 Live from London, England, it's the Cube, covering .NEXT Conference Europe 2018, brought to you by Nutanix. Hi, and welcome back. I'm with you, Piskar, and I'm Stu Miniman, and welcome to the CTO segment at Nutanix NEXT 2018. Welcome back to the program. To my right is Benny Gill, who's the CTO of Cloud Services, and to his right is Rajiv Muryani, not Biryani. You know, the CTO of Cloud Platforms. Gentlemen, thanks so much for joining us again. Thanks, Stu, for having us. Great being back. All right, Rajiv and Benny, Nutanix has been kind of busy since last time we've chatted. AOS got really a file system rewrite. There's been some M&A integration going on, as well as organic activity. So, you know, I love talking to the CTOs, just if you can bring us inside a little bit, you know, what's been happening, what your team's been working on, some of the hard challenges, I mean, things like AHV, nested hypervisor on top of GCP. These are some hard challenges, getting ready for NVMe over fabric. You know, some real massive things that happen underneath the cover, as well as some new products. So, Benny, want to start with you, as to, you know, what's keeping your team busy? Oh, the teams have been quite busy, especially, you know, once you have, you know, more than 10,000 customers, and a product that's earning a lot of revenue coming in, and at the same time, you have to change the guts of it, preparing for the next generation. So, it's a lot of work. I mean, if you're starting from scratch, it's much easier. But, you know, we've had a lot of experience bringing in new capabilities, making it transparent to the customer. One click upgrade is really important for us. So, learning from the past, we've been able to rewrite the engine for the storage, in a way that customers wouldn't notice, but it's going to run just faster. You know, put us to the team that they've pulled it off. And it goes across the board, when we are acquiring new companies that come into the fold of the Nutanix family, the whole idea is to make it look seamless to the customer, because that's one thing that, you know, customers know us for, like, hey, will it have Nutanix simplicity? So, a lot of learnings, we have created some thumb rules to guide people coming in, and those are working fine for us. Yeah, and there's, you know, a method to the madness over here. There is, in the end, one vision, that we want to provide a true hybrid cloud experience to our users. And to do that, we feel we're the first start by building the best private cloud. You can't have hybrid without private. And to do that, we need to have an infrastructure that actually works for our private cloud. So, we start with HCI as an initial platform. We build on top of that with private cloud features. And not just, you know, networking compute and storage, like in the past, but more platform services, like Aira and Carbon and so on. And then once we have that, we can then layer on the new hybrid cloud services. So, even though it looks like we're doing a lot of things, it's all guided by the one vision. So, tell me, you know, that hybrid cloud vision, you know, where does it lead us? Does it lead us to, you know, the public cloud in the end? Does it lead us to a Nutanix cloud? Where does it help customers go towards? Well, the way I look at it is that it doesn't lead to any one place. It leads to multiple clouds. There'll be private clouds, there'll be edge clouds, distributed clouds, big central public clouds. The important thing is, can you move applications and data between clouds? An analogy I use is, you know, 20 years ago, if you were writing applications for Solaris, you were pretty much locked into sun. If you were writing applications for HPUX, you were pretty much locked into HP. Once Linux came along and made it possible to write applications for any x86 server, you had got independence from underlying hardware. And the same thing will happen with cloud. Today, you have to write applications for Amazon, for GCP, for Azure. Who can build an operating system that actually commoditizes all that? That makes it possible for you to run on any cloud with the same set of applications. So that kind of sounds to me like you're, you know, doing vMotion and HA and DRS, but then, you know, for a new generation of technologies. Well, not vMotion, across clouds is, of course, the goal. It is the goal. But it's not just enough to move the applications around. You have to move data around, you have to move the management plan as to be the same. So there's a lot more to it than just simply copying bytes across. Have any of you want to add to it? Yeah, I mean basically, adding to what Rajiv said, if you ask where will hybrid cloud lead, I think it leads to a dispersed cloud. And some of it was also mentioned by Deeraj in the keynote, which is, you know, this big monolithic cloud concept has to atomize into much smaller pieces and distribute. And that's what's going to happen. Well, you start with solving it with hybrid and at least solve it with two. And from two, you go to many. And that's what's really exciting. Yeah, it's a really good point, Vinny. I want you to help expand on that a little. I think back to companies that build portfolios and you'd look at it and say, okay, well, I buy product A, B and C and boy, I don't know how to use those together because they have different interfaces and how do I work them together? Today, you know, I think micro services architecture, I think about APIs pulling everything together. What are those guiding principles that you give internally to teams to make sure that I can use the pieces that I want, they work all together, they work with, you know, this really broad ecosystem we have and all the multi-cloud environments. So, you know, as much effort we put in building architecture for the product design, I mean, we have to put the same amount in terms of how is it going to be consumed by the customer. And just having a long portfolio is no longer what customers are looking for, they're looking for simplicity. So to your point, one of the things we are really careful about is, especially when we're requiring technology inorganically, is how do you make sure identity and billing is the same, right, that's the most important thing. So you don't have to log in once in this product, once in that product, basic stuff, but if you get it, you know, right, it's delightful. The other thing is about experience, developer experience and user experience, these are the two other out of the four factors. User experience is around, like, do I have to learn this again? Like, if you look at companies like Apple, I mean, if I use the Mac, they try to make it very similar, such that even a two-year-old can figure out how to use it, and we would like to say that if you've been in IT industry for two years, you should be able to use any Nutanix product. Developer experience is around APIs. We have a standard, we have Geared, version three intentful APIs, and that is creating a standardization across. You saw a little bit of it in the demo today, where I went through COM, and E-POC, and Flow, and Prism Pro, all from one pane of glass. It didn't look like four different products. In fact, if I not mention they were four different products, it probably wouldn't have been obvious that they were. And that's important to us, keeping that experience seamless is very important. And that comes at a cost. I mean, we could have released it as soon as we acquired some of these things and punted it on to the customer to figure out how these pieces come together. But we know our customers have a higher expectation from us, so we take the time. Yeah, from that perspective, as a user, I'm used to working with different types of clouds, public, private, hybrid, anything in between, and the amount of interfaces I have to touch to get something working, to get a series of products to align, to do what I want it to do. That's becoming such a difficult task that having a single interface or having a familiar interface would actually help in that. So maybe you can talk a little about, use that UI to go into the public cloud or into the hybrid cloud as well to make that experience easier as well. I'll talk about a couple of things. One, whenever there's a proliferation of technologies and you're trying to glue it together, I mean, single pane of glass is one thing that people talk about. I think that's not the most important thing. I mean, obviously it's a requirement. It's a necessary condition, not a sufficient one. To make it sufficient, you also have to bring an opinion into the design, and the opinion is where we are taking some decisions for the customer, where the customer wouldn't care about learning about those things, and that's where Nutanix will come in and through our best practices, we put our opinion in the design of the product so that the number of decision points for the customer is minimized, and that's how you basically start consuming this diversity out there. At the end of the day, for the business, the only two things matter, the business logic and business data. Infrastructure is sitting in the middle, it's just like a necessary evil, so if we can hide it and make it seamless, customers are really happy about it. Yeah, I'm wondering, can you talk about that the feedback loop you have with customers? Things are changing very fast. It's hard for anybody to keep up. This week, even Nutanix has a lot of announcements and I'm sure will take people all the time to there. How do you get the feedback loop to customers to make sure they're getting what they need to understand your products and your understanding where they are in their journey and mature the product line? Yeah, I mean, we have a whole bunch of channels. We just had a customer advisory board yesterday. We invite customers and have a really deep, intimate conversation and frank conversation. What's working for you, what's not working. We have an engineering team on Slack channels and WhatsApp channels with our customers, especially the customers who are really, they complain about a product and they have opinions. So we just tried to short-circuit this thing and it's all about empathy. So getting our team know the customers better. Just absolutely, Rajiv, I definitely want your opinion. But just feedback, actually, I talked to a few customers and they said, I don't know how Nutanix does it, but for a company their size, I feel like I get personal attention and touch point. So congratulations, it's good to hear how that happens. Some of the stuff you saw today is a direct result of the feedback. The grouping of products into core, essentials and enterprise kind of also reflects the customer journey. A lot of customers start with us with the core. Once they get used to that, get the essentials products, build a true private cloud. And only then they start looking at multi-cloud. So right products for the right customer, something that we are taking very, very seriously at this point. So I want to dive into that, you know, right product, right customer. So one of the announcements you made is carbon at Kubernetes as a managed platform. So what customers do you service with that product? How do you go into customers like that and how do you help them? Yeah, go ahead. See, Kubernetes is one of the most fastest growing technologies in the IT space that we have seen in the recent years. And a lot of our customers, I would say, especially this year, we have seen they have developers using containers, right? And they're at a point where they're trying to decide how can I put it in production? Now production has many requirements there. Carbon is being used by our customers who are trying to see how they'll put containers into production. And what we are doing with carbon is we're providing native Kubernetes APIs as is there in open source, but we're solving the hard problems of upgrades, scale out, high availability, troubleshooting. These mundane things that usually people don't want to do and that's where we come in and help. So I've seen customers use our storage volumes for even databases, containerize to stateless things. It's all across the board, but still early years, I mean, for this kind of ecosystem. But it's headed into, it's going to be the future. One of the things I found really interesting to watch is over the last two decades, we've talked about intelligence and automation and infrastructure, but really things are happening fast now. When you talk about, whether AI or ML, there's really things that are creating some intelligence that it's not like, oh, I created some script and it does some thing, but it's working well. I know there's a number of places that that fits into your portfolio. Prism X play, it seemed to get some good resonance and cheers from the audience because maybe they've all played with the if-t-t-t, so start from there and how do you think about the AI and ML space? Yeah, so we look at computing evolving from manual, mostly manual in the past to more automated, but really you want to get to this autonomous computing that Sunil talked about. So think of it as, you know, cars used to be really difficult to drive in the past. It used to require knowing how the carburetors work and cleaning them out once in a while to the point where maybe 15 years ago, you pretty much didn't know anything about the internals of a car, but you could drive it, it was reliable, it would work, which is probably where we are today in IT, but the real goal is to get to the autonomous computing, the self-driving cars, the Tesla and Google and others are trying to develop, where you don't even have to be paying attention and the car will just drive itself, right? Yeah. IF-T-T-T and the X-Play stuff that we have is a step in that direction, it's obviously very early, but it's the beginning of a journey where you can then start taking feedback loops, learning what works, modeling that out, and extending capabilities on your own, and that is something we'll be looking at over the next few years. And you know, it's something where, I don't think it's, it's not cute and that's why it needs to be done. It's actually required. And if you look at Moore's Law, it applies to machines. So every year you'll have double the number of course, and you know, the same dollar can buy more. Look at humans, that's not true. I mean, every year they're only getting more expensive. In fact, a lot of our customers here say talent is scarce. So just by that definition you'll see machines are growing and the people who manage the machines are shrinking or static. So you have to put in a layer of machine which is smart. In between of the human and the large pharma machines and that if you don't do it, there is no data center. So it's inevitable and you'll see this happen more and more. So that kind of sounds like you're positioning your portfolio in a way that you enable the IT ops people to not care about infrastructure as much anymore, but help their employer, their customer do other stuff. So how does your portfolio relate to the freeing up of time for those employees, for those IT ops people? Some of it just goes back to the core design principle. I would go to the basic, you know, how do we start as a company? We were looking at storage and they were dual controller. A and B, A dies, B is running, but guess what? I'm worried that B will also die, it's the same age. So I have to run to fix A. Run to fix A is my weekend and the night wasted. If I had N, one dies, fine. I mean it's a capacity problem. So that goes to the core, like how do we design things that are scale out and web scale we talked about? So everything that we do, including now Prism Central is scale out, I don't have to rush to go fix things. Hardware will always fail, right? And that's, you know, it permeates in the entire organization in terms of how we design things. And then on top of that you can add automation and machine intelligence and all that, but fundamentally it goes to engineering. When you talk about, we talked about earlier in the discussion kind of the rewrite that went on for emerging applications and emerging technologies, I guess, what's exciting you these days, you know? The industry is a whole containers, you know, we looked at, you know, flash technology, containerization, you know, I looked at Nutanix when it first came out as was, you know, some of these waves coming together, hyperscale and software defined and flash, all kind of were the perfect storm for the original generation. What are those next waves coming together that you think will, you know, have a massive impact on the industry? Yeah, a lot of innovation going on on every layer of the stack. I mean, we started with the hardware, it's been coming for a while, but it's almost here now, the whole concept of having persistent memory. Essentially dim spots having memory that can persist across reboots and being byte addressable. So this is a big difference for the storage market, right? We've always had block addressable storage when it becomes byte addressable. Paradigms of computing will change, and warthems will change, how we write programs will change. So there's a whole big wave coming and getting prepared for that was very important for us. Yeah, and if I can drill into that a little bit, because, you know, what I thought about, you know, before it was I had, you know, my pull of storage and my pull of compute and I had my networking and I thought, well, you know, what your solution is, I just have a pull of infrastructure, but I need specific data in specific places and latency is really important. You know, Amazon just announced, you know, a new compute instance with 100 gigabit networking for, you know, the same type of application we're talking about, HANA and persistent memory and the like. So do we not think of it as a pull anymore? It's, I hear, you know, metadata and data are going to get more localized. So how should we think of your infrastructure going forward? You should think of it as a pull. We should worry about making it all work well, and that is essentially our job. If we can succeed at that, then you would never have to think about it as, well, this particular, you know, storage is allocated to this particular application at this current time. It's up to us to make that happen as the applications are running. From your point of view, should we pull, go ahead. You know, absolutely. Another thing that's happening in IT or in the space of compute is the upper limit of this pull is being hidden, right? So for example, in the old days, there was disk, then there was a virtual disk, but it had a capacity and you would format it. When you look at S3, it doesn't have a capacity, you don't format it. That's what's, and that's more to application design. When you don't think about the capacity of the pull that you're using, that's the direction where we need to go and hide all this, right? You know, so just in time purchase of the next hardware that you need to get, but the developer does not see the upper limit. Well, Rajiv and Binny, thank you so much for sharing all that. Thanks, congrats on all the progress and look forward to what we're going to bring on down the road. Thanks, thanks, thank you. For you, Piskar, I'm Stu Miniman. Lots more coverage here. And Nutanix.next, London, 2018. Thanks for watching.