>> Narrator: From Palo Alto, California, it's CUBEConversations with John Furrier. >> Hello, welcome to our special CUBEConversation here in Palo Alto, California. I'm John Furrier, co-founder of SiliconAngle Media and co-host for the CUBE. I'm here with Phu Hoang who's the co-founder and chief strategy officer of DataTorrent. Great to see you again. Welcome back >> Thank you so much, John. >> This CUBEConversation. So, you're now the chief strategy officer, which is code words for you are the CEO and co-founder of the company. You bring in a pro guy, Churchwood we know very well, former EMC-er, real pro. Gives you a chance to kind of get down and dirty into the organization and get back to your roots and kind of look at the big picture. Great management team. Talk about what your background is, because I think I want to start there, because you have an interesting background. Former Yahoo executive, we've talked before. Take a minute to talk about your background. >> Yeah, sure. You know I think I'm just one of those super lucky engineer. I got involved with Yahoo way early in 1996. I think I was the fifth engineer, or so. I stayed there for 12 years, ended up running about close to 3,000 engineers, and had the chance to really experience the whole growth of the internet. We build out hundreds of sites worldwide, so all of engineering team develop all of those websites throughout the world. >> You must have a tear in your eye at how Yahoo ended up. We don't want to go there. Folks that don't remember Yahoo during the web1.0 days, it was the beginning of a revolution. I kind of see the same thing happening, like blockchain and what's going on now. A whole new wild west is happening, but back then you couldn't buy off the shelves. You had to certainly buy servers, but the software, you guys were handling kind of a first generation use case. >> That's right. >> Folks may or may not know, but Yahoo really was the inventor of Hadoop. Doing Hadoop at large scale, honestly ... MapReduce written by Google, but the rest is, you guys were deploying a lot of that stuff. You had to deal with scale and write your own software for big data, before it was called big data. >> That's exactly right. It's interesting, because originally we thought that our job was really customer-facing website, and all of the data crunching and massaging that we would actually be able to use enterprise software to do that. Very quickly we learned at the pace of scale data that we were generating that we really couldn't use that software. We were kind of on our own, so we had to invent approaches to do that. The thing we knew a lot was commodity servers on racks. So, we ended up saying, "How do I solve this big data processing problem using that hardware?" It didn't happen overnight. It took many years of doing it right, doing it wrong, and fixing it. You start to iterate around how to do distributed processing across many hundreds of servers. >> It's interesting, Yahoo had the same situation. And ultimately Amazon ended up having, cause they were a pioneer. People dismissed Amazon web services. Like, "It's just hosting and bare metal on the cloud." Really what's interesting is that you guys were hardening and operationalizing big data. >> That's right. >> So, I got to ask you the question, cause this is more of a geeky computer science concept, but batch processing has been around since the mainframe, and that's become normal. Databases, et cetera, software. But now, over the past 8 years in particular, as big data and unstructured data has proliferated in massive scale, certainly now with internet of things you see it booming. This notion of real time data in motion. You have two paradigms out there, batch processing, which is well known and data in motion, which is essentially real time. Self-driving cars ... Evidence is everywhere, where this is going. Real time is not near real time. >> That's right. >> In nanoseconds, people want results. This is a fundamental data challenge. What's your thoughts on this and how does this relate to how big data will evolve for customers? >> I think you're exactly right. I think as big data came, and people were able to process data, and understand it better, and derive insights from it, very quickly for competitive reason, they find out that they want those insights sooner and sooner. They couldn't get it soon enough. So, you have those opposing trends of more and more data, yet at the same time, faster and faster insight. Where does that go? When you really come down to it, people don't really want to do batch processing. They do batch processing, cause that was the technology that they have. If they have their way, they don't want to just ... Information is coming into their business. Customers are interacting with their products constantly, 24 by 7. Those events, if you will, that are giving them insights are happening all the time. Except, for a long time, they store it into a file. They wait til midnight, and then they process it overnight. More and more there are now capabilities in memory distributed to do that processing as it comes in. That's one of the big motivations for forming DataTorrent. >> I want to get to DataTorrent in a minute, but I want to get some of these trends, cause I think they're important to kind of put together the two big pieces of the puzzle, if you will. One is, you mentioned batch processing in real time. The companies, historically, have built their infrastructure and their operations IT, and whatever, around that, how storage was procured and deployed. Now with IOT and the edge of the network becoming huge, it's a big deal. So, data in motion, it's pretty much well agreed upon amongst most of the smart people, this is a big issue. Let me throw a little wrench in the equation. Cloud computing kind of changes the security paradigm. There's no perimeter anymore, so there's no door you can secure, no firewall model. Once you get in, you're in. That's where we've seen a lot of attacks on ransomware and a lot of cyber attacks. The penetration is everywhere. Now there's APIs and everything. When you bring cloud into it, and you bring in the fact that you've got data in motion, what is the challenge for the customer? How do top architects get their arms around this? What's the solution? What's your vision on that? >> Well, I will start by saying it's a hard problem. I think you're absolutely right. I think we're still in the phase where the problems are very visible about how do you solve this. I think we're still, as an industry, figuring out how to solve it, cause you're right, the security issue ... Security is not this one point tool. It's an entire ecosystem process for doing that. The cloud opens up all of those opportunities for fraud and so on. It's still an ongoing challenge. I think the trend of memory becoming cheaper and cheaper, so that things are done more in memory and less in storage could actually help a bit on that. But overall, security internal, external processes are ... >> It's a moving train. >> Yeah, it's moving. Exactly. >> Let me ask you about the big other trend to throw on top of this. This is really kind of where you see a lot of the activity, although some will claim that the app store is not seeing as many apps now as they used to be. Certainly the enterprises, massive growth and application development. So, ready-made apps with DevOps and Cloud have built a whole culture of infrastructure as code, which is essentially saying that I'm going to build apps and make the infrastructure kind of invisible. You're seeing a lot of apps like that, called ready-made apps, however you want to call it. Those are the things. How are you guys at DataTorrent handling and supporting that trend? >> We are right smack in the middle of exactly that trend. One of theses that we had was that big data's hard. Hadoop is hard. Hadoop is now 12 years old. Lots of people are using Hadoop, trying Hadoop, but yet it's still not something that is fully operationalized and easy for everybody. I think that part of that is big data's hard, distributed processing is hard, how to get all that working. There were two things we were focusing on. One was the real time thing. The other one was, how do we make this stuff a lot easier to use? So, we focus a lot on building tools on top of the open source engine, if you will, to kind of make it easy to use. The other one is exactly that, ready-made apps. As we continue to learn in working with our customers, and starting to see the patterns, putting kind of, bigger functional block together, so that it's easier to kind of build these big data application at this next layer. Machine learning, rule engines, whatever not. How do you piece that together in a way that is 80 percent done, so that the customer only has a little bit, the last mile. >> So, you guys want to be the tooling for that? >> Yeah, I think so. I think you have to. This stuff, you have to kind of go through the whole six layer of what it takes to get the final business value out. You're not going to have the skillset to do it. The more we can abstract and get it to the top, the better. >> Every company's got their own DNA. Intel has Moore's Law. You're the co-founder of DataTorrent. What's the DNA of your company, as the founder? Talk about what's the, what do employees you try to instill into your culture that is the DNA that you want to be known for? >> Interesting. So, I start out sort of on the technical or product side. Actually, our DNA is all about ops. We think that, especially in big data, there's lots of ways to do prototypes and get some proof of concept going.. Getting that to production, to run it 24 by 7, never lose data, that really has been hard. Our entire existence is around how to truly build 24 by 7, no data, fast application. All of our engineers live and breathe how to do that well. >> Ops is consistent with stability. It's interesting, Silicon Valley's going through its own transformation around programmers and the role of entrepreneurship. It's interesting, in the enterprise, they always kind of were like, "Oh, no big deal." Because at the end of the day they need stuff to run at five 9. These are networks. The old saying that Mark Zuckerberg used to have is, "Move fast and break stuff." They've changed their tune to, "Move fast and be a hundred percent reliable." This is the trend that the enterprises will always put out there. How do companies stay and maintain that ops integrity and still be innovative without a lot of command and control, and compliance restrictions? How do they experiment with this data tsunami that's happening and maintain that integrity? >> My answer to that is, I think, as an industry, we have to build products and tools to allow for that. Some of that is processes inside a company, but I think a lot of that can be productize. The advances in that big data processing layer, and how to recover, get new containers, and do all the right things, allow for the application developer not to have to worry about many of those segments. I think technology exists out there for tools to be developed to deal with a lot of that. >> I love talking with entrepreneurs and you're the co-founder of DataTorrent. Talk about the journey you've been on from the beginning. You have a new CEO, which as the CEO, you want to lighten the load up a little bit. It gets bigger, you got to have HR issues, things are happening. You're putting culture in place and trying to scale out and get a groove swing. Certainly Uber could've taken a few tips from your playbook, as bringing in senior management. You did it at the right time. Talk about your journey, the company, and what people should know about DataTorrent. >> We're just a bunch of guys that are just still trying to make a contribution to the industry. I think we saw an opportunity to really help people move towards big data, move towards real time analytics, and really help them solve some really hairy problems that they have coming up with data. From a skillset and personally, I think kind of my particular strength was really about that initial vision. Be able to build out a set of capabilities, and maybe get a first set of half a dozen wins, and really prove point. To sort of make it into a machine that has all the right marketing tools, and business development tools, and so on. It will be great to bring in someone like Guy, who has done that many, many times over, and has been super successful at that, to take us to the next level. >> Takes a lot of self awareness, too. You probably had your moments where, should you stay on, be the CEO ... But, what are you doing now, cause you get down and you can get into the products. Are you doing a lot more product tinkering? Are you involved the road map? What's your involvement day-to-day now? >> I love it, cause it's exactly what I enjoy most, which is really interacting with customers and users and really continue to hone in on the product market fit. And continue to understand, what are the pain points? What are the issues? And, how can we solve it? All coming from, not so much a services mentality, but a product mentality. >> At the cloud ops, too. That's a big area. So, what's the big problem that you solve for the customers? What's the big, hairy problem? >> Really easy, how to productize, how to operationalize this data pipeline that they have, so that they can truly be accepting real live business data that they are getting in, and giving them the insight. >> Been a lot of talk about automation and AI, lately. Obviously, it's a buzzword. Wikibon just put out a report called True Private Cloud that shows all the automation's actually going at and replacing non-differentiated labor, which actually the racking and stacking gear. Moving to values, actually is going to be more employment on that side. Talk about the role of automation in the data world, because if you just think about the amount of data that companies like Facebook and Yahoo take in, you need machine learning. You need automation. What is the key to automation in a lot of these new emerging areas around large data sets? >> It's so funny, yesterday I was driving. I was listening to a KQED segment, and they were talking about in its next phase, AI and machine learning is going to do sort of the first layer of all the reporting. So, you actually have reporters doing much more sophisticated reporting, cause there's an AI layer that has a template of what are the questions to answer, and they can just spill out all the news for you. >> Paid by cryptocurrency. >> Yeah. I think machine learning and AI will be everywhere. We will continue to learn, and it will continue to get better at doing more and more things for us, so that we get to kind of play at that creative, disruptive layer, while it does all the menial tasks. I think it will touch every part of our civilization. The technology is getting incredible. The algorithms are incredible. The power, the computing power to allow for that is getting exponential. I think it's super interesting that the engineers are super interested in it. Everything we do now revolves around ... When we talk about the analytics layer at real time, it's all about machine learning scoring and how to, rules and all that. >> Great to have you here on the CUBEConversation. Give you the last word. Give a quick plug about DataTorrent. What should your customers know about you guys? Why should they call you? >> We're a company solely focused on bringing big data applications to production. We focus on making sure that as long as you understand what you want to do with data, we can make it super fast, super reliable, super scalable. All that stuff. >> Co-founder of DataTorrent here and the CUBEConversation here in Palo Alto. I'm John Furrier. Thanks for watching. (synth music)

(techno music) >> Announcer: Live, from Silicon Valley, it's "the Cube," covering Mobile World Congress 2017. Brought to you by Mintel. >> Okay, welcome back everyone. We're here live in Palo Alto, California, covering Mobile World Congress, which is later in Spain right now, in Barcelona, it's gettin' close to bedtime, or, if you're a night owl, you're out hittin' the town, because Barcelona stays out very late, or just finishing your dinner. Of course, we'll bring in all theCube coverage here. News analysis, commentary, and of course, reaction to all the big mega-trends. And our next two guests is Guy Churchward who is the President and CEO of Data Torrent, formerly of EMC. You probably recognize him from theCube, from the EMC world, the many times he's been on. Cube alumni. And Phu Hoang, who's the co-founder and Chief Strategy Officer of Data Torrent. Co-founder, one of the founders. Also one of the early, early Yahoo engineers. I think he was the fourth engineer at Yahoo. Going way back on the 90s. Built that to a large scale. And Yahoo is credited for the invention of Hadoop, and many other great big data things. And we all know Yahoo was data-full. Guys, welcome to theCube's special coverage. Great to see you. >> Thank you so much. So I'm psyched that you guys came in, because, two things. I want to talk about the new opportunity at Data Torrent, and get some stories around the large scales experience that you guys have dealing with data. 'Cause you're in the middle of where this is intersecting with Mobile World Congress. Right now, Mobile World Congress is on the collision course between cloud-ready, classic enterprise network architectures with consumer, all happening at the same time. And data, with internet of things, is that going to be at the center of all the action? So, (laughing) these are not devices. So, that's the core theme. So, Guy, I want to get your take on, what attracted you to Data Torrent? What was the appeal for the opportunity? >> You mean, why am I here, why have I just arrived? >> I've always data-obsessed. You know this. From the days of running the storage business on their data protection, before that I was doing data analytics and security forensics. And if you look at, as you said, whether it's big data, or cloud, and the immersion of IOT, one thing's for sure, for me. It was never about big data, as in a big blob of stuff. It was all about small data sprawl. And the world's just getting more diverse by the second, and you can see that by Mobile World, right? The challenge then you have is, companies, they need to analyze their business. In other words, data analytics. About 30 years ago, when I was working for BA Systems, I remember meeting a general of the army. And he said the next war will be one in the data center, not on the battlegrounds. And so you really understand-- >> He's right about that. >> Yeah. And you have to be very, very close. So in other words, companies have started to obsess about what I call the do loop. And that really means, when data is created, and then ingesting the data, and getting insight from the data, and then actioning on that. And it's that do loop. And what you want to do, is you want to squeeze that down into a sub-second. And if you can run your analytics at the pace of your business, then you're in good shape. If you can't, you lose. And that means from a security perspective, or you're not going to win the bids. In any shape or form. That's not a business-- >> John: So speed is critical. >> Yeah, and people say, speed and accuracy. Because what you don't want to do is to run really really fast and fall off a cliff. So you really need to make sure that speed is there and accuracy is there. In the good old days, when I was running security forensics, you could either do complex end processing, which was a very small amount of information coming in and then querying it like crazy, or things like log management, where you would store data at rest, and then look at it afterwards. But now with the paradigm of all the technology catching up, so whether that's the disk space that you get, and the storage and the processing, and things like Hadoop with the clustering, you now break that paradigm. Where you can collect all the information from a business and process it before you land the data, and then get the insight out of it, and then action. So that was my thing, of looking and saying, look, this whole thing's going to happen. In last year -- >> And at large scale, too. I mean, what you're talking about in the theoretical side makes a lot of sense, but also putting that into large scale, is even more challenging. >> Yeah, we had, when I was going through the processes, dating, you know, to see whether was a company that made sense, I chatted one of our investors. And they're also a customer. And I said, why did you choose Data Torrent? And they said, "We tested everything in production, we tested all the competitive products out there, and we broke everything except Data Torrent. And actually, we tested you in production up to a billion events per second, and you didn't break. And we believe that that quantity is something that you need as a stepping stone to move forward." >> And what use cases does that fit for? Just give me some anecdotal (snaps fingers) billion transactions. At that speed, what's some use cases that really take advantage of that? >> They were mastering in, what I would call, industrialization of IT. So in other words, once you get into things like turbines, wind generation, train parts. We're going to be very very soon, looking out of a window and seeing -- >> John: So is it flow data? Is it the speed of the flow? Is it the feed of all the calculations, or both? >> It's a bit of both. And what I'll do, is I'll give Phu a chance, otherwise, we'll end up chatting about it. >> John: Phu, come on, you're the star. (laughing) When you founded this company, you had a background at Yahoo, which you built from scratch, but that was a first-mover opportunity, Web 1.0, as they say. That evolved up and then, everyone used Yahoo Finance. Everyone used Yahoo Search as a directory early on. And then everything just got bigger and bigger and bigger, and then you had to build your own stuff with Hadoop. >> Yeah. >> So you lived it. The telcos don't have the same problem. They actually got backed into the data, from being in the voice business, and then the data business. The data came after the voice. So what's the motivation behind Data Torrent? Tell us a little bit more. >> It's exactly what you say, actually. Going through the 12 years at Yahoo, and really, we learned big data the hard way. Making mistakes month after month, about how to do this thing right. We didn't have the money, and then we found out that, actually, proprietary systems of the shelf system that we thought were available, really couldn't do their jobs. So we had to invent our own technology, to deal with the kind of data processing that we had. At some point, Yahoo had a billion users using Yahoo at any given point in time, right? And the amount of impressions, the amount of clicks, the amount of activity, that a billion users have, onto the system. And all of the log files that you have to process to understand what's going on. On the other side of that, we need to be able to understand all of those activities in order to sell to our advertisers. Slice and dice behaviors and users, and so on. We didn't have the technology to do that. The only thing we knew how to do was, to have these cheap racks of cheap servers, that we were using to serve webpages. And we turned to that to say, this is what we're going to need to do, to solve these big data problems. And so, the idea of, okay we need to take this big problem and divide it into smaller pieces, so that we can run on these cheap servers, sort of became the core tenant of how we do distributor processing that became Hadoop, at the end of the day, right? >> You had big data come in because you were, big data-full, as we say. You weren't building software to solve someone else's problem. You had your own problem, you had a lot of data. You were full with data. >> Exactly. >> Had to go on a data diet, to your point. (crosstalk) >> And no one to turn to. >> And no one to turn to. >> All right. So let's spin this around or Mobile World Congress. 'Cause the big theme is, obviously, we all know what device is. In fact, we just released here on theCube early this morning Peter Burris pre-announced our new research initiative called IOTP. Which stands for Internet Of Things And People. And so now you add the complexity of people devices, whether that's going to be some sort of watch, phones, anything around them. That adds to the industrial aspect of turbines and what not. Internet of Things is a new edge architecture. So the data tsunami coming, besides the challenges of telcos to provision these devices, are going to be very challenging. So the question I want to ask you guys is, how do you see this evolving, because you have certainly connectivity. Yeah, you know, low latency, small little data coming from the windmills or whatever. Versus big high-dense bandwidth, mobility. And then you got network core issues, right. So how does this going to look like? Where does the data piece fit in? Because all aspects of this have data. What's your thoughts on this, and architecture. Tell us about your impressions, and the conversations you've had. >> First of all, I think data will exist everywhere. On the fringe, in the middle, at the center. And there's going to be data analytics and processing in every path of that. The challenge will be to kind of figure out what part of processing do you put on the fringe, what part do you put at the center. And I think that's a fluid thing that is going to be constantly changing. Going back to the telcos. We've had numbers of conversationw with telcos. And, yes we're helping them right now with their current set of issues around capacity management and billing, all those things. But they are also looking to the next step in their business. They're making all this money from provisioning, but they know they sit on top of this massive amount of really valuable data, from their customers. Every cellphone is sending them all of this data. And so there's a huge opportunity for them to monetize, or really produce value, back to their customers. And that could come in form of offers, to customers. But now you're talking about massive analytics targeting. That is also real-time, because if you're sending an offer to someone at a particular location, if you do that slowly, or in batch, and you give them an offer 10 minutes later, they're no longer where they are. They're 10 minutes away, right? >> Well, first two questions to follow up on that. One, do they know they have a data advantage opportunity here? Do they know that data is potentially a competitive advantage? >> From our conversation, they absolutely do. They're just trying to figure out, so what do we do here? It's new to them. >> I want to get both your perspectives. Guy, I want you to weigh in on this one, 'cause this is another theme that's coming out of the reporting and analysis from Mobile World Congress. This has come also from the cloud side as well. Integration now, is more important than ever, because, for instance, they might have an Oracle there, there might be Oracle databases outside their network. That they might want to tap into. So tapping other people's data. Not just what they can get, the telcos. It's going to be important. So how do you guys see the integration aspect, how we, top of the first inning, national anthem going on. I mean, where are we in this integration? There's a pregame, or, what inning are we in on this? >> Yeah, we're definitely not on the home run on it. I think our friend, and your friend Steve Manly, I sat down with him, and I gave him a brief, you know, what we were doing, and he was blown away by the technology and the opportunity, but he was certainly saying, but the challenge is the diversity of the data types. And then where they're going to be. Autonomic cars. You know each manufacturer will tell the car behind it, what it just experienced, but the question is, when will a Tesla tell a Range Rover, or tell a BMW? So you have actually -- >> They're different platforms, just different stats, it's a nightmare. >> Right. So in other words, >> And trackability. And whether it's going to be open APIs, whether it's technologies like Kafka. But the integration of that, and making sure that you can do transformation and then normalize it and drive it forward. It's kind of interesting, you know. You mentioned the telco space, and do they understand it. In some respects, what Phu went through with Yahoo, in other words, you go to a webpage, you pull it up, it knows you because of a cookie and it figures out, and then sells advertising to you on that page. Now think about you as a location, and you're walking past a Starbucks, and they want to sell you a coffee for ten cents less than they would normally do. They need to know you're there then. And this is the thing, and this is why real-time is going to be so critical. And similarly, like you said, you look out the window and you see DHL, or UPS, or FedEx drones out the window. You not only have an insight issue. You also have a security issue, you have a compliance issue, you have a locational issue. >> I think you're onto something. And I think I actually had this talk today with Steve Manly EMC World last year, around time series data. So this is interesting. Everyone wants to store everything, but it actually might not be worth anything anymore. If the drone is delivering your package, or whatever realtime data is in realtime, it's really important right there in realtime, or near realtime. It might not be worth anything after. But yet a purchase at a store, at a time, might be worth knowing that as a record to pull in. You get what I'm saying? So there's a notion of data that's interesting. >> And I think, and again, Phu's the expert. I'm still running up onto it. It's just a pet hobby, an obsession of mine. But the market has this term ETL. In other words, Extract, Transform, Land. Or load. But in essence, it's always talked about in that (mumbles) batch. In other words, I get the data, transform it, drop it, and then I have a look at it. We're going upside-down. So the idea now is to actually extract, transform, insight, action, then landing. So in other words, get the value at the fresh data, before it's the data late. Because if you set the data late, by default, it's actually stale. And actually, then there's the fascination of saying, if you're delivering realtime data to a person, you can't think fast enough to actually make a live decision. So therefore, you've almost got any information that comes to you, has to tier out. So it comes to a process. You get that fresh use of it, and then it drops into a data lake. And so I think there's using both, but I think what will you see in the market, and, again, you've experienced the disk flash momentum that happened last year. You're going to see that from a data source from at-rest, advanced, to real-time data streams on our applications next year. So I think the issue is, the formative year, and back to your, you know, get it right, get the integration, but make sure your APIs are there, talking to the right technologies. I think everything's going to be exciting this year and new and fresh and people really want to do it. Next year is going to be the year where you're going to see an absolute changing of the guards. >> And then also the SLA requirements, they'll start to get into this when you start looking at integration. >> You're absolutely right. Actually, the SLA part is actually very very important here. Because, as you move analytics from this back world, where it has, you do it once a day, and if it dies, it's okay, you just do it again. To where it is now continuous, 24 by 7, giving you insight continuously about your business, your people, your services, and so on. Then all of a sudden, it has to have the same characteristics as your business. Which is, it's 24 by 7, it can never go down, it can never lose data. So, all of a sudden you're putting tremendous requirements on an analytics system, which has, all the way from the beginning of history 'til now, been a very relaxed batch thing, to all of a sudden being something that is enterprise-grade, 24 by 7. And I think that that's actually where it's going to be the toughest nut to crack. >> So tell about some of the things that you've learned. And pretend for a second, let's pretend that you, as a co-founder at Data Torrent, and Guy, and you are teamed up. You guys run this telco. Let's just make one up, Verizon. Or AT&T, or pick one. And you sit there saying, okay, you've got the keys to the kingdom. And you can do whatever you want (laughing). You can be Donald Trump, or you can be whoever you want. You can fire everybody, or you can pick it over and run it. What would you do? You know you've got IOT. So this is business model innovation opportunities. I want you to put the technical hat on, plus knowing what you know around the business model opportunities. What do you do? You know IOT's an opportunity. Amazon is going after that heavily. Do you bolt a cloud together? Do you go after Amazon? Do you co-op with Amazon? Do you co-integrate? Do you grab the IOT? Do you use the data? I mean, given where we are today, what's the best move if we were consulting with this. >> You know, I will be the last person to be talking about giving advice to a telco. But since we are, we own our own telco here, and then we're pretending, I would say the following. IOT is going to happen, right? Earlier, when I say a billion people, that's just human beings. Once you now talk about censoring, you can program how many times they can send you data per second, then the growth in volume is immense, right? I think there's a huge opportunity, as a telco, in terms of the data that they have available and the insight that they could have about what's going on. That is not easy. I don't think that, as a telco, in the current DNA of a telco, I can go ahead and do all that analytics and really open up my business to the data insight layer. I would partner, and find a way-- >> Well, we're consulting, we're going to sit around and say hey, what do we have? We have relationship with the consumer, big marketing budgets. We can talk to them directly, we have access to their device. >> But you'll bifurcate the business. We're in the boardroom here, this is nothing more than that. But I would look at it and say look, you've got a consumer business, the same as in IOT. There's really, for me, there's three parts of IOT. There is the bit that I love which, you can geek out, which is basically the consumer market, which, there's no money in for a large-scale tenant, right, enterprise. And then you have the industrialization of IOT, which is I've got a leaky pipe, and I want a hardened device, ruggedized, which is wifi, so, now as a telco, I could create a IOT cloud, that allows me to put these devices out there, and in fact, I use Arlo, the little cameras. And they've got one now, where I can basically float it with its own cellular signal. So it's its own cellphone. That's a great use of IOT for that. And then you step to the consumer side of, I've got a cellphone, and then what I'll do is literally, in essence, riff off what Yahoo did in the early days and say, I'm now the new browser. The person's the browser. So in other words, follow the location, follow where he is, and then basically do locational-based advertising. >> By the way, you have to license the patent from our earlier guest, he'll say will he leak, 'cause he's got th6e patent on personal firewall for personal server. He's built a mobile personal server. >> Yeah. >> But this is the opportunity around wireless. Why I love the confusion, but the opportunity around wireless right now is, you can get bandwidth at high capacity. You have millimeter wave four, that doesn't go through walls, but you have other diverse frequencies and spectrum for instance, you can blend it all together to have that little drip signal, if you will, going into the cloud from the leaky pipe. Or if you need turbine, full-fat pipe, you maybe go somewhere. So, I think this is an interesting opportunity. >> And they're going to end up watching the data centers as well. There's still the gamut of saying our customer is going to continue to support their own data centers, or are there going to be one to a hundred data centers out there? And then how does selling a manufacturer or a telco play into that, and do they want to be that guy or not? >> Guy, Phu, thanks for coming in. I want to give you guys a chance to put a plug in for Data Torrent. Thanks for sharing some great commentary on the industry. So, what's up with you guys? Give us the update. Are you hiring? You growing? What are you guys doing? Customers? What's the update? Technology, innovations? >> So we've got a release coming out tomorrow which is a momentum release. I can't talk too much about the numbers, but in essence, from a fact base, we have a thing called a patchy apex. So it's open sourced, so you can use our product for free. But that's growing like gangbusters. From a top-level project, that's actually the fastest-growing one, and it's only been out for seven months. We just broke through 50,000 users on it. From our product, we're doing very well on the back of it. So we actually have subscription for the production side. >> So revenue is a subscription model. >> Yeah, and we meet both sides. So in other words, for the engineer who writes it, you've got the open source. And then when you put it into production, from the operations side, you can then license our products to enable you to manage an easy-- >> So when it gets commercialized, you pay as you go, when you use it. >> And you don't have to, if you don't want to. You've got all the tools to do it. But, we focus for our products group of, time to value, total cost of ownership. We're trying to bring Hadoop and real scale, realtime streaming to the masses. So what's the technology innovation? What's the disruptive enabler for you guys? >> I think we talked about it, right? You've got two really competing trends going on here. On one side, data is getting more and more and more massive. So it's going to take longer and longer to process it. Yet at the other side, business wants to be able to get data, have insight, and take action sub-second. So how do you get both at the same time? That's really the magic of the technology. >> Thanks for coming in. Great to meet you, Phu. I'd love to talk about the old Yahoo days, a total throwback, Web 1.0, a great time in history, pre-bubble bursting. Greatness happening in the valley and all around the world, and I remember those days clearly. Guy, great to see you. Congratulations on your new CEO committee. And great to have you on theCube. This is theCube bringing the coverage, and commentary, and reaction of Mobile World Congress here, in California. As everyone goes to bed in Barcelona, we're just gettin' down to the end of our day here in the afternoon in California. Be right back with more after this short break. (techno music)

>> Why don't I spend just a couple minutes talking about what we mean by digital enactment, turning data in models and models into action. And then we'll jump directly into, I'll introduce the panelists after that, and we'll jump directly into the questions. So Wikibon SiliconAngle has been on a mission for quite sometime now to really understand what is the nature of digital transformation, or digital disruption. And historically, when we've talked about digital, people talk about a variety of different characteristics of it, so we'll talk about new types of channels and activity on the web, and a many number of other things. But to really make sense of this, we kind of felt that we had to go to a set of basic principles, and utilize those basic principles to build our observations up. And so what we started with is a simple observation that, if it's not digital, or if it's not data, it ain't digital. By that we mean fundamentally the idea of digital business is how are we going to use data as an asset to differentially drive our business forward? And if we borrowed from Drucker, Drucker used to like to talk about the idea that business exists to create sustained customers, and so we would say that digital business is about applying data assets to differentially create sustained customers. Now to do that successfully, we have to be able to, as businesses, be able to establish a set of strategic business capabilities that will allow us to differentially use data assets. And we think that there are a couple of core strategic business capabilities required. One is human beings and most businesses operate in the analog world, so it's how do we take that analog data and turn it into digital data that we can then process. So that's the first one, the notion of an IOT as a transducer of information so that we can generate these very rich data streams. Secondly we have to be able to do something with those data streams, and that's the basis of big data. So we utilize big data to create models, to create insights, and increasingly through a more declarative style, actually create new types of software systems that will be crucial to driving the business forward. That's the second capability. The third capability is one that we're still coming to understand, and that is we have to take the output of those models, the output of those insights, and then turn them back into some event that has a consequential moment in the real world, or what we call systems of an action. And so the three core business capabilities that have to be built are this capture data through IOT, big data to process it, systems of an action also through IOT, through actuators, to actually that have a consequential action in the real world. So that's the basis of what we're talking about. We're going to take Flavio's vision that he just laid out, and then we, in this panel, are going to talk about some of the business capabilities necessary to make that happen, and then after this, David Foyer will lead a panel on specifically some of the lower level technologies that are going to make it work. Make sense guys? >> Sounds good (mumbles). >> Okay, so let me introduce the panelists. Over, down there on the end, Ted Connell. Ted is from Intel, I don't know if we can get the slide up that has their names and their titles. Ted, why don't you very quickly introduce yourself. >> Yeah, thank you very much. I run Solution Architecture for the manufacturing and industrial vertical, where we put together end to end ecosystem solutions that solve our clients business problems. So we're not selling silicone or semiconductors, we're solving our clients problems, which as Flavio said, requires ecosystem solutions of software, system integrators, and other partners to come together to put together end solutions. >> Excellent, next to Ted is Steve Madden of Equinix. >> Yeah, Steve Madden. Equinix is the largest interconnection, global interconnection company and a lot of the ecosystems that you'll be hearing about, come together inside our locations. And one of the things I do in there is work with our big customers on industry vertical level solutions, IOT being one of them. >> Phu Hoang, from Data Torrent. >> Hi, my name's Phu Hoang, I'm co-founder and chief strategy of a company called Data Torrent, and at Data Torrent, our mission is really to build out solutions to allow enterprises to process big data in a streaming fashion. So that whole theme around ingestion, transformation, analytics, and taking action in sub second on massive data is what we're focusing on. >> And you're familiar with Flavio. Flavio, will you take a second to introduce yourself. >> Yes, thank you, I am leading a company that is trying to manifest the vision highlighted here, building a platform. Not so much the applications, we are hosting the applications (mumbles) the data management and so forth. And trying to apply the industrial vertical first. Big enough to keep us busy for quite a while. >> So in case you didn't know this, we have an interesting panel, we have use case, application, technol infrastructure, and platform. So what' we'll try to do is over the next, say, 10 minutes or so, we're going to spend a little bit of time, again, talking about some of these business capabilities. Let me start off by asking each of you a question, and I will take, if anybody is really burning to ask a question, raise your hand, I'll do my best to see you and I'll share the microphone for just long enough for you to ask it. Okay, so first question, digital business is data. That means we have to think about data differently. Ted, at Intel, what is Intel doing when they think about data as an asset? >> So, Intel has been working on what is now being called Fog, and big data analytics for over a generation. The modern xeon server we're selling, the wire in the electronics if you will, is 10 silicon atoms wide. So to control that process, we've had to do what is called Industry 4.0 20 years ago. So all of our production equipment has been connected for 20 years, we're running... One of our factories will produce a petabyte of data a day, and we're running big data analytics, including machine learning on the stuff currently. If you look at an Intel factory, we have 2,000 fit clients on the factory floor supported by 600 servers in our data center at the factory, just to control the process and run predictive yield analytics. >> Peter: So that's your itch? >> Our competitive advantage at Intel is the factory. We are a manufacturer, we're a world class manufacturer. Our front end factories have zero people in it, not that we don't like people, but we had to fully automate the factory because as I speak, tens of thousands of water molecules are leaving my mouth, and if one of those water molecules lands on a silicon, it ain't going to work. So we had to get people physically out of the factory, and so we were forced by Moore's Law, and the product we build, to build out what became Fog, when they came up with the term seven years ago, we just came to that conclusion because of cost, latency, and security, it made sense to, you know, look, you got data, you got compute, there's a network between. It doesn't matter where you do the compute, bring the compute to the data, the data to the compute. You're doing a compute function, it doesn't matter where you do it. So Fog is not complicated, it's just a distributed data center. >> So when you think about some of the technologies necessary to make this work, it's not just batch, we're going to be doing a lot of stuff in real time, continuously. So Phu, talk a little bit about the system software, the infrastructure software that has to be put in place to ensure that this works for them. >> I think that's great. A little bit about our background, the company was founded by a bunch of ex-Yahoos that had been out for 12, 15 years from the early days. So we sort of grew up in that period where we had to learn about big data, learn about making all the mistakes of big data, and really seeing that nowadays, it's not good enough to get insight, you have to get insight in a timely fashion enough to actually do something about it. And for a lot of enterprise, especially with human being carrying around mobile phones and moving around all over the place, and sensors sending thousands, if not millions of events per second, the need for the business to understand what's going on and react, have insight and react sub second, is crucial. And what that means is the stuff that used to be batch, offline, you know, can kind of go down, now has to be continuous, 24 by seven. You can't lose data, you got to be able to recover and come back to where you were as if nothing has happened with no human intervention. There's a lot of theme around no human intervention, because this stuff is so fast, you can't involve human beings in it, then you're not reacting fast enough. >> Can I real quickly add one thing first? >> Peter: Sure. >> We think of data at Intel in half life terms. >> Yeah, that's exactly right. >> The data has valuable right now. If you wait a second, literally a second, the data has a little bit of value. You wait two second, it's historical data you can run regressions, and tell you why you screwed up, but you ain't going to fix anything. >> Exactly. >> If you want to do anything with your data, you got to do it now. >> So that, ultimately, we need to develop experience, a creed experience about what we're doing. And the stuff we're doing in applications will eventually find itself into platforms. So Flavio, talk to us a little bit about the types of things that are going to end up in the platform to ensure that these use cases are made available to, certainly, businesses that perhaps aren't as sophisticated as Intel. >> Yes, so in many ways, we are learning from what is going on in the Cloud, and has to come through this continuum, all the way into the machines. This break between what's going inside the machine, and old 1980 microprocessor and the server, and the Cloud server with virtualization on the other side cannot leave. So it has to be a continuum of computing so you can move the same function, the same container, all the way through first. Second, you really have to take the real time very, very seriously, particularly at the edge, but even in the back so that when you have these end to end continuum, you can decide where you do what. And I think that one of the models that was in that picture with a concentric circle is really telling what we need to learn first. Bring the data back and learn, and that can take time. But then you can have models that are lightweight, that can be brought down to the front, and impact the reaction to the data there. And we heard from a car company, a big car company, how powerful this was when they learned that the angle of a screwdriver, and a few other parameters, can determine the success of screwing something into a body of a car, that could go well, or could go very, very bad and be very costly. So all the learning, massive data, can come down to a simple model that can save a lot of money and improve efficiency. But that has to be hosted along this continuum. >> So from a continuum, it means we still have to have machines somewhere to do something. >> Touching the ground, touching the physical world requires machines, actuators. >> Peter: Absolutely, so Steve, what is Equinix doing to simplify the thinking through of some of these infrastructure issues? >> Yeah, I mean, the biggest thing that people find when they start looking at millions of devices, millions of data capture points, transferring those data real time and streaming it, is one thing hasn't changed and that's physics. So where those things are, where they need to go, where the data needs to move to and how fast, starts with having to figure out your own topology of how you're moving that data. As much as it's easy to say we're just going to buy a platform and choose a device, and we'll clink them together, there's still a lot of other things that need to be solved, physics being the first one. The second one, primarily, is volumes. So how much bandwidth and (mumbles) you're going to require. How much of that data are you going to back haul to centralized data center before you send it up to a Cloud? How much of it are you going to leave at the edge? Where do you place that becomes a bigger deal. And the third one is pretty much every industry has to deal with regulations. Regulations control what you can and can't do in terms of IT delivery, where you can place stuff, where you cannot place stuff, data that can leave the country, data that can't. So all these things mean that you need to have a thought through process of where you're placing certain functions, and what you're defining as your itch between the digital and physical world. And Equinix is an interconnection company that's sitting there as a neutral party across all the networks, all the clouds, all the enterprises, all the providers to help people figure that out. >> So before I ask the audience a question, now that I'm down here so I can see you so be prepared, I'm going to ask some of you a question. When you think about the strategic business capabilities necessary to succeed, what is the first thing that the business has to do? So why don't I just take Ted, and just go right on down the line. >> Yeah, so I think this is really, really important. I work with many, many clients around the world who are doing five, 10, 15 POCs, pilots, and the internet things, and they haven't thought through a codified strategy. So they're doing five things that will never fit together, that you will never scale, and the learnings you're using, you really can't do that much with. So coming up with what is my architecture, what is my stack going to look like, how am I going to push data, what is my data... You know, because when you connect to these things, I can't tell you how much data you're going to get. You're going to be overwhelmed by the data, and that's why we all go to the edge, and I got to process this data real time. And oh, by the way, if I only have one source of data, like I'm connecting to production equipment, you're not going to learn anything. 98% of that data's useless, you got to contextualize the data with either an inspection step, or some kind of contextualization that tells you if this then that. You need the then that, without that, your data is basically worthless. So now you're pulling multiple sources of data together in real time to make an understanding. And so understanding what that architecture looks like, spend the time upfront. Look, most of us are engineers, you know five percent additional work upfront saves you 95% on the backend, that's true here. So think through the architecture, talk to some of us who have been working in this area for a long time. We'll share our architecture, we have reference architecture that we're working with companies. How do you go from industry 2.0 or industry 3.0, to industry 4.0? And there is a logical path to do it, but ultimately, where we're going to end up is a software defined universe. I mean, what's a cloud? It's a software defined data center. Now we're doing software defined networks, software defined storages, ultimately we're going to be doing software defined systems because it's cheaper. You get better capital utilization, better asset utilization, so we will go there, so what does that mean for you infrastructure, and what are you going to do from an architectural perspective, and then take all of your POCs and pilots, and force them to do that specifically around security. People are doing POCs with security that they don't even have any protocols, they're violating all their industry standards doing POCs, and that's going to get thrown out. It's wasted time, wasted effort, don't do it. >> Steve, a couple sentences? >> Yeah, essentially it's not going to be any prizes for me saying think interconnection first. A lot of our customers, if we look at what they've done with us, everyone from GE to real time facial recognition at the edge, it all comes down to how are you wired, topology wise, first. You can't use the internet for risk reasons, you can't necessarily pay for multiple (mumbles) bandwidth costs, et cetera. So low latency, 80% lower latency, seven times of bandwidth at half the cost is a scalable infrastructure to move (mumbles) around the planet. If you don't have that, the rest of the stuff (mumbles) breakdown. >> Peter: Phu? >> Well I would say that analytics is hard, analytics in real time is even harder. And I think with us talking to our customers, I feel for them, they're confused. There's like a million solutions out there, everybody's trying to claim to do the same thing. I think it's both sides, consumers have to get more educated, they have to be more intelligent about their POCs, but as an industry, we also have to get better at thinking about how do we help our customer succeed. It's not about let me give you some open source, and then let me spend the next 10 months charging you professional services to help you. We ought to think about software tools and enterprise tools to really help the customer be able to think about their total cost (mumbles) and time to value to handle this thing, because it's not easy. >> Peter: Flavio. >> Yeah, we're facing an interesting situation where the customers are ready, the needs are there, the marketing is going to be huge, but the plot, the solution, is not trivial. It is maturing and we are all trying to understand how to do it. And this is the confusion that you see in many of these half baked solution (mumbles). Everything is coming together, and you have to go up the stalk and down the stalk with full confidence, that's not easy. So we all have to really work together. Give ourselves time, be feeling that we are in a competitive world, preparing for addressing together a huge market. And trying to mature these solutions that then will be replicated more and more, but we have to be patient with each other, and with the technologies that are maturing and they're not fully there and understood. But the market is amazing. >> Peter: So we have a Twitter question. >> Man: It's being live streamed, the audience is really engaged online as well, digital. So we have a question from Twitter from Lauren Cooney saying, "Would like to know what industries would "be most impacted with digitization "over the next five years." >> Which one won't be? (men laughing) All of them, what we've seen, the business model is the data. I mean, our CEOs calling data the new gold. I mean, it's the new oil. So I don't know of anything, unless you're doing something that is just physical therapy, but that even data, you can do data on that. So yeah, everything, yeah, I don't know of anything that won't be. >> I think the real question is how is it going to move through industries. Obviously it's going to start with some of the digital native, it's all ready deep into that, deep into media, we're moving through the media right now. Intel's clearly a digital company, and you've been working, you've been on this path for quite some time. >> Let me give you a stat. Intel has a 105,000 people, and 144,000 servers. So we're about 1.5 server to people, that's what kind of computation we're (mumbles). >> Peter: We can help you work on that. >> If you do like the networking started by (mumbles) the internet, then content delivery, and media, hard media, et cetera, is gone. Financial services and trading exchanges pretty much show what digital market's going to be in the future. Cloud showed up, and now, I think he's right, it's effecting every industry. Manufacturing, industrial, health professional services are the top three right now. But people who shop to ask for help went from every industry on every country, for that matter. >> Our customers are, you know, the top players in almost every vertical. You start out as a small company thinking that you're going to attack one vertical, but as you start to talk about the capability, everybody (mumbles) wait, you're solving my problem. >> Peter: (mumbles) are followers, is what you mean. >> Yeah, because what business would say, hey, I don't want to know what's going on with my business, and I don't want to take any action. >> Add to that it's an ecosystem of ecosystems. No one, by themselves, is going to solve anything. They have to partner and connect with other people to solve the solution. >> So I'll close the panel by making these kind of summary comments, the business capabilities that we think are going to be most important are, first off, when we talk about the internet of things, we like to talk about the internet of things and people. That the people equation doesn't go away. So we're building on mobile, we're building on other things, but if there's a strategic capability that's going to be required, it's going to be how is this going to impact folks who actually create value in the business. The second one, I'll turn it around, is that IT organizations have gone through a number of different range wars, if you will, over the past 20 years. I lived through IT versus telecom, for example. The IT, OT conflict, or potential conflict, is non trivial. There's going to be some serious work that has to be done, so I would add to the conversation that we've heard thus far, the answers that we've heard thus far, is the degree to which people are going to be essential to making this work, and how we diffuse this knowledge into our employees, and into our IT and professional communities is going to be crucial, especially with developers because Flavio, if we are, right now, trying to figure stuff out, it really matures when we think about the developer world. Okay, so I want to close the first panel and get ready for the second panel. So thank you very much, and thank you very much to our panelists. (audience applauding) And if we could bring David Foyer and the second panel up, we'll get going on panel two. Oh, we're going to get together for a picture. (exciting rhythmic music)