 from Washington D.C., it's theCUBE covering .next conference brought to you by Nutanix. Welcome back to D.C., everybody. This is theCUBE, the leader in live tech coverage. This is our special coverage and presentation of Nutanix NextConf 2017. This is the third U.S. conference that theCUBE has done of Nutanix .next. Nicholas Radford is here as the Senior Vice President of Engineering and the CTO at Houston Mechatronics. Wait until you hear what these guys do. It's Satyam Vagani, who is the VP of Technology at Nutanix, gentlemen. Welcome to theCUBE. Hi, Nick. Let's get right into it. We're talking IoT, we're talking Edge. You guys do some pretty interesting stuff. Tell us about the company. Well, Robotics is a service company, primarily, but we do some intelligent automation and intelligent drilling in the IoT space. So it's pretty exciting and dynamic area, actually, and just imagine taking a bunch of different systems that haven't typically talked together before, and we've kind of glued them all together, and one of the big oil field services companies was attracted to our sort of thinking in this area and have given us some work and we're taking and running with it. Can you just explain what Robotics is a service means? Well, usually, you can kind of break this down in a couple different ways. There's a lot of people out there that sell robots and it's kind of a thin business, right? You might sell a robot and then the people you sell it to use it and make a bunch of money off of it and then you're starting, you're stuck trying to find a new customer to sell a robot to, but if you consume the robots, so to speak, that you build and then use them as a service, it's a much more lucrative position to have, and so we do technology systems development for partners, and then we also operate that robot in the field for them, so it's a good residual pull through revenue stream for us. So Satyam, the industrial technology world, the information technology world is coming crashing together. Absolutely. IT and OT, Nutanix has been talking about the edge, you know more and more, I mean, if you're not talking about the edge, not in the game, but give us an update on your strategy with regard to the edge and really you're thinking about companies like Nix. And a great question. I guess I have so much to tell. Yeah. Make it edgey. Make it edgey. But yeah, I guess we kind of sort of naturally fell into it because we saw that the future of computing might be more edgey, if you will, than we think is, you know, right now we are spending a lot of time and energy worrying about clouds, you know, private cloud, public cloud, how to consolidate them. But then at the same time, we are seeing that there's so many of these sensors being deployed in the world just this year. It's going to be roughly eight and a half billion sensors. If you count consumer and industrial reality together by 2020, it's going to be 20 billion sensors. And so all the data that these sensors are going to generate is going to be processed in real time, closer to where the sensors create the data as opposed to slightly farther away, which is in the cloud. Of course, the cloud remains relevant. The cloud is going to do much more longer-term processing and the edge is going to do real-time processing on the data. And so in that sense, we saw that as a natural step two of the hyper-convergence journey is if you think about step one of hyper-convergence as the convergence of compute storage and network resources inside a data center. Step two is the convergence of the edge and the cloud into one fabric, one OS, if you will. I wonder if you could help us unpack that a little because we saw kind of public cloud pulled at the data center for years and now the edge is pulling out the cloud. So the edge is different from the data center. So most people think of Nutanix. You know, I'm either living in a data center or maybe some service provider. So, you know, a different form factor. I know there's some announcements Nutanix made to kind of get to robo cases. Is that the same for the edge? Yeah, different form factors because, you know, some of this hardware needs to be ruggedized. It's on oil rigs or drones or military vehicles and so on. But also a slightly different and evolving storage stack because now the problem of deploying applications at the edge is about developers having to write code and not having to worry about how the code runs on the edge because as soon as they have to worry about that, developers become operators, infrastructure operators. And so this one will also have a slightly higher level of applications stack, you know, machine learning services or analytics services at the edge so that applications can directly consume those high level services as opposed to the lower level, you know. Which actually that's really intriguing because as part of our robotics as a service side of our business, we have a pipe inspection system that we're going to be deploying in quantity. And so what you, that's a type of edge device, right? That's a, I mean, robots are really nothing more than fancy data collection systems, right? And so we put them out into the world to collect the data but then what do you do with that data? How is it stored? What sort of post analytics are you doing on that data to then feed forward back to the intelligence at the edge so that they can make decisions better, right? So when you have our robot, we would call Pearl, a pipe inspection robot, you'll actually see a demo of it later, fingers crossed. As it's traveling through the pipe, it's collecting all this data, right? So, but all of the runs prior to that, it's afforded all of that knowledge on the decisions it's making right then and there, right? Because we've done all this back learning, if you will, on what deformities look like, which increases the quality of our inspections. And so then if you start looking at a ubiquitous deployment of these types of assets, where you might have 10, 20, 30 in the field, that's a massive amount of data that you're collecting right there, right where the sensors being taken. The processing of that data is determining whether the robot stops and maybe observes a little bit more. But then it's all being shipped back at a later date to the cloud for further analytics, then to feed forward in the next operation to perform that better. So it's this feedback process of learning between the application that's actually happening in real time and the later on analytics that will occur. So let's stay on the data for a moment. It is all about the data. Is it correct that much of the data in your world is analog data that you're able to now convert into digital or are you already there? At the end of the day, you're trying to take an analog measurement of some type, right? We live in an analog world and if I'm trying to measure the thickness of a pipe, I'm using a transducer that by nature is typically an analog device. I can then digitize that information and that's how I send it over in communication streams and whatnot and of course it's stored digitally. But at the end of the day, we're taking analog information, doing data processing on that, looking at what it means in the activity that we're trying to do, measuring the thickness of a tube and then we shift the data back at a more convenient time when we have more bandwidth back to the cloud for all of the deep learning, deep type of analytics. Nicholas, could you kind of explain that, kind of your stack? I was hoping you were going to explain it to me. Because how did you get to Nutanix? What goes into what public cloud, what services are you using there? Whatever you could share, if you kind of could understand. We're involved with Nutanix on our rig automation side. And so we use their storage, we use their storage in the way that they've created an excellent way of doing that. And so that's primarily how we interface with them and one of our big oil and gas partners is a huge client of theirs. And so that's our primary relationship with them. In fact, I sent Rich a picture of a Nutanix box that we just recently installed in our server room and I was like, giving him the thumbs up and I was like, hey buddy, you know. All right, in public cloud, do you have a specific one you're using, using mini-clouds? No, no, no, no. I mean, for the processing of data, you've seen some of it goes to the public cloud though. Yeah, no, I mean, it's more of the local area. I mean, this is the stuff we're using in turn. I mean, this is the security requirements that we have is. So your cloud is an on-prem cloud. Yeah, exactly. Okay, how much of the data? I mean, I know it can't be precise, but if you think about all this real-time decision-making that you're doing, how much of the data is actually going to go back to the cloud? I mean, these are even rough percentage terms. Well, we'd like to send it all back, right? 100%. It's just what you don't send then and there. Right, there might be a little stream of it coming back off of, let's say, our pipe inspection robot. But at some point though, you want to take that, download everything, store it back up. I mean, it's in all the big data analytic techniques and analyze it. I mean, you know. So you expect you want to persist the majority of the data, and you ultimately will not do that at the edge. You'll ultimately have to get it back up to the cloud. Yeah, that's the way we see it. You're going to use the Chevy truck. You have a different opinion. Use the Chevy truck access method to get it there. Go ahead, please. As I have a different opinion, kind of sort of a similar principle about a different opinion, which is, you know, in terms of volume, a very small fraction of the data is going to make it to the cloud. Now, in terms of intelligence, you know, almost 100% of the intelligence is going to make it. But it is how the edge participates in reducing the volume of the data. You know, just again, to give you numbers, you know, in the year 2020, it's projected, and this is, I think, the Cisco Global Cloud Index. They project that roughly 600 zettabytes of data is going to be created on the edge. And the public and private cloud combined in that year is going to roughly witness 15 zettabytes of data. And so the question is, where did the rest of it go? And I think my answer is, if you look at, for example, a smart billing, or a robot inspection kind of scenario, the robots taking pictures or video streams, which is ridiculously rich data, and changing it into a time series database of whether some anomaly was detected or not. You know, look at a smart airport example, we're going to take a lot of surveillance data and change it into whether a person of interest was detected or not. Or did you see a white van that you're looking for? And so really the information, the volume of information goes down, but the refinement goes up. Is, you know, the cloud really is interested to know, because presumably in the smart airport example, you have somebody sitting at a dashboard monitoring all California airports looking for a person of interest. And all they are worried about is whether somebody showed up or not. And so it's the metadata that shows up, as opposed to the raw data. So the needles go back? Needles go back, exactly. That's a good way to put it, not the A-stack. Nicholas, one of the things we look at IoT is it's really created a much larger, I mean, orders magnitude more surface area for security attacks. Is that something that concerns you, your clients, you know, how security fit in? It concerns our clients, very much so, absolutely. In fact, one of the first questions out of everybody's mouth is how are you going to handle security? So it's paramount and very important. Absolutely. All right, Satya, how are you going to solve that? Well, the running joke is blockchain. And, you know, people stop asking questions as soon as you say blockchain. But no, it's an unbelievable problem. In fact, something that probably we haven't, you know, kind of solved in generations. We are struggling with cloud security, with cyber security. And now we are talking about a number of devices that's going to be three hours of magnitude more than the number of servers that run in the cloud today. What about one of the things we haven't talked about is connectivity. How do you connect the edge? Is it just all wireless? Yeah, I mean, the ubiquity of the wireless networking systems are very high right now. I mean, it's all... How's the quality? You know, good. It's like wireless. It's like wireless. But is it a headwind? No, it's actually, you know, one of the issues that we're having with, honestly, our pipe inspection demonstration today is just being flooded. There's 4,000 people in the main hall, right? And so there's all this wireless activity. And sometimes, you know, our pipe inspection robot doesn't know who it should be quite listening to. And I mean, you know, you go to a concert and you look like you might head to a Metallica concert here and there. I do. And you know, sometimes you can't even send a text because there's just so many people and trying to connect and it's a big deal. So it's a challenge for you? Absolutely, it's a challenge. Excellent. I've seen some vendors, they are deploying special networks, right? They are deploying low bandwidth networks. Verizon's doing it, I think. The NTA is doing it. No pineapples. No pineapples, hopefully. That's like the most recent Silicon Valley episode, right? All right, gentlemen, listen, thanks very much. We really appreciate you coming on theCUBE. Thanks for having us. Great story and use cases. It's always a pleasure. Good luck with the demo. All right, keep right there, buddy, we'll be back with our next guest. TheCube are live from Nutanix. Next conference. We'll be right back.