 Live from Copenhagen, Denmark, it's theCUBE. Covering Nutanix.NEXT 2019, brought to you by Nutanix. Welcome back everyone to theCUBE's live coverage of Nutanix.NEXT here in Copenhagen, Denmark. I'm your host, Rebecca Knight, co-hosting alongside of Stu Miniman, of course. We're joined by Rajiv Murrani. He is the CTO Cloud Platforms at Nutanix. Thank you so much for coming on theCUBE, Rajiv. Glad to be here as usual. So you were up on the main stage, talking about your guiding force, your mission, to make enterprise computing so reliable, so ubiquitous that it's invisible. You don't think about it. You don't even, yeah, just tell, we thought that a little. No, I mean, look, if you look at any successful technology, the consumer of the technology doesn't need to be aware of the details around it, right? I mean, take cars, when cars were first introduced, you probably had to understand what a carburetor was, how it all worked, remember to fill water in the radiator. But most people don't know how to do that today. I wouldn't know how to do that today, even though my youth actually had to do some of those things. And it's gonna be the point where the cars are self-driving. Now, the same thing should happen to computing, and we spend way too much time today just doing things that keep the lights on, right? Just managing infrastructure, patching things, upgrading them. It really should be something that's not what IT is doing. IT should be focusing on the business, delivering applications that their customers need, and not so much on managing infrastructure. Rajiv, that one-click simplicity is so important, yet we know things are becoming even more complicated. Bring us inside a little bit. You've got 14,000 customers, and there's been a major, basically rewrite of AOS to be able to be ready for all of the new apps, the cloud-native, NVMe, Optane, you name it, all of these things. How do you balance getting all of this new stuff up with making sure that you keep that simplicity and don't break things for all of your customers while the jet engine's up in the air? It takes constant effort. It takes a conscious effort to make sure that things aren't drifting away from our goal of being simple, and to be frank, at times it has, and we periodically do audits of all our workflows, make sure they're as simple as we think they are, and haven't drifted over time, and occasionally we do find some real clunkers and have to go back and fix those things. What makes it different is that we start with a fairly opinionated view on how things should be done. The idea is to make it simple for 99% of the people while still offering all the advanced options that one person power users would need them, and making sure that we understand what 90% and 99% of people need, and focusing on that is very important. So a lot of customer focus studies, a lot of design reviews, but also this constant going back and you take a workflow like VM creation, like provisioning a VM. When we started it was pretty simple, right? Then as we started adding more and more options, is this thing going to use PCI pass through? Is it going to use this option, that option? Suddenly realize that now there are 30 things that people are filling into provision of VM, most of which nobody cares about. So you go back, rewrite, and keep doing that again and again and again. So Rajiv, it's one thing when you talk about living on different servers, whether it be Supermicro underneath, or Dell EMC, Lenovo, HPE, it's got to be a little bit different if you're talking about where you're going with HPE GreenLake, with the X in AWS that you talk about when I talk to people, it's like, oh, I'm trying to use Terraform and boy, I have to write it one way to work with Amazon, I have to write it another way if it's Azure or GCP. So will Nutanix be able to keep that simplicity and bring homogeneity to these dispersed environments? Absolutely, so the point is what's your layer of abstraction, right? The way we are going to public cloud, the way we go to every server vendor out there, the way we're doing things in XI, all of it starts with the hypervisor and the storage stack and networking with the three layers of abstraction. And if you have the same three components everywhere, everything that we build on top of that remains exactly the same, for example, it looks the same, our APS look the same, everything looks the same. So higher level constructs like Calm and so on don't have to be aware of what the actual substrate is. So we're creating a Calm Blueprint whether it works on XI on AWS or whether it runs on ESX or whether it runs on Nutanix AHE, the blueprint remains exactly the same. Now, if you want to consume more services, if you want to consume services on Amazon which are not available on premises, you want to use things like auto-scaling groups in EC2. Sure, you could then create blueprints that are customized for a particular substrate. So in terms of this, you said you start with a very opinionated place of where the customer is. Well, first of all, it's based on customer feedback and as you said, customer surveys. And so where are you right now in terms of where the customer is? Are you meeting the customer now or is the customer ahead or where would you describe the customer mindset? I think it depends on what you are looking at. If you're looking at the core products, absolutely the customers are with us. They're ready to consume. They actually drive a lot of the innovation that we're doing by feeding back changes that we could be doing to make things even simpler. On the private cloud side, that's getting there. I think we get a lot of feedback on files, on Prism Pro, on Calm, on Flow, because those have started getting a lot of adoption in the market and we do get a lot of feedback on them. On some of our newer products which have started evolving more recently, it's a more collaborative process. There we are actually working with customers directly, understanding their problems and moving our roadmap forward based on that. So while it's early in terms of adoption in production, the whole process is very collaborative in those situations. So we really are very close to the customers there. What are some of the biggest customer problems right now that they're facing? What's keeping them up at night? And therefore keeping you up at night. Security is always a big one. Complexity, people, skills, all those things are big problems. In fact, one of the biggest things right now is just training enough people to handle all the complexity in a dataset. So the more you can remove from that, the more they don't have to focus on that. The easier you make their lives. The other thing is just the amount of time they spend on routine activities which require disruption to services, right? This is something we've talked about before. Why does it require a legacy three tier system to have maintenance windows and downtime to do an upgrade? Google never has downtime. Google is never down for maintenance. Doesn't mean they aren't upgrading, they're upgrading all the time. So how do we bring that same kind of capabilities on premises that's been a focus for a long time? So Rajiv, when I'm talking to users out there, when they talk about all of the items that are out there that they need to deal with and the routine tasks, automation is something that keeps coming up. So tell us where automation fits into some of the new things that you were talking about today and this week with your customers. Yeah, automation, for us, we look at automation in three steps. Anything that's automatic better be safe. First of all, safety is paramount. It starts with security. It has to be simple and we talked about how calm provides for that. And then you can start adding in this new wave of technologies around artificial intelligence and machine learning. And really it's not so much about automation right now, it's all about autonomy. It's autonomous operations, not just scripts that do a task. And that's an area we invested in very heavily early on with our Prism Pro product. We built our own patented algorithms for machine learning, applied them to operational metrics like capacity planning and what if modeling and dynamic alerts. And what we've been doing with that is extending that more into the application layer so that not only can you apply these algorithms to CPU and memory, you can actually get insights into, hey, the latency on this particular application looks somewhat unusual or the amount of cash available on a SQL server is unusually low. And act on those. And the other part is acting on alerts, right? So if something happens, does a human being need to get involved to solve for that? And if it does, then well, it's not really automatic, right? So that's the other part that we introduced which is our cross-play product which lets you define these action chains that are automatically triggered when an event or an alert takes place. They can go ahead and fix the problem but also simple things like send you a Slack notification or an email, log the event, maybe create a snapshot of your VM so that you can go back and debug problems later. All that sort of thing made really, really simple. Yes, it goes back to the simplicity and the invisibility to that aspect. Yes, yes. Autonomous data centers by definition have to be invisible, right? If you're to get involved in managing an autonomous data center, then well, it's autonomous, right? What's the point? Exactly. So the whole idea is that human involvement in day-to-day operations becomes so low that everybody's focused on applications and light up business use cases. Rajiv, when it comes to those applications, you talked about some of the new enhancements like NVMe and Optane, where are your customers today? Are there any interesting applications that you're seeing them deploying today? A pattern I've talked about for the last couple of years of the Nutanix show is modernize the platform and then modernize the apps on top of it. Things like containerization. I'm sure it's a bell curve and a journey for all the customers, but what are some of the patterns that are starting to emerge and where are they finding success? Yeah, there's a whole wave of new applications around data pipelines with Kafka, Spark, things like that, that Apache stack effectively, which are putting more of a load on storage in particular, so that's one area we see customers looking for more performance. But even some of the traditional applications like SAP HANA and Epic and Meditech Expans, they also have patterns which can benefit greatly with some of the advancements that we have been making. It gets a little technical, it's the size of the working set, whether it all fits in on SSD versus fitting on magnetic drives, but some of the work we have been doing is removing the overheads. If you do have a miss and you go to slower media, you still get good performance and that's really how we're getting good to the new products. Well, yeah, maybe without getting, we don't need to go drill down into the core of the Intel chip and everything, but Nutanix doesn't just take off the shelf stuff and put a box together, it's software and there's work that happens with your partners in the ecosystem. Give us a little flavor as to where you're making the investments and where some of those partnerships and integrations are key. Yeah, yeah. So, on the platform side, a lot of the investments happen in validation of the platforms, making sure that we are ahead of the curve in adopting technologies, but also feeding back from our side things we would like to see in the platform, right? So, how do you adapt things like RDMA to handle not just the traditional workloads but a hyper-converged workload? How do you essentially look for things in this new class of memories that would benefit from data locality for us? So, that's one class of partnerships that we have with the hardware vendors, with HPE, with Intel, with IBM, a whole bunch of people, but then we're also looking at partnerships up the stack these days with companies like ServiceNow, with Veeam for backup for our mind product. I think you saw a little bit of that today. It's a whole bunch of things happening across all the areas. One of the things that it really comes across at this conference is just how strong Nutanix's culture is, the company culture, the humble, honest, hungry, and another word that's creeping in now is resilient. I'm wondering if you could talk a little bit about your division of the company and describe how the company's resilience, the employees and the company itself has really displayed itself. So, as with any company, anytime you go through the kind of growth we have, the forward momentum that everybody sees, there's also a lot of setbacks that people don't really see, and there have been a whole bunch of us of these in our history. There have been areas where literally the product has had a flaw which is so fundamental, we call it internally, a near-death experience. What's been really great, again, from an engineering point of view is how the teams come together at that point. We set up these war rooms, we've been working weekends, everybody's there, everybody's on it, and nobody talks about, hey, look, where's my work-life balance or things like that at that point. Especially when there's a customer involved. If there's a problem that's causing customers' outages, our engineers will give up everything. They'll give up everything and not just their work but their lives to make sure that gets fixed. And that has helped us get past these setbacks, get back in stride. Happens less now, it used to happen a lot earlier, but to build this real culture of resilience from the big make and the early engineers. All right, Rajiv, what's exciting you going forward? You don't have to touch on this one, but when I saw at the end the Zyclusters and that hibernate feature was something that was like, oh, yeah, I don't think I've seen that as to how I can make sure I save my data, I'll be able to shut things down, maybe start there, but give us a little bit, look forward as to where the team's playing. No, that's the kind of thinking that, the detailed thinking that you have to do when you want to launch a new product, right? Okay, look, the whole point of doing Zyclusters on Amazon, one of the biggest use cases is cloud bursting. Well, cloud bursting is not just about increasing your workload size, scaling it up, at some point you want to scale it down. How do you do that for stateful workloads? For stateless, it's easy. All I'm doing is putting web servers over there, throw them away, they're gone, but I have a database that I scaled out over there, well, that data can't go away. So we had to find ways to essentially solve for those problems, so that's how the Hibernate feature came around. In general, the bigger question that you asked about, you know, what I'm most excited about, I think this whole convergence of private and public cloud with the same stack on both sides, is a new thing, it really hasn't existed before. The fact that applications can now move back and forth seamlessly between public cloud and private cloud, without rewriting, without refactoring, without a big lift and shift is very, very interesting, but by itself it's not enough. The flip that's missing is what about data movement? What about if you have your data in Amazon and now want to move it on premises? That, there really are no good solutions to them. Amazon doesn't give you APIs and the tools to do that. So I think data movements want to be a big thing, and then building a common services stack on both sides, because it's not just networking storage and compute anymore, right? If you're on Amazon, you're probably using all kinds of networking services, security groups, around 53, how do you take those kind of services and also make them available on premises? So Rajiv, is there anything you've learned as a team when you look at AWS Outposts or Oracle Cloud at Customer? We've had a few years of people talking about, but not a lot of deployment yet, so not saying you're late to the market, but what have you been able to sit back and learn from what's been done so far? So the reason we are a little bit late to the market is that we, I think, are solving a problem which the other vendors are not, which is kind of what I alluded to before, that is that how do you support both the legacy applications and the cloud applications? How do you provide for migration both ways? Applications that were born in the cloud, and now you want to move them on premises, or applications that were born on premises and you want to move them to the cloud? Outposts is great, I think it's a good product. Technology is that AWS thinks that hybrid is the right strategy, but it's also one problem. It's also the problem of cloud applications running on premises. It has not solved for the problem of legacy applications running in cloud, right? That is just as difficult as ever. That's not become any easier with Outposts. Similarly, if you look at what VMware is doing with AWS, it's also legacy applications going to AWS, but in doing so, they don't have access to all the different networking services that AWS offers because you're not running NSX and you're kind of running a different networking stack than Amazon is. So we thought long and hard about this problem and we said, hey, look, we are not going to take the easy answer over here. We're going to take our stack which is known to run both legacy and cloud native applications. We've proven that and integrated natively into Amazon so that you can use the Amazon services, you can use our services, you can run legacy applications, you can run modern applications without giving up anything. And that I think is why it's taking a little longer but I think it's a more powerful solution for the long term. Thank you so much for coming on theCUBE. It's always a pleasure having you on. Hope you'll come back and see us again. I'm Rebecca Knight for Stu Miniman. Stay more of theCUBE's live coverage of .NEXT, Nutanix here in Copenhagen. Thank you.