 From around the globe, it's theCUBE with coverage of KubeCon and CloudNativeCon Europe 2020, virtual, brought to you by Red Hat, the CloudNative Computing Foundation and ecosystem partners. Hi, I'm Stu Miniman and this is the Kube coverage of KubeCon, CloudNativeCon 2020 in Europe, the virtual edition, and you've reached the final stage. This is our last interview, so hopefully learned a lot, talking to the CNCF members. We've had a few great practitioners, of course, some of the important vendors and startups in this space, and when we talk about what's happening in this CloudNative space, one of the things that gets banded about a lot is scale. What does that mean, when it first rolled out, of course, there is only one Google out there and only a handful of true hyperscalers, but there absolutely are some companies that really need scale, performance, global, and so happy to bring in, here's the final boss, it is Alexandre McLean. He's a technical architect at Ubisoft, and yes, I do have a little bit of background in gaming, but here is someone that is helping enable one of the largest gaming companies in the globe, so Alexandre, thanks so much for joining us. Hey, thanks for the invitation, happy to be here. All right, so you're no novice to this ecosystem. I know you and I have both been at many of the Dockercons, the Kubecons over the years, so if you could just give our audience a little bit of your background and what is your team responsible for at Ubisoft? Okay, sure, so I am part of one of the IT teams inside Ubisoft, so we're responsible mainly to provide cloud computing resources and Kubernetes infrastructure for the whole company. So, okay, and if you want to know more, well, basically I've been leading the Kubernetes industry the past few years right now, so we started the journey maybe in 2016. We were already pretty busy working on the growth for the cloud, the cloud initiative inside Ubisoft, for the growth of the expansion in different data centers and supporting the needs of the different teams and development teams inside Ubisoft. And one thing we wanted to do back then was really to enable and accelerate the adoption of cloud-native, the cloud-native mindset and cloud-native architecture, so what we did back then is we did a short analysis of our different technologies that was available at the time and we did decided to jump full head on Kubernetes and make this as the foundation for the different workloads, container workflows that will enable drive of adoption inside Ubisoft to grow and boost the productivity of many things. All right, I'm really glad you brought up that cloud-native mindset. If you could just up-level it a little bit for the business leaders out there, they hear about Kubernetes and they won't know how to spell it. They hear something like a cloud-native mindset and they're saying, I don't understand, what does this mean for our business? So what architecturally are you doing and what does that mean for your games and ultimately your end users? Yeah, so I would say that basically, if you want to have a cloud-native architecture, really want to make your application, first of all, very portable, very easy to deploy and manageable and at the same time, very resilient to failure. So you want to make sure that your application, once it's deployed, that it's highly resilient to failure, that it was built for failures and that you can manage the project and the service to meet the expectation of either the gamers or the service owners, basically. Yeah, absolutely. I'm curious. Here in 2020, we see the ripple effects of what the global pandemic has. I have to imagine that from a gaming standpoint, that has had an impact. So maybe if we can use that as an analogy, if it's valid from your standpoint, I have to imagine more people are using it. What did this mean to your infrastructure? How were you ready, from an IT's perspective, to support that increased usage, kind of rippling around the globe as more people are home all the time? Yeah, that's a good question. I guess, I mean, we really have like two kind of, I would say, audience inside Ubisoft in the IT team that I serve. So we have the people who are building the softwares and the applications who help the developers to, I mean, the game developers in general. So we have different services, internal services and tooling that needs to be hosted somewhere. And we need to enable these people and these teams to have a way to manage your applications efficiently. And the other side, we are looking at right now, I mean, the game server and the game industry is really, I think there's a shift right now in the way that, a shift of paradigm of the way that you're going to manage the game servers in the future. And I would say that back then, there was a lot of in-house tooling, things that were really, I mean, proprietary to each gaming company. But right now, what we wanted to do in the past few years, we worked for instance on a solution called Hagoness. So we were involved in the beginning to design this kind of next-gen game server, dedicated server hosting infrastructure that was all built around Kubernetes. So in the future, we already started to work on that and the next gen of games are going to be differently hosted in top of Kubernetes, which is going to enable a lot more efficiency of resource usage. And at the same time, we would say manageability and operability about all these services. Because I think that one key thing about CloudNative and Kubernetes is that, once you look at Kubernetes, I mean, basically it's very easy to onboard new people in the team, in the project, because they know what is Kubernetes and how to operate it. So it will be much more efficient in the future for all the workflows that we have internally and the next game server infrastructure as well to be a stable in Kubernetes. It's going to be much more easy to standardize and unify that whole stack. Well, skill sets are so critically important and it's great to hear you say that onboarding somebody in Kubernetes is easier than it might have been a couple of years ago. If you could bring us inside a little bit, what's your stack look like? Can you say what cloud or cloud you use when it comes to Kubernetes, what are the key tools that you're using and partners that you have? Yeah, sure. So early on, I would say almost 10 years ago, we really started to focus on having on-prem cloud infrastructure and the technology that we chose back then was OpenStack. So we have a large footprint of OpenStack Cloud installed internally and different data centers all around the world. So people and different teams and anyone at Ubisoft can easily have compute resources available for them. And with Kubernetes, we initially, we wanted to make Kubernetes a commodity. We wanted to have people be very in a position to easily experiment new things, new applications on top of Kubernetes. And for that, we decided to go with Rancher. So Rancher is an open source solution. They are made by Rancher Labs and we initially, although we started to build an in-house solution the first year because we talked back then, the landscape was quite different and we talked, it was the best choice for us to do. But we realized shortly after, I mean, when Rancher 2.2 came out, I think it was in something like April, 2008 thing that we will benefit a lot to go with this kind of solution which was open source. There was a lot of traction behind it and it will enable us to accelerate the adoption of Kubernetes and cloud native in general, much more faster than the in-house solution that we had built at that time. So we went with Rancher and right now we have, I would say, I mean, we have maybe 10 data centers with the cloud installed on top of it, much more data centers are going to grow in the next couple of months and years and we have over 200 clusters and 1,000 nodes that are managed by Rancher and people can just deploy on demand to run Kubernetes cluster and get started with it if they want to. Okay, so if I heard you right, it's Rancher on top of the open stack solutions in your data centers. Yes. You talk about how many clusters you have. What's the state of managing those environments? You said you're using Rancher, that's one of the things we've seen a lot of discussion over the last couple of years is, you know, went from managing containers to managing, you know, pod or cluster to now multi clusters around multi sites. You know, what's the maturity today? Anything that you're looking for that would make your life easier to manage such a broad environment? Yeah, well, I would say that's one of the drawbacks. I mean, when we enabled that solution with Rancher with DSE, I mean, the keys of use of launching and provisioning new clusters is that right now we have a lot of clusters, maybe too many because we try to consolidate. I mean, the next logical step for us is we try to consolidate the workloads maybe as much as possible and see if there's really a need for people to have their own dedicated cluster for them. And initially there was a lot of demand for that because people basically they came to us and they said, you know, we want to use Kubernetes and what we want to do is we want to have full administrative access to it. We want to be able to do whatever we want to be able to upgrade it at our home pace. And I don't want to have any neighbor on it. I want to be completely isolated in terms of compute resources. So we said, all right, we're going to make a solution that's going to provision new clusters on demand for everyone. And Rancher stuff stepped in very well. But now after a while, we some people and we especially as an IT provider and operator we realized that, you know, maybe people don't have to be completely alone on their cluster. Maybe we should try to consolidate that a little bit. So we're trying to migrate workloads from certain services and tooling and say maybe you can instead of running your own cluster you can use this one that is going to be shared and there will be a team dedicated, I mean, dedicated to support and operate this cluster for you because we want to, in the end, we want to afloat the burden of infrastructure and Kubernetes although it's, I mean, it brings a lot of abstraction and simplicity, you still have to manage your cluster in the end. So we'd rather have people focus on the application side than on the Kubernetes and infrastructure side. So we will start a path of maybe try to consolidate the different workloads and see if we can reduce the amount of clusters that we have and also to unify the way that people are using the different providers because although we have a huge open stack cloud offering internally on prem, there are still people who need to use GKE or AKS and external cloud providers. So for these people, some of them are not using really Rensher, although it's possible with Rensher, they're just directly using the providers. But what we want to do is try to unify the way that you're going to get access to this cluster, try to make a central governance model for people to pass through a central team to get access and project the cluster so they will be standardized and we will be able to add more, maybe security policies and compliance and rules and everything so that cluster will be created in certain ways and not too much fragmented as they are today. Yeah, that's ultimately what I was trying to understand is most customers I talk to, they have hybrid environments, they use multiple clouds. If you're using Kubernetes, how do you get your arms around that? So I'd love to get your viewpoint just because you've been involved since kind of the early Kubernetes days. What's better now than it was a few years ago? I heard you say that you looked at possibly creating a solution to yourself, a company like Rensher helps simplify things. So when you look at the maturity, how happy are you with what you have now and are there any things that you say, boy, I'd love my team to not have to worry about this. Maybe the industry as a whole would be able to standardize or make things simpler. Well, when we started to use Rensher, maybe there were a couple of things that we wanted to simplify for the users, because what Rensher does, it essentially is that there's a lot of configuration options. It's very flexible because it supports many providers. So the first few things that we did was try to simplify the user experience. So we modified Rensher's certain ways to make it simpler to be consumed. And although the experience is much more simpler than it was, let's say two years ago when we started, we still want you to simplify it even further. We want to ideally provide a fully managed experience so people don't even have to worry about the control plane components that is currently being deployed with their commit as clusters. So we want to remove that away from them. So there will be, once again, fully focused solely focused on the application side of development. And I think one other aspect that we need to maybe improve in the future is that once you want to deploy your application and make it resilient and geographically distributed, then you need to manage multiple clusters and you need to deploy your applications on top of multiple clusters. So the whole multi cluster aspect of things like how do I deploy my application from a version? How do I make it consistent between the different clusters that where it needs to be deployed? How do I make service discovery possible? Or do I mesh everything, all the application together to make sure that it's easy to operate? It's easy for the developers and that it's resilient in the end. So we'll start to look at the, I mean the multi cluster multi region aspect for Kubernetes because that's a big challenge to us. All right, well, Alexandre, I want to shift for a second. Let's talk about the conference, KubeCon, CloudNativeCon. Obviously it's virtual this year so there is a little bit of shift but you've attended many of these in the past. What, are there projects that you're interested in learning more? Are there peers of yours that you're looking to collaborate with? What have you seen in the past that you're hoping you still get from a virtual event like we have this year? Well, I think that it has become so big. It's hard to keep up with everything that's happening at the same time nowadays but things that we're looking at really is maybe like, I think that there's maybe in terms of service mesh, there are a lot of technologies. I think it's measuring slowly so we'll always start to have a look about what is the best fit for us and the use cases that we have and some people probably are using Kubernetes and some other people are using more traditional stacks so we try to bridge that together and see what's possible to migrate the existing workloads from the traditional cloud VMs and cloud applications toward Kubernetes and everything. So maybe try to see what's possible to bridge that path and migrate gradually for the users that we have and other things in general, I think that it will be very interesting to see all the whole Bay of SecOps evaluate right now and see how we can try to add conformance and compliance rules to different clusters that we have to manage to make sure that it's no longer like this ad hoc matter of I want to create a cluster, I get access to it. We need to centralize the governance, we need to centralize the rules of all, everything's going to be managed in the end and make sure that security is a big aspect to it. So make sure that there's no vulnerabilities and everything's being audited and especially for the game servers is going to be a big factor for us. So we're definitely interested into all the security discussions that's happening right now. All right, no shortage of lots of information. Alexandre, by the way, that there's no way that anybody can keep up on everything that's happening in this very robust community, but thank you so much for sharing your journey. It's always great to hear from the practitioner. Thanks so much for joining us. Thanks for having me. Awesome, thank you. And thank you for joining us for all the coverage. Be sure to go to the cube.net. You can see not only all the interviews from this show, you can go search, find previous shows as well as see what events we will be at of course right now all virtually. So I'm Stu Miniman and thank you as always for watching the cube.