 Live from Boston, Massachusetts, it's theCUBE, covering OpenStack Summit 2017, brought to you by the OpenStack Foundation, Red Hat, and additional ecosystem support. And we're back. I'm Stu Miniman with my co-host, John Troyer. Happy to welcome the program, first time guest, a non-Christian who's the EVP and general manager of Cloud Insight Canonical. Thanks so much for joining us. Thank you. All right, so we had Mark Shuddleworth on this morning talking about how it used to be kind of three groups. And now there's kind of two main groups of the focus. You know, I think he said Ubuntu, winning everywhere. Of course we expect the man who's stepping into the CEO role to give that positioning. Most people I talk to know Ubuntu. I mean, I talk to neighbors, I remember back, you know, like John and I have background in VMware. It was like, you know, VMware desktop all over the place. Every, you know, person that uses Linux absolutely uses Ubuntu. Let's talk about cloud in your space. Where do you guys position yourselves? Where are customers using you? Let's start there. So a bit of background, as Mark says, three things that have sort of shrunk down to two. Canonical's origins were on the desktop. As a Linux for the desktop, as a Linux for developers. And that part of our DNA is as strong as it ever was. And so developers still, I think overwhelmingly, developed for Ubuntu on Ubuntu. And one of the target environments they would typically develop for is Ubuntu on the cloud, or increasingly Ubuntu on the edge. And those are the two sides of the company now. You have two divisions. I run the cloud division and Mike Bell, a counterpart to me, runs the IoT side of it. Now, if you think of cloud, for any, it's probably best to see this from a customer standpoint. For any of our customers, their data center has two sides to it. There's all the stuff they know how to do, which they figured out how to do 10 years ago or five years ago. This stuff has a lot of windows, my old world, a lot of Red Hat, a lot of VMware, and that stuff is just fine. And they have no desire to change it, other than keep it alive and keep it running. The new stuff, the stuff that fills their inboxes, is words like containers, cloud, infrastructure as a service, public cloud, hybrid, yada yada, all of that new stuff. That's the world which is evolving. It is a different class of software for them, and that's where they will find us more often than not. So, if you look in the public cloud, we were early to the public cloud. We didn't just treat it as yet another server, just virtual. We treated it as its own environment. And today Ubuntu will make up 70 to 80% of all the guest OSs on a public cloud, no matter which one you pick. We have relationships, we know all of them, we know we have relationships with about 35 partners in the public cloud space. And you'll find a similar mirroring as well of Ubuntu presence when you look at that class of workload on-premise. So, take Cloud Foundry, 90 plus percent of the time, that'll be deployed on Ubuntu, and a 70% of Docker out there runs on Ubuntu, as we see from the OpenStack survey every six months, something of the order of 55% of all the OpenStack out there, especially the large ones will be running on Ubuntu. We tend to be the default environment customers will go for as an operating system when they're looking at that new class of workload, containers, big data, et cetera. So, that's what the cloud division is really about, is helping people build first class, net new infrastructure which can be on-premise, it can be hybrid, it can be completely public cloud, we are indifferent to it. Our biggest customers, I mean just about anybody that you can think of that is a logo that you recognize the name of, the Netflix, the Uber, the Spotify, the Dropboxes of the world are all running on Ubuntu. Anybody you can think of with a Fortune 500 that is doing infrastructure scale is almost guaranteed to have significant estate now running on Ubuntu. So, that's the spread of the business and our journey if you will over the last five years. Great, that's quite a spread that you've got there. What's different in the cloud versus kind of where Ubuntu came from? I love your comment kind of about cloud, you know there's stickers, it's just somebody else's computer when you do there. Why is it, what's more to that? Why, what do you have to do special? Why do you have such strong affinity there? Our understanding of it has evolved because initially you're right and I've been in the cloud space for 10 years. It did start out as it's just somebody else's server, I just rent it now, I don't have to buy it. That evolves to, oh no it's actually, there are a set of things technically that are materially different when it's in a VM versus on a physical box and those are some of the first things that they did to a Linux distribution that nobody else did. Some of the reasons why Ubuntu became so popular in AWS was because we were optimized for it. Increasingly now, it isn't about the operating system, it's about the operation system. So I think Mark talked a little bit this morning. The class of software people are putting on these installations is just not your daddy's SAP installation or ERP system or CRM or some such. It is software composed of multiple pieces, typically integrated and configured per use specific to a custom user, scaled out across tens, maybe hundreds or even thousands of machines and with a chain cycle that is light years away from what you used to have. Nowadays we patch things every month, we upgrade them every six months and that is normal. Entities like an Amazon or a Google had to deal with this problem before they could deliver the services to deliver to us at scale. They figured out how to solve this 10 years ago. They did it at probably a massive cost but they all have internal infrastructure that runs a certain way. Most IT organizations haven't made that leap. They're still looking to operate an open stack the way they used to operate their legacy virtualization infrastructure. That is what is really new about this. Solving that problem will define the next two decades of how IT runs. And on its interest, we had theCUBE at both DockerCon and at Red Hat Summit and now we're here and you hear that whole containers are Linux, Linux is containers, who has the best expertise of course, where it all sits, what canonicals take on that container and who has the expertise, who's helping clients through all of this? Yes, so we've had containers as an option for a while. The bulk of our production open stack deployments are containerized and deploy open stack in containers. It's just something we never made a big deal of because to us it was just the way you would do it. Today there's a broader set of choices that customers have. Our conversation with customers ultimately hinges on the workload. Somebody wants to deploy media workloads fast, they have licensing rights that expire and they want to be in market in three months. They are indifferent honestly to whether or not that is done on an open stack or if it's done with Docker managed by Kubernetes, they just want it to be efficient. We try to deliver as many options to them as possible. So we've got large customers using us for open stack. We've got those same customers also starting to use us for Kubernetes. We deliver a distribution of Kubernetes that is supported by Economical. We don't bias the conversation one way or the other. And when it comes to containers, different strokes for different folks. Docker is great. We are one of their biggest partners. The vast majority of Docker deploy on us. We resell Docker and sell large deals to customers that will stand behind the installation and Docker is right beside us. But just as much, you'll get a bank and all they want is a better, faster VM. They want that VM to run faster and more of it on a single machine. The answer to that looks like Lexity which by the way, yes, happens to be a container. But that's kind of besides the point, they're looking for a better VM. So it isn't just about containers versus something else. We think VMs to different types of containers are all options. And the important thing is for customers to be able to have a choice and to change their minds about that choice. I love the way that you talk about this in a multi-valent way, in a multi, it's a multi-cloud, multi-platform world. It doesn't really matter, right? The customer at the end of the day, the business owner wants to be up and running and be able to change quickly. Yeah. I want to talk about buying center and design center for a little bit. In the olden days, infrastructure folks got to talk to the infrastructure folks and the sysadmins and we didn't really talk to the developers and that sort of, those sorts of people. Right. You know, Ubuntu, like as we started, right, long history working with developers. Yeah. When you go in and talk to a CIO right now, who's at the table with them? Are there, is it ops? Is it developers? Is it both? I mean, how are they looking at building this stuff? Yes. They have to, let me give you a few cases. It's probably the best way to illustrate it. We see a lot of organic Ubuntu because developers love us. It'll spin up as something that is a POC and it'll become production. Walmart is a large customer of Microsoft and of Red Hat and I'm sure VMware's as well. They're a very large customer to us. All of walmart.com runs on Ubuntu. That was not a decision we coaxed Walmart into making. That was made without us in the room. When they reached out to us, they were already running thousands of nodes of Ubuntu and Ubuntu OpenStack and powering the website with it. And that was the basis for the commercial relationship we had. Since then, the operations guys have told us that one of their biggest problems is bare metal provisioning at scale, which is something we happen to do very well in our tooling. So there's an ops conversation that expands beyond the original scope. So that's a good example of that kind of organic. We see a lot of this in the public cloud, certainly. Anybody large in the public cloud is almost certainly doing Ubuntu. They got there without asking us, but now they're asking us for specific set of things that matter at that scale, we can solve it. All the way at the other end of the extreme, you get telcos, big monstrous projects that are defining the next generation of telephony, big RFP process, everybody comes to the party, it's months before you kind of get to that process. There it is absolutely a IT and business led conversation that checks all the boxes. One really interesting thing across all of this, and speaking on the customer side now, if you go into a Spotify and ask them if there is an IT department versus a business division, they're probably going to look at you funny. Most legacy enterprise IT, enterprise organizations are figuring out now that they're all software companies. And so you see this a lot in telco, where our dialogue is not just with IT, with infra, it is just as much with the business guy that owns monetizing telephony, but now it relies on deploying telephony as applications on a cloud delivered by what used to be a service and is now a partner to him. They are both at the table, both dictating demands and very willing to go off and do their own thing if the other partner doesn't cooperate. That blurring at the line between business and IT is I think a necessary dynamic in the market. It is what allows ultimately a sky to compete with the Netflix in the longer run. Yeah, it's a fascinating discussion. I remember I think four or five years ago we did like a big info graphic on the changing role of the CIO. And it's like as cloud comes, are they a broker, they have some inside, some outside, that they handle a whole lot of this. You know, you could get a comment a little bit more. What do you see the role of the CIO going forward? You know, business versus CMO versus, you know, the application and the IT people? No, let me tell you a story. So Mobile World Congress earlier this year, Deutsche Telekom, one of our biggest customers, they were one of the first large telcos to start the journey into NFV, network function virtualization, which think of it as telephony done in software versus black boxes. DT has been a big partner to us, we're very much involved in that project. It is a monstrously complicated multi-country engagement. They invited all of the entities selling them applications that are part of the project to basically an open meeting at their booth. And there must have been about 40 people sitting in the audience. And up on stage was initially the head of the whole project, but subsequently two people. One heading up in for, and the other one more responsible for the business side of it. And they collectively had a single message back to the crowd, which is all about how the software ecosystem needs to iterate its own business models in order to meet the business need. Someone needs to roll out a new service in a country and needs to do that across four countries now. You are not going to have completely different procurement processes for each one of those. That's a requirement of the software vendor. In that conversation for me, without the titles, it was really difficult to tell whether I was speaking, hearing from the business guy or the IT guy. And that seamless side-by-side engagement is to me very much the kind of template I think most CIOs need to start thinking of. It isn't about being a broker. That world is gone. You absolutely have to be a partner and in some ways quasi-owner of the business. You know, WhatsApp has done the telephony. 32 engineers servicing 500 million users. You don't get that in a world where somebody owns the user and somebody else owns the infrastructure. Narn, you talked about that. Cloud Foundry tends to be on your platform a lot. Some of the other services. What's your talk about business and technology? How about the application and the infrastructure? Where does kind of that application modernization fit in with Canonical? How do you help customers? Or is that mostly through partners? Yeah, so this kind of goes right to the back of what I was saying earlier about operations. Nobody's building a cloud just to have one. They want to run stuff on it. The core, the real learning lesson for the customer comes when they start using their cloud to deploy things on top of it. Making that easier. The vision is that you'd want somebody to take components off a catalog, stitch them together in a way that makes sense for that user, and then deploy them to one of many target environments. That's what you want. What you have in practice is I'll put a change request in, a bunch of forms get filled out, some scripts get deployed. That is only specific to this one environment. You've changed your mind under that's too tough, that's another three months, that's to an RFP. That is the world we live in. This elegant model-driven user-initiated process is what we want to get to. The heart of that is where we think IT has to go from being run by people to being run by software, being run by a model that is designed by people. That's how Amazon runs their data center. That's what every organization has to do. And the heart of it, what you're really doing is not building infrastructure, but letting people deploy applications on top of it. That model-driven approach is at the heart of what we do, and it goes beyond just having Ubuntu. It's really about the tooling around it. All right, Anand, what conversations are you having with the users here? What's their state of mind? What would you want to share with our audience? So here, and this has been the case to an increasing degree in the last two OpenStack summits, my first ODS was two and a half years ago. This is my, you know, fifth one now. Two and a half years ago, it was very difficult to explain to somebody why building a cloud was the easy part. The really hard questions were, how many people does it take to run your cloud? A great answer would be one or two. A good answer would be four or five. You're north of 10, you're in ugly territory already. Many customers would tell you it's 15, 20, 25. It is very difficult to explain why headcount cannot scale with server count, why operations has to run on an economic benchmark set by AWS, like that is what you should be comparing yourself to in cost per cycle. Now we find two conversations, two types of conversation. We've got the customer that tried it the old way, has stopped the project after phase one went live and is now back and they want to do it the right way. And we're selling into customers that we didn't succeed in selling into two years ago. That is one class of customer at ODS. And then the other end of the spectrum, we get the customer that comes up and says, I want to do this, it's got to be on premise. Honestly, I don't care if it's open stack or not. I just want it to be done as a service and done quickly. For two years now, we've had a service called Bootstack. Build, operate and optionally transfer an open stack. It's a managed open stack. They just care about the workload functioning and they just want that delivered on a timeline. And those are the two. So in one case, it's very much, I love open stack, but I've learned what not to do about it. In the other case, it's very much a, I just want this done as a service. I want IT as a service. Can you make that happen for me? You talked about moving application development, old school, new school, the same conversation does also happen in public cloud, on-prem. Are you, as canonical, obviously works well with both, right? Partner with all the big players. It's still a Windows world and part of the world. I mean, how do you see your customers adapting to that? Are they, it seems to be hybrid, multi-cloud. I mean, is that a real factor in 2017? So I guess we see, there's bleeding going on in three directions. You can take your core infrastructure and throw it up into the public cloud and get hybrid. And we see a lot of customers. A lot of our public cloud customer base started out small, entirely on the public cloud. We'll then get to a size where it makes sense for portions of their infrastructure to come off and run on-premise. Box is a customer of ours, as many other logos, that did that last year, right? They came off of AWS for a portion and they're truly hybrid in how they run. That's one way, sort of, you think about both balance. Here's where Netflix didn't start out on Ubuntu. They were on Amazon Linux. Back when Amazon Linux was the default Linux on Amazon, they took the effort to port to Ubuntu. One reason is because we do a set of things that keep them at a cutting edge, but one reason is also because it gives them the flexibility then to be truly hybrid in their choices. And that's important. So that's one choice you see, and there in both the tooling, but also having a consistent experience across on-premise and off, is very important to the customer. It's important to us as well. The other kind of thing you see then is laterally, where someone wants to keep OpenStack, but they also want to give that customer a bare metal option, because the drill performance runs on bare metal or a container option. Multiple substrates is how we think of it. Right now, those look like independent projects. What we're trying to enable for them, what we do with JuJu, is the ability to take that workload and decide at deployment time where you want it to go and make that truly seamless behind a single-painter class. That's what the model of it allows you to do. Those two, everybody talks about. The third one, the most nascent one, and the most interesting one in some ways, is the spectrum from that data center down to the edge. So take a telco. Yes, I've got lots and lots of servers and I've got applications deployed. I've also got a different type of server. It's called a set-top box. It just happens to be a server because it's really got that much compute in it. I've got this spread out across millions of households, and right now it's sold as a dumb unit which comes to you and gets upgraded once every two years of the firmware. You want to put apps on it and configure those apps and treat them just like you do a server. That is a different class of infrastructure. There's no reason for that class of infrastructure to be treated differently from the one that's in the data center. And again, Ubuntu Core is how we think of that as a platform and the management of it to us, it's in the same class of problem. So would it be too much to say you'd like people not to think of Ubuntu as an operating system anymore but as a set of platform and services that can give you a consistent application environment in all these different areas you talked about? That's a great way to put it. We think of it as an operation system for at scale infrastructure. That to us is the defining problem in IT for the coming decade. A Linux operating system is absolutely a great part of it but it is a part of the solution. All right. And on Christian, perfect place to end there. Thank you so much for joining us and we'll be back with more coverage here from the OpenStack Summit. You're watching theCUBE.