 Hi everybody, this is Dave Vellante. Welcome to this CUBE conversation where we're going to go back in time a little bit and explore the early days of Kubernetes, talk about how it formed, the improbable events, perhaps that led to it, and maybe how customers are taking advantage of containers and container orchestration today, and maybe where the industry is going. Matt Provo is here, he's the founder and CEO of Stormforge and Chandler Hoisington, is the general manager of EKS Edge and Hybrid at AWS. Guys, thanks for coming on. Good to see you. So Chandler, you were the Vice President of Engineering at Mesosphere, is that correct? Yeah, well, Vice President of Engineering at Mesosphere and then I ran product and engineering for D2IQ at Mesosphere. Yeah, okay, okay, so you were there in the early days of container orchestration and Matt, you were working at a Docker Swarm Shop, right? Yeah. Okay, so, I mean, a lot of people were using your platform, it was pretty novel at the time, it was more sophisticated than what was happening with Kubernetes, take us back, what was it like then? Did you guys, I mean, everybody was coming out, remember there was, I think there was one DockerCon and everybody was coming, I think Kubernetes was announced, you guys were there, Docker Swarm was announced and there were probably three or four other startups doing kind of container orchestration. What were those days like? Yeah, I wasn't actually at Mesosphere for those days, but I know them well, I know the story as well. I came right as we started to pivot towards Kubernetes there, but it's a really interesting story, I mean, obviously they did a documentary on it and people can watch that, it's pretty good, but I think from my perspective, it was really interesting how this happened, you had basically, you had this advent of containers coming out, right? So this new novel technology and Solomon and these folks started saying, hey, wait a second, what if I put a UX around these couple of Linux features that got launched a couple of years ago, what does that look like? Oh, this is pretty cool. So you have containers starting to crop up and at the same time, you had folks like ThoughtWorks and other kind of thought leaders in the space starting to talk about microservices and saying, hey, monoliths are bad and you should break up these monoliths into smaller pieces and any Greenfield application should be broken up into individual scalable units that a team can own by themselves and they can scale independent of each other and you can write tests against them independently of other components and you should break up these big, big monoliths and now we're kind of going back to monoliths but that's for another day. So you had microservices coming out and then you also had containers coming out at the same time. So they're like, oh, we need to put these microservices in something perfect, we'll put them in containers. And so at that point, you don't really, before that moment, you didn't really need container orchestration. You could just run a workload in a container and be done with it, right? You don't need Kubernetes to run Docker. But all of a sudden, you had tons and tons of containers and you had to manage these in some way. And so that's where container orchestration came from and Ben Hindman, the founder of Mesos was actually helping schedule Spark at the time at Berkeley. And that was one of the first workloads was Spark for Mesos. And then his friends at Twitter said, hey, can you help us do this with containers at Twitter? He said, okay. So he wouldn't help them do it with containers at Twitter and that's kind of how that branch of the container wars was started. And it was really, really great technology and it actually is still in production in a lot of shops today. More and more people are moving towards Kubernetes and Mesosphere saw that trend. And at the end of the day, Mesosphere was less concerned about, even though they named the company Mesosphere, they were less concerned about helping customers with Meso specifically. They really wanted to help customers with these distributed problems. And so it didn't make sense to just do Meso so they would took on Kubernetes as well and I helped them do that. I remember my co-founder, John Furrier, introduced me to Jerry Chen way back when Jerry was his first VC investment with Greylock was Docker. And we were talking and he was very obviously very excited about it. And as Chen was just saying, Solomon and the team simplified containers, simple and brilliant. So you guys saw the opportunity where you were a Docker Swarm Shop. Why? Because you needed more sophisticated capabilities. But then you switched, why the switch? What was happening? What was the mindset back then? We ran into some scale challenges in kind of operationalized or productizing our kind of our core machine learning. And we saw kind of the challenges, luckily a bit ahead of our time. And we happened to have someone on the team that was also kind of moonlighting as one of the original core contributors to Kubernetes. And so as this sort of shift was taking place, we saw the flexibility of what was becoming Kubernetes. And I'll never forget, I left on a Friday, came back on a Monday and we had lifted and shifted to Kubernetes. The challenge was, at that time, you didn't have what you have today through EKS and those kinds of services where just getting that first cluster up and running was super, super difficult even in a small environment. And so I remember we finally got it up and running and it was like, nobody touch it, don't do anything. But obviously that doesn't scale either. And so that's really being kind of a data science focused shop at StormForge from the very beginning. That's where our core IP is. Our team looked at that problem and then we looked at, okay, there are a bunch of parameters and ways that I can tune this application and why are the configurations set the way that they are and is there room to explore? And that's really where StormForge came in. Because Mesa said much greater enterprise capabilities as did DockerSwarm, at least they were heading in that direction. But you still saw that Kubernetes was attractive because even though it didn't have all the security features and enterprise features because it was just so simple. I remember Chen Goldberg, who was at Google at the time saying, no, we were focused on keeping it simple and we're going for mass adoption. But is that kind of what you saw? Yeah, and we made a bet, honestly. We saw that the growing community was really starting to, we had a little bit of an inside view because we had someone that was very much in the original part. But you also saw the tool chain itself start to come into place a little bit. And it's still hardening now. But yeah, as any startup does, we made a pivot and we made a bet and this one paid off. Well, it's interesting because we said at the time, I mean, you had obviously Amazon invented the modern cloud, Microsoft has the advantage if it's got this huge software stage, hey, just now run it into the cloud. So they had their entry point. Google didn't have an entry point. This was kind of a Hail Mary against Amazon. And I wrote a piece, the improbable rise of Kubernetes to become the OS, the cloud. But I asked, did it make sense for Google to do that? I'd never made any money off of it, but I would argue they'd be irrelevant if they hadn't done that. But it certainly didn't hurt Amazon. You get EKS and you do containers and your customers, you embraced it, right? I mean, I don't know what it was like early days. I remember I've talked to Amazon people about this, it's like, okay, we saw it, but then talk to customers. What are they doing? That's kind of what the mindset is, right? Yeah, I've been at Amazon a couple of years now and you hear the stories, oh, we're customer obsessed. We listen to our customers like, okay, okay, we have our company values too. You get told them and when you get first hired and first day, you never really think about them again. But Amazon, that really is preached every day. It really is. And we really do listen to our customers. So when customers started asking for Kubernetes, we said, okay, we built it for them. So I mean, it's really that simple. And we also, it's not as simple as just building them a Kubernetes service. Amazon has had a big commitment now to start getting involved more in the community and working with folks like Storm Forge and really listening to customers on what they want and they want us working with folks like Storm Forge and that's why we're doing things like this. Well, it's interesting, because of course everybody looks at the ecosystem and says, oh, Amazon's going to kill the ecosystem. And then we saw an article the other day in, I think it was CRN did an article, great job by Amazon PR, but talking about Snowflake and Amazon's relationship. And I have said many times, Snowflake probably drives more EC2 than any other ISV out there. And so yeah, maybe the Redshift guys might not love Snowflake, but Amazon in general, you know, they're doing great three things. And I remember Andy Jassy said to me one time, look, we love the ecosystem, we need the ecosystem. They have to innovate too. If they don't, you know, keep pace, you know, they're going to be in trouble. So that's actually a healthy kind of a dynamic. I mean, as an ecosystem partner, how do you feel about that? Well, I'll go back to one thing. Without the work that Google did to open source Kubernetes, Storm Forge wouldn't exist. But without the effort that AWS and EKS in particular provides and opens up for developers to innovate to continue kind of operationalizing this shift to Kubernetes, you know, we wouldn't have nearly the opportunity that we do to actually listen to them as well. Listen to the users and be able to say, what do you want, right? Our entire reason for existence comes from asking users like, how painful is this process? Like how much confidence do you have in the, you know, out-of-the-box defaults that ship with your database or whatever it is? And how much do you love manually tuning your application? And obviously nobody said, I love that. And so I think as that ecosystem comes together and continues expanding, it opens up a huge opportunity, not only for existing, you know, EKS and AWS users to continue innovating, but for companies like Storm Forge to be able to provide that opportunity for them as well. And that's pretty powerful. So I think without a lot of the moves they've made, you know, the door wouldn't be nearly as open for companies like Storm Forge who are, you know, growing quickly, but are smaller to be able to, you know, to exist. Well, and I was saying earlier that you're, and I wrote about this, you're going to get better capabilities. You're clearly seeing that cluster management. We've talked about better automation, security, the whole shift-left movement. So obviously there's a lot of momentum right now for Kubernetes. When you think about bare metal, you had servers and storage, and then you had VM virtualization, VMWare really, and then containers, and then Kubernetes as another abstraction. I would expect we're not at the end of the road here. What's next? Is there another abstraction layer that you would think is coming? Yeah, I mean, for a while it looked like, and I remember even with our, like board members and some of our investors said, well, you know, well, what about serverless? And you know, what's the next Kubernetes? And nothing, as much as I love Kubernetes, which I do and we do, nothing about what we particularly do. We are purpose-built for Kubernetes, but from a core kind of machine learning and problem-solving standpoint, we could apply this elsewhere if we went that direction. And so time will tell what will be next and there will be something that will end up expanding beyond Kubernetes at some point. But I think without knowing what that is, our job is to serve our customers and serve our users in the way that they are asking for that to take place. Well, serverless obviously is exploding when you look again at the ETR survey data. When you look at the services within Amazon and other cloud providers, the functions off the charts. So that's kind of an interesting and notable. Now, of course, you've got Chandler, you've got Edge in your title, you've got Hybrid in your title. So this notion of the cloud expanding, it's not just the set of remote services just only in the public cloud now, it's coming on premises. Actually, Andy Jassy in my headspace, he said one time we just look at data centers as another Edge location. Okay, that's the way to look at it and then you've got Edge. So that cloud is expanding, isn't it? The definition of cloud is evolving. Yeah, that's right. I mean, customers want to run workloads in lots of places and that's why we have things like local zones and wavelengths and outposts and EKS anywhere, EKS distro and obviously probably lots more things to come and there's, I always think of like Amazon's Kubernetes strategy on a manageability scale where on one far end of the spectrum you have EKS distro which is just a collection of the core Kubernetes packages and you could take those and stand them up yourself in a broom closet in a retail shop. And then on the other far end of the spectrum you have EKS Fargate where you can just give us your container and we'll handle everything for you. And then we kind of tried to solve everything in between for your data center and for the cloud and so you can really ask Amazon, I want you to manage my control plane, I want you to manage this much of my worker nodes, et cetera and oh, I actually want help on-prem and so we're just trying to listen to customers and solve their problems where they're asking us to solve them. Go ahead. No, I would just add that in a more vertically focused kind of orientation for us, like we believe that optimization capabilities should transcend the location itself and so whether that's part public, part private cloud, that's part of what I love about EKS anywhere, you should still be able to achieve optimal results that connect to your business objectives wherever those workloads are living. Well, don't wince. So John and I coined this term called super cloud and people laugh about it, but it's different. People talk about multi-cloud but that was just really kind of vendor diversity, right? I got to running here, I'm running there, I'm running anywhere, but individually and so super cloud is this concept of this abstraction layer that floats wherever you are, whether it's on-prem, across clouds and you're taking advantage of those native primitives and hiding that underlying complexity and that's what, when we reinvent, the ecosystem was so excited and they didn't call it super cloud, we called it that but they're clearly thinking differently about the value that they can add on top of Goldman Sachs, right? That to me is an example of a super cloud that they're taking their on-prem data and their software tooling, they're connecting it to AWS, they're running it on AWS, but they're abstracting that complexity and I think you're gonna see a lot more of that. Kubernetes itself in many cases is being abstracted away. So there's a bit of a disappearing act for Kubernetes and I don't mean that from an adoption standpoint but Kubernetes itself is increasingly being abstracted away which I think is actually super interesting. Kubernetes doesn't really do anything for a company, like we run Kubernetes, like how does that help your bottom line? And then in the day, companies don't care that they're running Kubernetes, they're trying to solve a problem which is I need to be able to deploy my applications, I need to be able to scale them easily, I need to be able to update them easily and those are the things we're trying to solve. So if you can give them some other way to do that, I'm sure that's what they want. It's not like a big bank is making more money because they're running Kubernetes, that's not the goal. And it gets subsumed, it's just gonna become invisible. Exactly. You guys back to the office yet? What's the situation with work and hybrid work? You know, I work for my house and we go into the office a couple times a week so it's a crazy time, it's a crazy time to be managing and hiring and it's definitely a challenge but there's a lot of benefits for working at home. I got two young kids so I get to see them grow up a little bit more working out of my house so it's nice. Awesome, you back in the office yet? So we're in, even as a smaller startup, we're in 26, 27 states, Canada, Germany, we got a little bit of presence in Japan so we're very much distributed. We have not gone back and I'm not sure we will. Okay, so you think it's permanently remote potentially? Yeah, I mean, we made a pretty, like for us the timing of our series B funding, which was where we started hiring a lot was just before COVID started really picking up. So we thankfully made a pretty good strategic decision to say we're gonna go where the talent is and yeah, it was harder to find for sure especially and we're competing, it's incredibly competitive but yeah, it was a good decision for us. We are very intentional about getting the teams together in person as often as possible and in as safe as way possible obviously but it's been a pretty interesting journey for us and something that I'm not sure I would change to be honest with you. Well, Frank Slutman moved Snowflake's HQ to Montana, does that tell you? And folks like Michael Delce and hey, same thing as you, wherever they wanna work, bring yourself and wherever you are, it's cool. And Chandler, do you think that the hybrid mode for your team is kind of the operating mode for the foreseeable future? You know, I think there's a lot of benefits in both, working from the office, I don't think you can deny like the face-to-face interactions, it feels good just doing this interview face-to-face, right? And I can see your mouth move. So it's like, there's a lot of benefits to that. Over a chime call or a zoom call or whatever, that also has advantages, right? I mean, you can be more focused at home. And I think some version of hybrid is probably in the industry's future. I don't know what Amazon's exact plans are, that's above my pay grade, but I know that like in general, the industry is definitely moving to some kind of hybrid model. No question. And like Matt said, I'm a big fan, at Mesosphere we ran a very remote workforce. We had a big office in Germany, but we'd get everybody together a couple of times a year for engineering week or something like this. And you get 100 people just dedicated to spending time together at a hotel in Vegas or Hamburg or wherever. And it's a really good time. And I think that's a good model. I like that model. Yeah, and I think just more ETR data, the current thinking now is that hybrid is the number one sort of model. 36%, the CIOs believe 36% of the workforce are gonna be hybrid permanently, is kind of their call. Couple days in, couple days out. And the percentage that is remote is significantly higher, probably high 20s, whereas historically it's probably 15%. So permanent changes. That changes the infrastructure you need to support it, the security models and everything and how you communicate. Yeah, I remember when COVID really started hitting in 2020, the big banks for example had to, I mean, you wanna talk about innovation and ability to shift quickly to the bigger banks that have, in fact adopted Kubernetes, were able to shift pretty quickly. Systems and things that were historically, it was in the office all the time. And some of that's obviously shifted back to a certain degree, but that ability, it was pretty remarkable actually to see that take place for some of the larger banks and others that are operating in super regulated environments. I mean, we saw that in government agencies and stuff as well. Well, without the cloud, this never would have happened. Yeah, and I think it's funny, I remember some of the more old school managers were thinking people aren't gonna work less when they're working from home, they're gonna be distracted. I think you're seeing the opposite where people are working too much, they get burned out because you're just front of your computer all day. And so I think that we're learning. I think the whole industry is learning. Like what does it mean to work from home really? And it's a fascinating thing. It's a case study we're all a part of right now. I was talking to my wife last night about this and she's very thoughtful. And when she was in the workforce, she was at a PR firm and a guy came in, a guest speaker, it might even be in the CEO of the company asking what on average, what time, who stays at the office until, who leaves by five o'clock, a few hands went up, or who stays until like eight o'clock and the hands went up. And then so he asked those people, like why can't you get your work done in an eight hour work day? I go home, why don't you go? And I said to her, that's interesting because she's always looking at me like, can't you do, get it done? And I'm saying the world has changed. It really has where people are just on all the time. I'm not sure it's sustainable, quite frankly. I mean, I think that we have to, as organizations, think about, and I see companies doing it, you guys probably do as well. Take a four day weekend just for your head. But it's, there's no playbook. Like I said, we're a part of a case study. It's all so hard because people are distributed now. So you have meetings on the East Coast, you wake up at seven four, and then you have meetings on the West Coast, you stay until seven o'clock therefore. So your day just stretches out. So you've got to manage this. I think we'll figure it out. I mean, we're good at figuring this out. There's a rise in asynchronous communication. So with things like Slack and other tools, as helpful as they are in many cases, it isn't always on mentality. And people look for that little green dot, and if you're online. So my kids, we have a term now for me, because my office at home is upstairs, and I'll come down. And if it's during the day, they'll say, oh, dad, you're going for a walk and talk. Yeah. Which is like, it's my way of getting away from the desk, getting away from Zoom, and even in Boston, getting outside, trying to at least get a little exercise or walk and get my head away from the computer screen. But even then, it's often like, I'll get a Slack notification on my phone or someone will call me, even if it's not a scheduled walk and talk. And so it is an interesting shift. A lot of ways to get in touch. Productivity's presumably going to go through the roof, but all right, guys, I'll let you go. Thanks so much for coming to theCUBE. I appreciate it. Thank you for watching this CUBE Conversation. This is Dave Vellante, and we'll see you next time.