessions are great and interesting. But it's the hallway chatter and interacting with people face to face, and not just to meet them, to actually have real technical deep discussions with them here at the conference, because everybody's here. You can do that much better face to face than you can over a Zoom call or something else. And the productivity from that level is just astronomical. I love it. Yeah, I totally agree. And one thing I would add, just my observation interviews in the hallways is that we're living, and we talk about this on theCUBE all the time, a modern software architectures here. And it's got some visibility around. It's not filled in yet, but I think this clear visibility, cloud, microservices, interoperability, portability, pretty clear. And people are engaged, people excited. So you get the progressive new guard coming in on board here. So great job. Thanks for coming on theCUBE. We appreciate it. Thank you. You're on. Iguazio and IBM here on theCUBE, breaking down KubeCon 2018 Europe. More live coverage. Stay with us. We'll be right back after this short break.