 Hi everyone, welcome back to theCUBE's coverage of DockerCon 2021. I'm John Furrier, host of theCUBE. Got a great segment here, one of the big supporters in open source, Amazon Web Services, returning back second year on DockerCon virtual, Deepak Singh, Vice President of the Compute Services at AWS. Deepak, great to see you. Thanks for coming back on remotely again. Soon we'll be in real life. Reinvent is going to be in person. We'll be there. Good to see you. Yeah, good to see you too, John. It's always good to do these sites. I don't know how often I've been at theCUBE now, but it's great every single time. You're a legend and getting on there. A lot of important things to discuss. You're in one of the most important areas in the technology industry right now. And that is at the confluence of cloud scale and modern development applications as they shift towards, as Andy Jassy says, the new guard, right? It's been happening. You guys have been a big proponent of open source and enabling open source as a service, creating business models for companies. But more importantly, you guys are powering, making it easier for folks to use software. And Docker's been a big relationship for you. Could you take a minute to first talk about the Docker AWS relationship and your involvement and what you're doing? Yeah, and actually it goes back a long way. Just in, we announced ECS at Reinvent in 2014. And ECS at that time was very much managed orchestration service on top of Docker. That time I think were the first really big one. Out there from a cloud provider and since then of course the world has evolved quite a bit and our relationship with Docker has evolved a lot. The last thing I'd like to talk to is something that we announced with Docker last year. I don't remember if I talked about it on theCUBE at that time. But last year we started working with Docker on how can we go from Docker run, which customers love or Docker desktop, which customers love and make it easy for people to run containers on ECS and Fergie. So most new customers running containers on AWS today start with ECS and Fergie, or half of them. And we wanted to make it very easy for them to start with where they are on their laptops, which is often Docker desktop and have running services in the AWS. So we started working with Docker and that collaboration has been very successful. We want to keep, we look forward to continuing to work on evolving that where you can use Docker compose, Docker desktop, Docker run, the tool that Docker customers use and then we will run production services on AWS side, which is a part that we hang up on. So I think that's one area where we work really well together. The other area where I think the two companies continue to work well together as open source in general. As some of you know, AWS has a very strong commitment to container D. EKS, our Kubernetes service is moving towards container D. Fergie actually runs all on container D today. And we collaborate Docker on that on the OCI specification because the OCI image spec is becoming the de facto packaging format at AWS. This morning we launched, well yesterday we launched a service called AppRunner and the main expected input for AppRunner is an OCI image. We did this with Lambda as well where the OCI image is now a way of packaging for Lambda. And I think the last one I'd like to call out and which has been an amazing partnership and it's an area where most people don't pay attention is image signing. There's a project called Notary V2, the second version of the Notary spec for image signing and AWS, Docker and a couple of other companies have been working very closely together on bringing that spec, finalizing Notary V2 so that at least in our case we can start building services for our customers on top of that. You know, it's a great relationship and I expect to see it continue. Well, I think one of the themes this year is developer experience. So good call out there and the new announcements on the tools you have and software, because that seems to be a great developer integration with Docker. The question I have for you is how should the customers think about things like ECS versus EKS, AppRunner, Lambda for kind of running their containers? How do they understand the differences? What's the thought process there? What's the thinking? It's a great question actually, when we announced AppRunner I think that was one of the questions I started getting on Twitter. You know, let's start at the very beginning. Anyone can pick up a Docker container and run it on EC2 today. Like you can run it on EC2 or you can run it on LightSale but Docker on works just fine. It's a Linux machine. Then people want to do more complex things. They want to run large scale orchestrated services. They want to run their entire business on containers. We have customers who do that today. You know, you have people like Vanguard who runs a significant portion of the infrastructure on ECS and Fargi and all your fidelity which is a heavy user of EKS or Kubernetes service. So in general, if you're running large scale systems you're building your platforms you're most likely to use ECS and EKS. If you come from a Kubernetes background you're running Kubernetes on-prem or you want the flexibility and control that Kubernetes gives you you're going to end up with EKS. That's what we see our customers doing. If you just want to run containers you want to use AWS to its fullest extent where you want the container API to be part of the AWS API set. Then you pick ECS and I think one of the reasons you see so many customers start with ECS and Fargate is with Fargate you get the significant ease of use from an operational standpoint and we see many startups and enterprises especially security-focused enterprises leaning towards Fargate. But there's a class of customers that doesn't want to think about orchestration that just wants, here's my code, here's my container image just run my service for me. And that's where things like app runner comes come and that's one of the reasons we launched it. Lambda is a little bit different. Lambda is a unique service you have to buy into an event-driven architecture if you do that and you can fit our application into this that's where it should start, it's magic. The container part there is what Lambda announced that we invented where they now support container packaging. So instead of zip files you can package up your functions as containers and Lambda will run them for you. The advantage it gives you is all the tooling that you have to build your containers now works for Lambda as well. So I won't call Lambda container orchestration service in the same sense that ECS, ECS or app runner are but it definitely allows the container image format as a standard packaging format. I think that's the sort of universal common theme that you'll find across AWS at this point of time. You know, one of the things that we are observing at this event here is a lot of developers KubeCon and Linux foundations, a lot of operators too Kubernetes hits that, but here's developers. And the thing is I want an ease of use simplicity experience, but also I want the innovation. I want all of it. When I ask you what does Amazon bring to the table for the new equation? What would you say? Yeah, I mean, for me it's always, you've probably heard me say this a hundred times maybe a thousand times, it's Fargate. Fargate's unique to us. It takes a lot of what we have learned about operating infrastructure at scale. The question we asked ourselves, you know in many ways we thought about Fargate even before we launched ECS, but we had to learn on what it meant and what customers really wanted. But the idea was when you're running clusters of instances or machines to run containers on you have to start thinking about a lot of things that in some ways VMs in the cloud have taken away capacity, what kind of infrastructure to run it on? Should I have been packed? Should I not been packed? Where is my container running? Those are things you suddenly started having to think about. Those felt kind of backwards almost. So the idea was how can we make your containerized bundles? So ECS task or a Kubernetes pod, the thing that you talk to and that is the main unit that you operate on. That is the unit that you get built on and meet it on. That's why Fargate comes in. And it allows us to do many interesting things. We've effectively changed the engine of Fargate since we've launched it. We run it on EC2 instances and we run it on Firecracker. We've changed the Fargate agent architecture. We've made a lot of underneath the hood changes that even take that take advantage of the broader innovations in AWS. We did a whole bunch more to launch Aprona which runs on top of Fargate. And customers don't have to think about it. They don't have to worry about it. It happens underneath the hood. It's always you upgrade the engine as you go along. And it takes away all the operational pain of managing clusters, of running into picking which instances to use, figuring out, trying to figure out how to bend back and get efficiency. That becomes our problem. So that is an area where you should expect to see us do a ton more. It's becoming the fabric of so many things that AWS now. And in some ways we're just starting. There's a lot more to do. Yeah, and it's a really good time. A lot more wave of developers coming in. One of the things that we've been reporting on on SiliconANGLE and theCUBE with our CUBE videos is more developers keep on coming on, more people are coming in and contributing to the open source community, even end users, not just the normal awesome hyperscalers. You're talking about like classic, I call main street enterprises. So two things I want to ask you on the customer side because you have kind of two customers. You have the community, open source community and you have enterprise customers that want to make it easier. What are you seeing and hearing from customers? I know you guys work backwards from the customer. So I got to ask you, work backwards from the community and work backwards from the enterprise customer. What's going on in their environment? What's the key trends that they're riding? What's the big challenges? What's the big opportunities that they're facing and saying for the community? Yeah, I'll start with enterprise. That's almost an easier answer, which is we are seeing increasingly enterprises moving into the cloud wholesale. Like in some ways you could argue that the pandemic has just accelerated it, but we have started seeing that before and they want to move to the cloud and adopt modern best practices. If you see my talk at Reem when last few years I've talked about modernization and all the aspects of modernization. And that's 90% of our conversations with enterprises. I've walked into a meeting supposedly to talk about containers, but about half a conversation is spent on how does an organization modernize? What does an organization need to do to modernize? And containers and serverless play a pretty important part in it because it gives them an opportunity to step away from the shackles of sort of fixed infrastructure and the methods and approaches that built in. But equally we are talking about CI CD in a fully automated deployments. What does it mean for developers to run their own services? What are the, how do you monitor and instrument your services? How do you do observability in a modern world? So a lot of those are the challenges that enterprises are going towards and we are spending a ton of time helping them there. But many of them are still running infrastructure on premises. So, we have our posts for them. This last week you're talking to a bunch of our customers and they have lots of interesting ideas and things that they want to do with our posts. But many of them also have their own infrastructure and that's where something like ECS anywhere came from which is, hey, you like using ECS in the cloud. You like having this API that just orchestrates containers for you it does it in an AWS region. It'll do it in an outpost, it'll do it on wavelength, it'll do it on a local zone. How about we allow you to do it on whatever infrastructure you bring to us? You want to bring a Raspberry Pi, you can do that. You want to bring your on premises, data center infrastructure, we can do that. Or a point of sale device. As long as you can get the ECS agent running and you can connect to an AWS region, even though it's okay to lose connectivity every now and then. We can orchestrate your containers for you over there. And the same customer that likes to use the ECS and the simplicity really resonated but that message really resonates with them. So I think where we are today with the enterprise is we've got some really good solutions for you in AWS and we are now allowing you to take those APIs and then launch containers wherever you want to run them whether it's the edge or whether it's your own data center. Now I think that's a big part of where the enterprise is going, but by and large, I think yes, a lot of them are still making that change from running infrastructure and applications the way they used to to a modern sort of if you want to use the word cloud native way and we helping them a lot with that. The community is interesting. They want to be more participatory. That's where things like co-pilot comes from. But honestly, the best thing we've ever done in my org is probably our open road maps where the community can go into the roadmap and engage with us over there. Whether it's an open source project or just trying to tell us what the feature is and how they would like to see it. It's a great engagement and it's taught us a lot. It's helped us prioritize correctly and think about what we want to do next. So I think that's a great part of being out there. That must be very hard to do for opening up the Kimono on the roadmap because normally that's the crown jewels and it's secretive and now it's all out in the open. I think that is a really interesting experiment and what's your reaction to that? What's been the feedback on the roadmap piece because I mean, I definitely want to see the roadmap. So we do it pretty much for every service in our organization and we've been doing it now for three or so years. I forget, I think about three years. And it's been great. Now, we are very upfront, which is security and availability, our job, zero, zero, zero and 100 times out of 100 if I were to choose between a new feature and helping our customers be available and safe, we'll do that. And which is why we don't put dates in there. We just tell you directionally where we are and what we are prioritizing. There every now and then we'll put something in there that we will not choose not to put a feature in there because we want to keep it secret until it launches. But for the most part, 99% of our roadmap's out there and people engage with it. And it's not proven to be a problem because you've also been very responsible with how we manage and be very transparent on whether we can commit to something or not. And I think that's helped a lot. I got to ask you as a leader, a threaded leader on this group, open source is super important. As you know, and you continue to do it for many years. How are you investing in the future? What's your plan, plans for your team, the industry? Obviously very inclusive, which is very cool. It's going to resonate well. What's the plans? Give us some details on what you're investing in. What's your priorities? What's your first principles? Yeah, so it goes in many ways. One, I also have the luxury also on the Amazon open source program office. So I get the chance to, my team rather, not me, help Amazon engineers participate in open source. That's a team that helps create the tools for them, makes it easy for them to contribute, creates some, you know, manages all the licenses, et cetera. I'll give you a simple example. You know, there's this thing called ECR credential helper that was written by one of our engineers. And he kind of just wrote it because he felt it was something that we needed to do and we made it open source. In general, in many of our teams, the first question we asked is, should something be open? And why is this thing not open source? Especially if it's a utility or some piece of software that runs along our services. So that was step one. But we've done some big things also. I, you know, a couple of years ago, we launched a Linux operating system called bottle rocket. And right from the beginning, it was very clear to us that bottle rocket was two things. It was both an AWS product, but first it was an open source project. We'd already learned a little bit from what we've done with Firecracker, but making bottle rocket an open source operating system was very important. Anyone can take bottle rocket, we open sourced our build tooling, you can run it wherever you want. If you want to take bottle rocket and build a version and manage it for another provider, or another provider wants to do it, go for it. There's nothing stopping you from doing that. So you'll see us do a lot there. Obviously there's multiple areas you've seen AWS invested in on the open source side. But to me, the wins come from when engineers can participate in small things, release little helpers, or get contributions from outside. And I think that's where we're still, we can always get better. We're going to continue to strive to make it better and easier. And as I said, I have me and my team, we have an opportunity to help there with inside the company and we continue to do so. But that's what gets me excited. Yeah, that's great stuff. And congratulations on investing. The community really enjoys it. And I know it moves the needle for the industry. Deepak, I got to ask you while I got you here, DockerCon, obviously developers. What's the most important story that they should be paying attention to as a developer? Because a lot's going on, shift left for security, day two operations, also known as AI ops, GitOps, whatever you want to call it. Ongoing, you get serverless, you got Lambda. I mean, all kinds of great things are going on. You mentioned Fargate. What should they be paying attention to that's going to really help their life, both innovation-wise and just the quality of life. Yeah, I would say, look at, in the end, it is very easy, developers in particular want to build their builders. And it's very easy to get tempted to try and get, learn everything about something, have access to all the bells and whistles and knobs. But in reality, if you want to run things, you want to focus on what's important, the business application, the end-user application. And I think a lot of what I tell developers, and I think it's a lot of where the industry is going is, we have built a really solid foundation, whether it's Kubernetes or ECS and Fargate or container industries out there, we have a very solid foundation that our customers and developers all of the work can use to build upon. But increasingly, we are going to provide tools that sort of take that, wrap them up and provide them a nice package solution after another great example. Our collaboration with Docker around Docker desktop for a great example where, forget all the muck, focus on the application and build on top of that and you can get so much done. I think that's one trend you'll see more and more. Those things are no longer toys, they're production grade systems that you can build real world applications on, even though they're so easy to use. The second thing I would add to that is GitOps. It is, you can give it whatever name you want and there's nuances there, but I actually think GitOps is the way people should be running the infrastructure. This is my bias and my personal, it's something that we believe a lot in our side, is how do you go towards immutable infrastructure and infrastructure automation. We think GitOps plays a significant role. I think developers naturally gravitate towards it. And if you want to live in a world where development and operations are tightly linked, I think GitOps is a huge role to play in that. And it's actually a big part of how we're planning to do things like ECS anywhere, for example, and GitOps is a significant player in that. Take AWS Proton, I think GitOps will be a significant trend in the future of Proton as well. So, I think that's the other trend. If you wanted to pick a trend that people should pay attention to, that's what I believe in a lot. Yeah, well, you're an expert. So I want to get you a quick definition. What is GitOps? How would you define it? Because that's a big trend. What is it? What does that mean? Alexis Richardson will probably shoot me for getting this wrong. Actually, how I think about it, which is, in many cases, when you're doing deployments, you're pushing a deployment. GitOps is more of a pull deployment when you are pushing code to a Git repository. You have a system that knows that the event has happened and then pulls from there and triggers the thing, as opposed to you telling it, hey, I have this new piece of code, now go deploy it everywhere. So to me, a biggest change is that two parts. One is it's more of a pull mechanism where you're pulling because something has changed. So it needs systems like container orchestrators to keep them in sync. And the second part is a natural evolution of infrastructure as code, which is basically everything is code. Configure as code, infrastructure as code, code is code, and everything is getting stored in a software repo and the software repo becomes your store of record and drives everything. So for a class of customers, that's going to be a pretty big deal. Yeah, when you're checking in code, again, it's like a compiler for the compiler, a container for the container, you got things for each other. Automation is ultimately what we're talking about here and that's to me, we're machine learning kicks in. So again, having this open source foundational fabric, as you said, forget about the muck or the undifferentiated heavy lifting. This is what we're talking about, automation, isn't it, Deepak? Yes, I mean, as I said, the thing where we hang our hat on is there's such good stuff out there in the world which we like to contribute to. But the thing we like to hang our hat on is, why don't you run this? How do you do it this in ways where you can uniquely bring capabilities to customers where there's things like Nitro or things that Nitro opens up? Or the fact that we have built up this operational infrastructure over the last decade plus or in the container space over the last seven years where we really, really know how to run these things at scale and have made all the investments to make it easy to do so. That's where we hang our hat. Keeping people safe, helping them run highly available applications. They're a new startup that just completely takes off in over a weekend for whatever reason, because you're the next hot thing on Twitter. And our goal is to support you whether you are enterprise that's moving things up from the end frame or you are the next hot startup that's growing virally. And we've done a lot to build systems to help both sides and we are pretty proud of that. It's interesting if you think about open source where it's come from. I mean, I remember that days when open source wasn't open, I would be peddling software, hey, there's a free copy of Lynne Unix in college and now it's all free. But I mean, just what's changed now it used to be just free software. Download software, you got it. Now it's a service. Service now can be monetized quickly. And what you guys are offering with AWS and CloudScale is, you've done all these things as I don't have to, I'm a developer. I get the benefits of the scale. I can bring my open source code to the table, make it a service, integrate it in with other services and be the next Snowflake, be the next company that could scale. And that's the innovation, right? That's the, this is a new phenomenon. So it also changes the business model. So- Yeah, actually, you're quite right. Actually, I want more things to it. If you look at how a lot of enterprises use containers today, most of them are using something like ECS and Farvee or EKS to build an internal developer platform and the internal developer portal. And then the question then become this, how do you scale this modern development practices to an entire organization? What is your big bank that's been around as thousands and thousands of IT staff that may not all be experts at running Kubernetes how do you scale it out? That's where systems like Proton come into play. That was actually the inspiration is, how do you help an organization where they're building these developer portals and developer infrastructure, developer platforms? And how do you make it easy for them to build it? B, almost use it as a way to get these modern practices into the hands of all the business units where they may not have the time to become experts at the modern ways of running infrastructure because they're busy doing other things. And I think you'll see a lot more happen in that space. There's stuff happening in the open source community. There's Proton. There's a bunch of interesting things happening here. And we're interested to see how that evolves. And also, the communal aspect of not just writing code together but succeeding, building something. I mean, that's when you start to see the commercial meets open kind of ethos of communal activity of working together and sharing. A big part of this year's DockerCon is sharing, not just running and shipping a code, but sharing. Yeah, I mean, if you think about it, Docker's original value was you build, run and ship, right? You use the same code to build it. You use the same code to ship it, same sort of infrastructure interface. And then you run it. And the fact that the Docker image is such a wonderfully shareable entity and that can run everywhere is such a powerful, I mean, so it's called, it's an OCI image now. I still call it Docker images because it's this easier. But to me, that was a big deal. And I think it's becoming an even bigger deal over the years. I came from, before Amazon, I used to work in the sciences in bioinformatics and the ability to share code, share dependencies and package all of that up in a container image is a big deal. It's what got me, one of the reasons I got fascinated with container seven, eight years ago. So it'll be interesting to see where all of this goes. It's great stuff, great success. And congratulations, Deepak. Great to always talk to you. Got a great finger on the pulse. We lead a really important organization at AWS and Docker has such huge success with developers, even though the company has gone through kind of a changeover and a pivot to what they're doing now. They're back to their open source roots, but they have millions and millions of developers use Docker and new developers are coming in. Dotnet developers are coming in. Windows developers are coming in. And so it's no longer about Linux anymore. It's about just coding. Yeah, big job. And it's part of this big trend towards infrastructure automation and development and deployment practices that I think everyone is going to adopt faster than we think they will. But companies like Docker and open source projects that they're involved in are critical in making that a lot easier for them. And then folks like us get to build on top of that and or with them and make it even easier. Well, great, Tesla wanted Docker that you guys based your ECS on Docker. Docker has a critical role in the developer community by Run Compose and their hub with Docker desktop and we'll be watching Amazon and the community activity and see what kind of experiences you guys can bring to the table and continue that momentum. Thank you, DPEG for coming on theCUBE. Oh, thank you, John. It's always a pleasure. Okay. This is theCUBE's DockerCon 2021 virtual coverage. I'm John Furrier, host of theCUBE. Thanks for watching.