 Live from San Francisco, it's theCUBE. Covering Red Hat Summit 2018, brought to you by Red Hat. Okay, welcome back everyone. We are here live, the CUBE in San Francisco, Moscone West for the Red Hat Summit 2018. Exclusive coverage, I'm John Furrier, the co-host of theCUBE. I'm here with my co-host, John Troyer, who's the co-founder of Tech Reckoning, an advisory and community development firm. Our next guest is Jonathan Donaldson, technical director, office of the CTO, Google Cloud, former CUBE alumni, formerly with Intel, been on before, now at Google Cloud for almost two years. Welcome back, good to see you. Good to see you too, it's great to be back. So, had a great time last week with the Google Cloud folks at KubeCon in Denmark, Kubernetes, rock on the world. Really, when I hear the word de facto standard and abstraction layers, I start to get my bells going, hmm, let me look at that. Some interesting stuff, you guys have been part of that from the beginning with the CNCF, Google, Intel, among others, really created a movement, congratulations. Thank you, yeah, I mean, it really comes down to the fact that we've been running containers for almost a dozen years, right? Four billion a week, we launch and collapse, and we know that at some point as Docker and containers really started to take over the new way of developing things, that everyone's going to run into that scalability wall that we had run into years and years and years ago, and so Craig and the team at Google, again, I wasn't at Google at the time, but they had a really, like, hey, let's take what we know from internally here, and let's take those patterns and let's put them out there for the world to use, and that became Kubernetes, and so I think that's really kind of the massive growth there, is that people are like, wow, you've solved a problem, but not from a science project, it's actually from something that's been running for a decade. And that's called Bore, that's the tools that Google used to, that their SRE site reliable engineers used to massively provision management, and they're all software engineers, it's not like they're operators, they're all like Google engineers, but I want to take a minute if you can to explain because you're new to your Google Cloud, you're in the industry, you've been around, you've helped form the CNCF, which is the Cloud Native Foundation, you know Cloud, you know tech, Google's changed a lot, and Google Cloud specifically has a narrative of, well, you know, they're one big cloud, and they have an application called Google Stuff, and the enterprises are different, you've been there now for almost a year or more. A little over a year, yeah. What's Google Cloud like right now? Break the myths down around Google Cloud, what's the current state? I know personally, a lot of cloud DNA is coming in from the industry they've been hiring, making some great progress, take a minute to explain the Google Cloud. Yeah, so it's really interesting. So again, it comes to back from where you started from, and so Google itself started from a scale, consumer, you know, SaaS type of business, and so that they understood really well, and we still understand obviously, you know, uptime and scalability really, really well, and I would say, you know, if you backtrack several years ago, as the enterprise started to really look at public clouds and Google Cloud itself started to spin up, that was probably not, they probably didn't understand exactly all of the things that an enterprise would need, as really at that point in time, really no one cloud understood any of the enterprise specifically, and so what they did is they started hiring in people like myself and others that are in the group that I'm in, they're, you know, former CIOs of large enterprise companies or former VPs of engineering, and really kind of our job in the Office of CTO for Google Cloud is to help with the product teams, to help them build the products that enterprises need to be able to use the public cloud, and then also work with, you know, some of those top enterprise customers to help them adopt those technologies, and so I think now that if you look at Google Cloud, right, they understand enterprise really, really well, certainly from the product and the technology perspective, and I think it's just going to get better. I interviewed Jennifer Lin at a one-on-one with her, I didn't publish it, it was more of a briefing, she's runs product management on the security side. It's fantastic. So, you know, we, she's checking the boxes, so the table stakes are kind of like set for Google, I know you got to do some basic things to catch up to get in the cloud, but also you have partnerships, Google Next is coming up, theCUBE will be there, Red Hat's a partner, talk about that relationship with Red Hat and partners, so you're very partner-centric with Google Cloud, and that's important in the enterprise. Yeah, there tends to be kind of two main areas that we focus on from what we consider, the right way to do cloud. One of them is open source, right? So having the, which again, aligns perfectly with Red Hat is putting the technologies that we want customers to use and that we think customers should use in open source, right? Kubernetes is an example of those, Istio and others that we've put out there, examples of those, you know, a lot of the open source projects that we all take for granted today were started from white papers that we had put out at one point in time, kind of explaining how we do those things. Red Hat, and from a partner perspective, I think that that follows along, right? We think that the way that customers are going to consume these technologies, certainly enterprise customers, are through those partners that they know and trust, and so having a good flourishing ecosystem of partners that surround Google Cloud is absolutely key to what we do. They love multi-cloud too, I mean, can't you? They love multi-cloud, and we do too. I mean, the idea is that we want customers to come to Google Cloud and stay there because they want to stay there, because they like us for who we are and for what we offer them, not because they're locked into a specific service or technology, and, you know, things like Kubernetes, things like containers, being open source, right, allows them to, you know, take their tool chains all the way from their laptop to their own cloud and inside their own data center to any, you know, cloud provider they want, and we think, you know, hopefully they'll naturally gravitate towards us over time. One of the things I like about the cloud is that there's a flywheel, if you will, of expertise, like I look at Amazon, for instance, they're getting a lot of metadata of the kinds of workloads that are on their cloud, so they can learn from that and turn that into an advantage for them or not, or for their customers and how they can do that. That's their business decision. Google has a lot of flywheel action going on. A lot of Android devices connected in the Google system. You have a lot of services that you can bring to bear in the cloud. How are you guys looking at, say, from a security standpoint alone? That would be a very valuable service to have. I can tap into all the security gutting goodness of Google around what spearfishing is out there or things of that nature. So are you guys thinking like that in terms of services for customers? How was that laid out? So we're very consistent on what we consider is privacy is number one for our customers, right? Whether they're consumer customers or whether they're enterprise customers, right? The where we would use data, like you mentioned a lot of things, but where we would use some data across customer bases are typically for security things, right? So where we would see some sort of security impact or an attack or something like that. This started to impact many customers and we would then kind of aggregate that information. It's not really customer information, it's just like, like you said, metadata or themes or trends, right? And yeah, we're not monetizing it, but we're actually using it to protect customers. But when a customer actually uses Google Cloud, that instance, that is their hermetically sealed environment. In fact, I think we just came out recently with even the transparency aspects of it where it's almost like the two key type of access for if our engineers have to help the customer with a troubleshooting ticket, that ticket actually has to be open. That kind of unlocks one door, the customer has to say yes, that unlocks the other door, and then they can go in there and help the customer do things to solve whatever the problem is. And each one of those is transparently and permanently logged, and then the customer can at any point in time go in and see those things. So we are taking customer privacy from an enterprise perspective. And you guys are also in a whole building from Google proper. Like it's a completely different campus. So that's important to note. It is. And a lot of this just chains on from Google proper itself. I mean, if you understood just how crazy and fanatical they are about keeping things inside and secret and not proprietary, but like not allowing the customer data out, even on the consumer side, it would be. Well, you got to amplify that, I understand. But also I see a good side of that, which is there's a lot of resources you're bringing to bear, or learnings. Absolutely. The SRE concept, for instance, is to me really powerful, because Google had to build that out themselves, right? This is now a paradigm we're seeing a cloud scale here with the cloud native market bringing in all new capabilities at scale. Horizontally scalable, fully asynchronous, microservices architecture. This future is a complete game changer on functionality at the different scale points. So there's no longer the operators running provisioning storage here. And this is what we've been doing for years and years and years. That's how all of Google itself has all search and ads and Gmail and everything runs in containers, all orchestrated by Borg, which is our version of Kubernetes and stuff. We're really just bringing those learnings into Google Cloud and to those customers. Jonathan, machine learning and AI has been a big topic this week on OpenShift. Obviously that's a big strength of Google Cloud as well. Can you drill down in that story and talk about what Google Cloud is bringing and machine learning on OpenShift in general? Give us a little picture of what's running. So I think they showed some of the service broker stuff and I think they showed some of the Kubeflow stuff, which is kind of taking some machine learning and Kubernetes underneath OpenShift. Like I think those are like very, very interesting for people that want to start getting into using auto ML, which is kind of roll your own machine learning or even the voice or vision APIs to enhance their products. And I think that those are going to be keys. Easing the adoption of those, making them really, really easy to consume is what's going to drive the significant ramp on using those types of technologies. One of the key touch points here has been the fact that this stuff is real world and production ready. The fact that the enterprise architects are now rolling out apps within days or weeks. One of those things that's now real is ML and even in the opening keynote, they talked about using a little bit of it to optimize the scheduling and what sessions were in which rooms. As you talk to enterprises, it does seem like this stuff is being baked into real enterprise apps today. Can you talk a little bit about that? So it certainly can't give any specific examples because I think what you're seeing is that a lot of enterprises or a lot of companies are kind of like, they're looking at that like, oh, this is our new secret sauce. It always used to be like they had some interesting feature before that a competitor would have to keep up with or catch up with, but I think they're looking at machine learning as a way to enhance that customer experience, right? So that it's a much more intimate experience. It feels much more tailored to whomever is using their product. And I think that you're seeing a lot of those types of things that people are starting to bake into their products. We've, again, this is one of these things where we've been using machine learning for almost 10 years inside Google things, like for Gmail, even in early days like spam filtering. I mean, something just kind of mundane like that or we even used it, turned it on in our data centers to drop our, because it does a really good job of lowering the PUE, which is the power efficiency in data centers. And those are very mundane things, but we have a lot of experience with that and we're exposing that through these products and we're starting to see people, customers gravitate to grab onto those to go. Instead of having to hard code something that is a one to many kind of thing, right? I may get it right or I may have to tweak it over time, but I'm still kind of generalizing what the use cases are that my customers want to see. Once they turn on machine learning inside their applications, it feels much more tailored to the customer's use cases. Machine learning as a service seems to be a big hot button that's coming out. How are you guys looking at the technical direction from the cloud? We're going there first, because you have three classes of enterprises. You have the early adopters, the power, front cutting edge, then you have the fast followers and you have kind of everybody else. The everybody else and fast followers, they know about Kubernetes, some might even, what is Kubernetes? So you have kind of the level of progress where people are. How are you guys looking at addressing those three areas? Because you could blow them away with, TensorFlow is a server. Whoa, I'm just trying to get my storage lungs moving to a cloud operation. So there's different parts of the journey. Is there a technical direction that addresses these? What are you guys doing? So typically we'll work with those customers to help them kind of chart the path for all those things and making it easy for them to use and consume. Machine learning is still, unless you are a stats major or you're a math major, a lot of the algorithms and understanding kind of linear algebra and things like that are still very complex topics. But then again, so is networking and BGP and things like OSPF back a few years ago. So technology always evolves and the thing that you can do is you can just help pull people along to continue on there by making it easy for them to use and to provide a lot of education. So we work with customers on all ends of the spectrum. If it's just like, how do I modernize my applications or how do I even just put them into the cloud? We have teams that can help do that or can educate on that. If there are customers who are like, I really want to go do something special with maybe refactoring my applications. I really want to get the cloud native experience. We help with that. And those customers that say, I really want to find out this machine learning thing. How can I actually make that an impactful portion of my company's portfolio? We can certainly help with that. And it's just a, there's no one, and typically you'll find in any large enterprise because there'll be some people on each one of those camps. Yeah, and they'll also want to, they'll put their toe in the water here and there. The question I have for you guys is you've got a lot of goodness going on. You're not trying to match Amazon speed for speed, feature for feature. You guys are picking your shots that is core to Google, that's clear. Is there a use case or set of building blocks that are highly adopted with you guys now? And as Google gets out there and gets some penetration in the enterprise, what's the use, what are the key things that you see with successes for you guys out of the gate? I mean, is there a basic building that Amazon's got EC2 and S3? What are you guys seeing as the kind of the core building blocks of Google Cloud from a product standpoint that's getting the most traction today? So I think we're seeing the same types of building blocks that the other cloud providers are. I think some of the differences is we look at security differently because of where, again, where we grew up, we do things like live migration of virtual machines if you're using virtual machines because we've had to do that internally. So I think there are some differences on just even some of the basic block and tackling type of things. But I do think that if you look at, just moving to the cloud in and of itself is not enough. I mean, that's just kind of, that's a stepping stone. We truly believe that artificial intelligence, the machine learning, cloud native style of applications, containers, things like service meshes, those things that kind of reduce the operational burdens and improve the rate of new feature introduction as well as the machine learning things. I think that that's what people tend to come to Google for. And we think that that's kind of the, a lot of what people are going to stay with us for. I overheard a quote I want to get your reaction to. I wrote it down. It says, I need to get away from VPNs and firewalls. I need user and application layer security with un-fishable access, otherwise I'm never safe. So this is kind of a user perspective or customer perspective. Obviously with cloud there's no perimeter. So you've got phishing problems, right? You know, speed phishing is one big problem. Security, you mentioned that. And then another quote I had was, Kubernetes is about running frameworks and it's about changing the way applications are going to be built over time. That's where I think SRE and Istio is very interesting in Kubeflow. This is a modern architecture for- We're doing a Kubevert out there where you can run a VM inside a container, which is actually what we do internally too. So that's, there's a lot of different ways to slice and dice. How relevant is that, those concepts? Because are you hearing that as well on the customers? Because that's pain point. But also the new modern software development is a future way to do things. So there's pain point, I need some aspirin for that, and then I need some growth with the new applications being built and hiring talent. Is that consistent with how you guys see it? So which one should I tackle? So you talked about VPNs first, that's my favorite one. So one of the most, can I give you the backstory? So one of the most interesting things when I came to Google, having come from other large enterprise vendors, before this was there's no VPNs. We don't even have it on our laptop. They have this thing called BeyondCorp, which is essentially now productized as the identity where proxy, which is it actually takes, we trust no one or nothing with anything. It's not the walled garden style of approach of firewall type VPN security. What we do is based upon the resource you're going to request access for, and are you on a trusted machine? So are you one that corporate has given you? And do you have two factor authentication that corporate, not only it's what you have and what you know, right? And so they take all of those things into awareness. Like is this the laptop that's registered to you, right? Is this, do you have your two factor authentication? Have you authenticated to it? That's a trusted platform. Boom, then I can gain access to the resources, but they will also look for things like, if all of a sudden, you were sitting here and I'm in San Francisco, but something from some country in Asia pops up with my credentials on it, they're going to slam the door shut going, there's no way that you can be in two places at one time. And so that's what the identity where proxy or beyond corp does, kind of in a nutshell. And so we use that everywhere, internally, externally. And so that's the way that we, one of the ways that we do security differently is without VPNs. And that's actually in front of, you know, a lot of the GCP technologies today that you can actually leverage that. So I would say, we take- Just rethinking security. Rethinking security, again, based upon a long history and not only that, but what we use internally from our corporate perspective. And now to get to the second question. Istio, Kubeflow is more of the way software gets run. I mean, with one quote from one of the ex-Googler who left Google and went out to another company, she goes, she was blown away, but this is the way people ship software. She was like, she was a fish out of water. She's like, oh my God, where's board? You know, like we do waterfall. So there's a new approach that opens source communities and people expect. Right. That's the notion of Kubeflow and orchestration. So that's kind of a modern, that requires training and commitment. Right. That's the upside, right? Fixed to aspirin, so identity proxy cool, future of software development architecture. Yeah, I think one of the strong things that you're going to see in software development is, I think the days of people running differently in development and then in sandbox and then in testing QA and then in prod are over, right? They want to basically have that same experience no matter where they are. They want to not have to do that, crossing your fingers. Remember, it used to get, now it gets red-edited or it got slash dotted way back in the past and things would collapse. Those days of people being able to put up with those types of issues are over. And so I think that you're going to continue to see the development and the style of microservices containers orchestrated by something that can do auto scaling and healing like Kubernetes, you're going to see them then start to use kind of that base layer to add new capabilities on top, which is where we see Qubeflow, which is like, hey, how can I go put, you know, scalable machine learning on top of containers and on top of Kubernetes? And you even see, like I said, you see people saying, well, I don't really want to run two different data planes and do kind of the inception model. Like, if I can lay down a base layer of Kubernetes and containers, then I can run bare metal workloads against the bare metal. If I need to run, launch a virtual machine, I'll just launch that inside the container. And that's what Qvert's doing. So we're seeing a lot of this very interesting stuff. Creativity. Creativity. Great. Talk about your role in the office of the CTO. I know we've got a couple of minutes left. I want to get it out there. What is the role of the CTO? Obviously, Brian, Steve, formerly a Red Hat executive. Yeah, Brian's our CTO. Used to run a big chunk of the engineering at for Google Cloud, absolutely. And so what is the office charter? You mentioned some CIOs, former CIOs are in there. Is it the think tank? Is it the command and control ivory tower? I mean, what's the role of the office? So I think a couple of years ago, Diane Green and Brian Stevens and other executives decided, if we want to really understand what the enterprise needs from us from a cloud perspective, we really need to have some people that have walked in those shoes. And they can't just be Diane or can't just be Brian, who also had a big breadth of experience there. But two people can't do that for every customer or for every product. And so they instituted this. The office CTO tapped Will Granis, again, had been in Boeing before, been in the military, and so tapped him to kind of build this thing. And they went and they looked for people that had experience. Former VP's of engineering, former CIOs. We have people from GE, oil and gas. We have people from Boeing. We have people from Pixar. Like you name it, across each of the different verticals, health care. We have those in the office CTO, and probably, I think, 25 to 30 of us now. I can't remember exact numbers. And really, what our day-to-day life is like is working significantly with the product managers and the engineering teams to help facilitate more and more enterprise-focused engineering into the products. And then working with enterprise customers, kind of the big enterprise customers that we want to see successful and helping drive their success as they consume Google Cloud. So being kind of that conduit directly into the engineering. So in market with customer, big-nut customers, getting requirements, helping facilitate product management function as well. And it's from an engineering perspective. So we actually sit in the engineering organization. Make sure you're making the good bets. Yes, exactly. Great, well, thanks for coming on theCUBE. Thanks for sharing the insight. Great to have you on. Great insight, again, Google. Always great technology. Great enterprise mojo going on right now. Of course, theCUBE will be at Google Next this July. So we'll have live coverage from Google Next here in San Francisco at that time. Thanks for coming on, Jonathan. Really appreciate it. Looking forward to more coverage. Stay with us for more day three as we start to wrap up our live coverage of Red Hat Summit 2018. We'll be back after this short break.