 Live from Austin, Texas, it's theCUBE. Covering DockerCon 2017, brought to you by Docker and support from its ecosystem partners. Hi, I'm Stu Miniman joined with Jim Kobielus. Happy to have at the end of our two days of live coverage here with DockerCon 2017 on theCUBE, we've got a practitioner. And also was one that did a great presentation in the keynote this morning. Happy to welcome Swami Kosher Lakota, who is the global head of infrastructure and operations with Visa. So, Swami, welcome in what's in your wallet. Yes. I all have Visa cards in the wallet, right. It's in your container, you're in DockerCon. That's a good one, Jim, I like that. So, we were really impressed. I tell you, social media was lighting up, going through your case study, talking about how kind of the before, very much a virtualized environment looks like many data centers I went through that digital transformation, if you will, as to what you're doing. Before we get into kind of your case studies, tell us just a little bit about you, your role, how much you're flying around the world with the infrastructure and ops. Right, so I'm responsible for Visa Inc's operations and infrastructure. So, my responsibility is to kind of run everything that's inside Visa that does the payment processing. So, that's my responsibility. I travel a lot, we have a global team, Visa is a global company. So, I'm on the road a lot. Yeah, so, I love the case study you did today because it's always what we want to do as analysts. But, let's talk about the pain points, let's paint the before, you know, how did everything go through and point the after. So, we want to encapsulate a little bit of that. As I said, highly virtualized environment. What was the pain point? What was the objective? You know, I don't think any executive came down and said, hey, containers are awesome, let's just do stuff because it sounds cool. What was the real business driver for what you were doing? Right, so like I mentioned in the keynote, the number one priority that we have for my organization is to make developers productive. You know, we take this as a challenge where any employee, a developer who joins Visa, we want them to be able to write code and publish code into production on the same day. That's what we're aspiring to go. That was the driver. So, we're looking at every minute of what it takes to get the provisioning and we're trying to streamline it so that we can deliver that vision quickly. Great, and a virtualized environment, you know, going to containerize, can you talk, how big a scope is this, you know, did it change your underlying infrastructure itself or, you know, can you maybe flesh that out a little bit for us? Right, so when I talk about provisioning or in general managing operations for a large organization or large enterprise, I look at it from two dimensions. One, the one-time provisioning, but then most of the challenges and the opportunities are in the life cycle, right? Yes, virtualization solves the one-time provisioning, but we haven't really solved the managing the life cycle. Right, you know, it still takes, even if it takes one day to two days to provision the virtual image, the pre-provisioning task, post-provisioning task, and care and feeding of it, whether it's patching and maintenance, doing tech refresh, it's still very intrusive and very painful. So when we looked at the whole problem, we wanted to solve all of them at once, and that's when the containerization and microservices attracted us. Swami, when this rolled out, I mean, is this 100% your environment in this new one, or have you been doing a phased approach, you know, how does it look today? I wish it's 100%, but we are in the early stages, you know, we have one application. It is now a Tower of Success, and we have about five other application teams that are looking into it, and then we build more towers of success, and then this becomes kind of like the part of the standard offering. So the initial implementation was a pilot, a showcase of the technology, or explain where the idea for this initial implementation came from, was it driven from the business level or from the technical level? It's a partnership between the application team and the infrastructure teams. The boundaries between the two teams are kind of blurring, and at Visa, one of the great things that we have is that we collaborate very well internally. So when we wanted to have this mission of making developers product here on the same day, when an application team is going through the refactoring process, we basically said, hey, maybe we can join forces, and we have a good collaboration, and we said, let's do it together. So there was a refactoring project already underway, and then containers and Docker came in to the overall equation, essentially, or that project. They both came at the same time. It's a perfect marriage at the time. The timing of when we want to do refactoring, and the timing of when containers came at the same time. So you say it was a success, and essentially it was proved out internally within your development organization, and tell us, what other areas of Visa are likely to become containerized fairly soon in terms of the core applications? There are about five other groups that we're looking at that still is work in progress. The next use case that I'm excited about is the kind of like the batch use case, right? That's about as much as I can say about the rest of the five work those, right? Swamy, there's a certain set of data services that you get when you have a virtualized environment. Can you talk to us a little bit about that difference going to containers as to how much was seamless? How much did you have to plan? I think about things like high availability, security that I'm sure are important to you guys. Right, see with the way we have done the implementation is to kind of preserve the same high standards that we have with availability and security as well. In fact, I would argue that availability is higher because now we have instrumented those microservices in a way where we know exactly when we need the help of another service, right? And then we can create this. So from an availability perspective, it's better than what we have today, which is already good, right? Because we can scale up and down. We know when the system needs more resources. And from a security perspective, even before the implementation, we made sure that the area is rock solid and it has the right controls for us. One of the slides we liked that you did in the presentation this morning was talking about utilization. We know that most companies are not utilizing most of it. First of all, forecasting what you're going to use, when you're going to use it is really tough either. Overdo it, you'll underdo it. You've got way too much gear sitting there. You're really transforming yourself to be an internal service provider. Do you have any key metrics as to how much greater utilization, what that means to the business to total cost that you need to be concerned with? See, the absolute numbers of how we'd like to see our infrastructure be 90% utilized, 80% utilized, these kind of old school in my mind. That's audacious. That's what you like to do. That's your goal. Say that again. That's your target utilization. Well, I don't have a target utilization. That's what I'm saying is that using a particular watermark as your target utilization is old school. It should be elastic. Because sometimes when we do campaigns, we don't know what type of a workload we would get. We just are focusing on just run enough and then only grow when you need. This is why we call it as just in time infrastructure. That way it is only provision when you need it. When you're done with it, we'll do provision. With the microservices and how fast we can get the containers, you could do that. Okay, how about the operational impact to what you're doing? How much retraining did you do with your professional services you need to have come in? Was it changing roles or was there any changing headcount? I know it's just the one application you've done so far, but where is it today? What do you see as you roll this out further? From an operational perspective, the skill set mix will change. Instead of having eyes on glass when you have an operational issue, you are working in a predictive environment where you can proactively say that a particular outage can occur. The skill set may change, but in terms of the size and scale of the operation at least in the early stages that we are in, we don't see a whole lot of shifts there. Were there any surprises when you rolled this out or anything that you look back that you say, okay, well, now when I go to the next five groups, I'll be able to say, oh, we'll do this faster. You need to plan this a little bit differently. What lessons learned can you share? So there are four lessons learned that I mentioned. One is very important to have the right granularity for your service. When you take a monolithic service and then divide it into 10, 12 microservices, you got to make sure that the granularity is right. That's number one. And you don't want it to be overly granular. You don't want to keep it monolithic. And the second thing is you are reducing that to microservices. However, from a memory perspective, you have to make sure that you're not asking for a whole lot of memory as well. You cannot have the same heap size as that of a big monolithic service. Microservices needs to have smaller footprints so that you can run more and get the more utilization of your hardware. Because everything is memory bound. Even in your virtual environment, it's not a CPU, it's all memory bound. All right, Swami, you've been interacting with a lot of people. What did you take of the show so far? Interaction with your peers. Are you able to find a lot of other companies that have similar challenges that you can share experiences with? So a couple of things that I liked about the announcement that Docker has made. One, the enhancements that they're making on the security are very valuable. And then the whole notion about secure supply chain is very relevant. And the second thing that I liked is how easy it is now to take a virtual workload and then putting them in the container. So I think the announcement that they have made was attractive. As far as the explore floor is concerned, we're not putting our workloads in public clouds anytime soon. We're still building private clouds and then hosting it inside there. There's a lot of people on the expo where they're offering services for the cloud. For example, DataDog and stuff like that. I wish we had seen more on how to manage large container deployments from an operations perspective and any innovation there. But I also haven't had a chance to completely sweep through the floor also. You know, in the keynote I heard it was, containers are everywhere you want to be. Did they take the tagline from Visa everywhere you want to be, you know? Maybe, maybe, I did not notice it, but yeah. All right, Swamy, I want to give you the final word. As you look out, you know, things you're excited about in the ecosystem or anything, the feedback that you'd give to be able to make your job easier, you know, help you move this forward even more into your environment. You know, I think the one thing that I would say is that in order to be able to transform enterprise IT and take the innovation as rapidly as we would like to have, every enterprise, the infrastructure and application teams have to partner. And the barriers between them should be collapsed and they should, you know, innovate together, collaborate together, that I would say is number one. And number two, the ecosystem is becoming complex, right? It's difficult to navigate, you know, what should be, because for every single member and partner of the ecosystem, you have more than one choice. So picking at the right stack is very important as well. Swamy, really appreciate you joining us, sharing online and, you know, the really great kind of encapsulation notes. I had said, you know, early on in this wave of Docker, it was like, oh, maybe Docker can help free us from the infrastructure, but of course, we know there's relationships they need to go together. And, you know, as we're maturing, that complexity is getting better. Thank you so much for joining us and sharing with this community and ours. Jim and I will be back with our wrap up here from the two days of live coverage. Thanks for watching theCUBE.