 Live from the FIA Barcelona Grand Vía Compensator in Barcelona, Spain, it's The Cube at HP Discover Barcelona 2014. Brought to you by headline sponsor HP. Okay, welcome back everyone. We are live in Barcelona, Spain for HP Discover 2014, the European edition that we were in Las Vegas earlier this year for HP Discover in North America. This is The Cube, our flagship program. We go out to the events and extract the signal from the noise. We talked to the execs, we talked to the product managers, we talked to the geeks, entrepreneurs, CTOs, and we had a great segment here. We grabbed up day three, getting down to the wire, day three of three days of coverage. We got Fernando Lucini, CTO of Big Data, and Chris Goodfellow, CTO of Idol on Demand. Basically the tech guys, the entrepreneurs who invented the products here inside The Cube, so great to have you guys. Thanks for coming on The Cube. No, thanks for being here. Thank you. So Fernando, we were just talking before we came on, totally geeking out on crowd-chatting data, Idol on Demand, so much good stuff going on. Under the hood, some stuff's trivial, some stuff is not. Before we get into some of the things around Idol on, Haven on Demand and Idol on Demand, what shared the folks out there? Just some of the complexities that goes on in Big Data. I mean, like, it sounds easy, but there's magic that happens. What's some of that tech going on? How hard is it? I mean, it's hard, but hard is a bad word, right? We're trying to simplify something that's actually quite fun for everyone, but so a good example of this is as we, so we're a company that deal with information, right? We dream information, we work with information, we look at these problems, we look at video, voice, and some of it is harder than others, so the image recognition is tough stuff, you know? You're looking at something that is even difficult to interpret with your eye, let alone with a machine, right? Then we have all the tech stuff that has such richness and context, this magic. I don't see it as difficult, I just see it as rich. Well, first of all, geeks love this stuff because it's a geek dream, right? There's a lot of stuff to get your hands on. One of the things that we talk about all the time on theCUBE is around Big Data is the diversity of geekness. I mean, there's machine learning, there's data science, there's large-scale systems. Talk about that for a minute. What do you see as a spectrum of skill sets and challenge areas and Big Data? The engineering standpoint and development standpoint. So the variety thing is important and we should spend a second on that, right? So variety historically has been a barrier to try and get things done, right? So if you have to worry about all the specialist knowledge and hardware that you need to deal with video, all the specialist knowledge you have to deal with, voice, or text, it becomes a barrier for you to achieve what you want to do, which ultimately is do something amazing with that thing, right? So what I think is definitely changing and we're changing at HP for sure is making much of this very approachable, right? Now it just requires for you to have a great idea. It just requires for you to know that you want to use video and understand that the outcomes and then it just takes the execution to become easier. So if you need to consume, and I use simple examples, right? So we've got customers, I was talking to somebody in Oslo and got one of our customers who processes about two billion documents a year. I mean, these are human beings processing information and we did the math on that and it would take like 200 people, 24 years to read this data, right? And they do this 50 people, right? A year. So these guys don't want to know how much hardware they throw at it or they want to just do it. And I think that's what's really changed. Now it's about what do you want to do? X, Y, and Z. Well computation is certainly awesome. I mean, you can spin up now with cloud, computation plus is at will. Well, it just boom. Oh, we do it all the time. We press the button in helium and we get ourselves a thousand nodes and we're off to the races just like that. You get special keys to the kingdom with helium? Tell the truth. Wow. Not special keys to the kingdom, but we're brothers and sisters and we do it. You get special backdoor access, family class. Oh no, nothing like that, nothing like that. But it's there. I mean, that's your question. That's what's there now. So if I want to create a business for Chihuahuas to process a large amount of data and I just want to have that computer, I press the button, the computer's there, I get on with my life. It's super exciting. And Chris, I want to get you in here because idle on demand is free and people can try it out. But it's a fun time right now to be in big data, be a geek in coding away because there's so much action happening. I mean, when you guys did the product, I mean, you're a small team, talk about the process. What did you guys do? You guys sitting just whiteboarded out? Did you back in the napkin? How did this all come together? Well, as you say, I mean, it's an exciting time. We've been doing big data for years. I mean, what everyone is now talking about, I mean, we've been building systems that index millions of emails for billions of emails for a long time. But what we did with idle on demand is we realized that we could take that powerful technology and get it out into the hands of everybody, the developer in their bedroom, the developer who's been given a small project that, okay, perhaps he doesn't have the funding to go and set up a server farm and all that. So it was really about opening that up. So what we really did is we sat down and started to work out, okay, how do I make this available to a developer? We went through the process of getting things out there as soon as possible. I mean, so from project kickoff to our first release was literally just three months. So we could get it out there. We could start talking to developers, start going to developer events, hackathons, and getting it into the hands and start to work that out. Now, at the same time, we had to, in the back end, as you were saying, there's some interesting problems here. We've solved a lot of the functionality pieces through kind of what we've been used to. So we've got a lot of experts in how to handle image, how to handle video, extract the meaning out of text. But obviously, building something on demand where you are dealing with hyperscale volumes, knowing how to scale, a lot of interesting problems there, knowing how to protect yourself in false quotas, ensure that fairness of service. Things that are very much part of the modern cloud era. So I got to ask you guys here. We've got a question on our crowd chat. Go to crowdchat.net slash hbsgubber, our big data app, it's supposed to be an engagement container because we was just showing you the big data stuff. A question came in from the crowd father, one of our anonymous handles. Crowd father. The crowd father. We confused with the crowd captain and the crowd doctor and then there's the crowd business, crowd exec, the crowd guy. So the crowd father would like to know, what about the evolution of reasoning? You know, Watson at IBM is certainly here. He's got the machine as their shiny object. Meta reasoning, metadata, the role of data, active data, passive data, reasoning. As geeks, how do you guys look at that? What are you guys doing? What's the hot area around reasoning? Because that's an area to simplify. If you can reason, you get some personalization, collective intelligence, roll the crowd. Well, there's plenty to choose from. And I know we've got the Watson's on one side, we're doing a great job of promoting the idea of something intractable and untouchable, but we've been working on what we call meaning-based computing for 10 years. And the principle hasn't changed that much, which is it's the user that's changed. The principle was that human information is something that we humans understand because it's based on context. So for us, you're geeking out on technology, somebody who's in our world can sit in front of this recording and say, I get these guys, I know the context. But my wife can sit there and look at us like we're mad because she's context-less. So today, this is still the case. So today, reasoning is all about the context of information, but more and more it's understood by the guy with his phone is the mini analyst. He can see information and demands that there's context. And interacts with it. Correct. The other part of this is the whole question and answer thing, the conversational software. So we all want to talk to our phone and get an answer. We never get the right answer, but we get an answer. We all demand that. So what I think is going to be very interesting in Christian Chippen as well, is that the evolution is demand-driven. So now we use consumers at demanding the machines understand or applications in your application, which we were having a look at. And it's very interesting. We demand these things speak to us and do something more useful to us than just give us it. So what we're obsessed about in the big data division is of course the backend tools that allow us to shift information in the trillions of objects and all that jazz, right? Understanding the data in whichever size, volume and variety, but we're really obsessed about how is it going to be delivered to the same user who is demanding that it does more for them. And there is where, you know, I think everybody has to do a lot of work. Watson has some work to do. We have work to do. Well, you nailed it. I think this demand interaction is interesting. This is where real time comes in because but Dave was here. Dave would be like all over this because we talk about real time. What does that mean? If you're crossing the street and you get hit by a taxi certainly in Barcelona, you might, but that's if you're like off by a second, if you're getting bad data and you walk out in the middle of the street, you're dead, right? So real time is interesting. So near real time, real time, that's a big part of this. The eyes of the beholder, right? So you're the consumer, you got the phone, you didn't get that answer when you wanted it. Is it real time? It doesn't matter whether it's real time or not. It's not right for me. It's the context. You nailed it. I love that. So let's drill down on that. So what does that mean to businesses? Obviously it's a spectrum of demand on real time. So that's why near real time, it's good enough. It's close, you can talk about a grenade or a bomb. It's close enough, okay, it still kills the target. That's my old analogy. But when it comes down to the actual platforms, what is the pattern right now that you guys see from a current state of the art? Because you got visualization, visualization is cool. Around the real time and the BI, this is the new market. What are some of the state of the art things that you guys see that have crossed the chasm in customer minds, like okay, that's cool. I don't think we still know the answers to even the questions half of the time. So if you take real time as part of that program, right? We all want real time in the sense that I think we all want instant gratification, right? We all want the answer now. But we, again, I'd love to see both of you use on this, but it feels that we don't in many cases know the question to begin with. So we're in a bit of a strange place. For real time stuff, I think more than real time, what businesses want is that they want to create this year 10 applications on their data as real time as possible, whereas last year they were building one in the whole year. I think the demand is I want to create five apps on my data, because my data's rich and has all the value. And I want to do it be as fresh as possible because the value of information is lower and lower and lower and lower, right? So cut the time back up and commission it. And that's what they're demanding. So these things, the on-demand platforms, the pay-as-you-go platforms, the rest-based functions where everything is the pure value and not the geekiness of how the face recognition works. That's where these things come in, because you're telling the guy sitting in his bedroom or in the company, hey, I'm going to do all the heavy lifting for you. You just have the idea. We've got some, Chris can talk to a couple of great examples of people creating apps. We'll talk about Sparky for a bit, have a go at that. Let's catch that point, because I just made a note of that on the crowd chat, was one, the app evolution is going to be data driven. So a data fabric has a development kit. Data as a development tool is a new concept. Not to us, but to you guys, not to you guys, but to the world. Oh, I can actually program with data as an input. And two, the third party developer, this comes in to more of a development platform. So when Robert Young-Johns was on last year, we asked him that direct question, what's your development strategy? Because that's huge, right? If you can enable people to build apps. So can you guys talk about that? What's your experience just in the industry, with the data and your platform? Is it the customer's going to build the apps? Certainly, I agree with you. I think apps is going to be a tsunami of apps, at some point, once they get out with their virtualization. Well, Chris, you build those development platforms. Yeah, I mean, you talk about real-time. In many ways, it's not just the real-time of being able to ask the query, it's the real-time of being able to change what I'm asking. So being able to get the business to the point where I no longer have to have a six-month project to introduce a new application or introduce a new project. We've had some great examples where we've had both enterprises and the indie developers out there developing some great applications over the course of 24, 48 hours and starting to ask new questions and being able to evolve much more quickly than the traditional model of kind of being three months behind. So your business is real-time, not necessarily in the sense that it returns the answer in 10 milliseconds, which we do have a few customers doing, but more than, okay, I know I need to do something new. I can do that in a week or two weeks rather than in three months. So we've got some great examples, whether it's a Mexican startup who within two weeks had built an application to do social media monitoring for parents and started presenting it to VCs to get funding. How good is that, right? So we're looking whether our kids are being bullied in Facebook, come on. And you want to kill that fast if it's not going to be a business or you want to push it if it's going to be a business. You don't want to waste six months coding to get that out. You want to get it out in two weeks and see if it's got traction. Yeah, exactly. It's the show me now. Now it's the business and I think bosses in enterprise are going to turn around and say, I've had a great idea, which is so-and-so. I'm going to take this data and make a sign. And they're going to say show me, show me. Yeah, I just want to post on this on force called the marketing cloud, evaluating Oracle. They actually have a good, that's not fully there yet, but what you guys are doing is the same thing, which is I totally agree. This notion of upfront licenses is gone. I mean, it's going to be like, show me, I'll do a little buy here, I'll taste. And if I don't like it, I'm not buying. But so, okay, if I like it, yeah, I'll take more of that, right? I buy the drink, whatever you want to call it. I mean, it is- You show me. I can say it wasn't a philosophy. I've actually tried it out, the technology. I've created a little UI, which I can show the principle and off I go. Well, there was a CIO telling us the other day that he's changed his RFP process. So it now includes a hackathon. So they run a hackathon for 48 hours with the software and see what you can produce in 48 hours. And he says it tells you so much more than the traditional process. We should talk about hackathons because for us it's been, I mean, we come from enterprise world, right? We sell enterprise tools and we find ourselves in the place where we're thinking, actually, if you open up the Kimono, right? And you open up the tools and you put them out there and you ask customers, you know, spend 24 hours with your people and they just, rather than just sitting in a room idealizing and doing things as well, actually just use the platform to have some mail and get the ideas going. You know what the outcome of that is? Applications. Yeah, and also loyal things. People, it's like, it's like cars. I mean, you drive a good car. You're like, I like this car, right? I like this to be good at cars. Let's continue with that. Yeah, I mean, you know, this is, but this is what it's like. I mean, this idea of locking, grab by the throat, the old lead gen. They are like, oh, here's a six month migration path and a million dollar upfront license. And the market's moving too fast. And I think the SaaS thing's interesting. I want to get your take on this because I have a lot of tech friends like you guys that are out there and we're doing the same thing where SaaS is the ultimate iteration, DevOps for entrepreneurship. You can put something out there fast and get instant feedback. And then iterate quickly, agile, lean, start whatever you're going to call it. But most entrepreneurs don't actually go out and do the SaaS business model in parallel. Meaning, if your iteration cycles on the tech is fast, you have to understand that it's a land and expand business model, which means if you do land and have to expand, you got to think that out. So if you do get successful. I see what you're saying. So I see a lot of guys that still don't know where I live, it's like, dude, you have a land and expand leverage model that if you do land and grow, you got lightning in a bottle, you've got a rocket ship. So what's your customer average of the cost? I don't know how to, what would I sell it? I'm excited. Don't you see the opposite model as well where you have the old problem we have, you and I created that problem like you guys created, right? And you've got an element in the old world that you've got IP that sits in the back to solve. They'll have a lifting problems and you have a delivery part where you're creating something the user might want to touch, right? What these guys see now is they don't have to worry about the creating the IP side. So you've removed an entire part of your delivery and how you're going to expand it. The only thing you're going to do is say, look, I'm going to create the UI which is compelling, you know, the average gal or chap that's going to use it going to love it. Oh, I've got a thousand users. I start paying. I don't have to worry about creating IP. Well, I think that's a huge point. So here's my take on that, right? So there's two levels of land grabbing, right? You know, the more IP you build that does heavy lifting, if it's good heavy lifting that has a market, you'll do make more profits. But what you're bringing up is on my belief the monetization of the first, a real new monetization opportunity for developers. And that is if your platform does a lot of heavy lifting, if I'm an entrepreneur, I can shed that off. I'm essentially a reseller partner in a way. So if you take the old business model concept, what is a reseller? They're front-ending someone else that has reseller selling gear. So the developer model is interesting. I can be an expert at a UI or interaction experience, whether it's voice, you know, drones or something else and provide real value that is worth potentially a very big business. Let me give you something that blows my mind, which is that, so we look at our on-demand platform. On-demand platform is all about not only about the heavy lifting that we do, all the stuff we do with structured data and human information, video voice, it's not only about what we do there. So one of our philosophies is to be able to give our partners, our developer community, everyone, the ability to expand that platform. So find yourself in a situation where you look at the 50 APIs that we provide today, as well as the SQL access, right? And you, as a developer, can say, actually, I've written a bit of technology that is core IP and I want to surface it through you as a bit of functionality, as a risk. It's not a marketplace, I just, I want to use it together with your stuff. That's where we're going, where it's going to be, hey, if you got something valuable, add it to the name. We were talking before you came on, before we went on about what we're doing, and more importantly, my pragmatic view is simply this, and you said it, Fernando, use case. The business innovation actually is IP, right? Some people patent processes, I mean, Amazon patented one click, I mean, who let that happen? I mean, come on, one click, everyone clicks once. It's not really a unique idea. But they branded, they trademarked it, but there's little innovations like that that I think are going to be huge. And just this data world's so big, like just your command center stuff, the haven on demand, so I just think it's really early. I think we're going to look back at this time and say, hey, remember when we had one app on big data? And now- And the business innovation is so critical. So what is growing my mind is we create all these amazing back-end systems that do heavy lifting, we provide the platform, and all these guys come along, and every time we do a hackathon, we look at the ideas of people when you sit there, we say there, you can't make this stuff up, because back to your car analogy, the car manufacturers build mules, right? Which is some strange-looking thing that has different wheels and an engine thing popping off the top. These guys are building a real car, it's just not very polished. And the use case is amazing. They're making boat cars, right? They're doing all these things that you wouldn't have thought of. So what's going on with you guys? Tell us a story about how you guys started the product, how it all came together, some of the challenges, things you've learned. I can start a little bit and you take it over for yourself. I'll plug it in when he starts, right? So we were, when Robert, young folks, we know, right? Robert joined us whenever that was. Robert came and said, look, I've got a dream that the world is going to consume information through services, and we're looking, yeah. That sounds like it, that sounds like it, that sounds like it. We're on board with that, yeah. Yeah, we get that. And he literally said, guys, why don't you build me, take some of the cool functionality of idle, for example, to give a face recognition, or I can't remember what, you probably know this. Give me a couple of these functions just to have a play. So suddenly we find ourselves with a couple of functions. That was the first, dude. And then we containers start, and now they're going to let you take over from there. Guys, I love containers. Suddenly it's a whole brand new concept. What do we do with this? So what did we do with this? I mean, the big problem we had is just, where did we start? There's so much technology we have in the Hayden platform. Do I want to do NPR? Do I want to do face detection? Do I want to do conceptual search? Do I want to do, it was, where did we start? So one thing we realized quite early on is we needed to build a platform that allowed us to basically democratize people adding in those extra pieces. So we could say, okay, image experts, you add in your pieces, and here's just the framework we're going to deliver in. For onboarding new tech? Yeah, for onboarding new tech. Okay, and scaling it or not? Okay. So we built a platform that took care of us handling, okay, who's coming in? What do they have access to? Am I allowed to do this? Are they able to pay? Do I have voters? Do I, how do I distribute the data? And basically we've got a framework which we call workers or kind of like little containers where the image teams, the experts who know how to get that face out of that image can concentrate on building their piece and packaging it up and delivering it up in a standard format so we can turn just exposing it. And that's what's allowed us to do this pretty much in amazing time. You basically built the Docker model internally for yourself. Yes, Docker now is all over the place, but for us, here's the- Containers have been around Pat Gelsinger, he was like, Pat Gelsinger. Oh yeah, we look back at Linux, sorry, at some of the Unix products and containers and some of us old timers go on here. Rapper, container, whatever you want to call it, it's been there. So we found ourselves using OpenStack of course and Helium a little bit later and we thought to ourselves, this whole container thing sounds like fun but I tell you what, we kind of need them to talk to each other because in data you start doing some analysis here and you pass it on to this and you do some outcome here and all these things need to talk to each other. So we found ourselves creating this worker fabric which is, I think it's working, sorry. So at that point, I think that is when, on demand is born. On demand is born the moment that we don't have a few REST functions for some servers and we have a worker fabric where these things talk to each other for years. So persistence versus non-persistent, you don't care. Both we have. Communication, say I want to build something on social media and I'm getting 100 tweets a second, I can't be doing that in an inefficient way. I need to pass that data, perhaps I want to do entity extraction, then I want to index it, perhaps I want to do some notification. The data's got to flow through the platform, it's got to know where to go. It's basically, it's like, you've got to build that bus. You've got to build like a sort of bus for containers, right, you've got this bus for containers that moves the data around shopping. Some data motherboard. The data motherboard, that's a good way. I'm going to print that out, you know, take a note. Data motherboard. I mean, you have people talking to each other, you've got a backbone, you've got a backplane, you've got to connect it. And we don't do it once. This is something that anybody can do out there and say, I'm going to use 10 of your APIs and I don't care how you do it, but these things need to talk to each other because that's the demand. There's no policy in there. You've got to break it. Whatever it is, whatever combination. So that ends up being on demand. As well as we have our vertical on demand as well and then the problem is complex as well is how do you have a warehouse in the cloud that can do everything that we can do at a scale that we can do, right? So what about developers? What are you guys doing right now? So, I mean, this is not a business question, but more like a technical question. Is the focus on developers? This is for them. This is for developers. On demand is for developers. Whereas Vertica on demand is obviously our warehouse product. The idle on demand piece is for developers. If you go to the site, I'll put a plug in there, www.idleondemand.com. Idle on demand.com? Idle on demand.com. It's purely for developers. It's something that they can immediately get a logging, immediately start looking at a community, immediately start playing with the functions, do tries. They can create an application. I mean, and this is as simple as that, but the focus when you have it going and you're putting it on your screen there is you'll immediately be able to see that it is built for developers. There's no marketing, no, I'm going to say, no BS, right? As they would say, right? No, no executive. No marketing speech. So I'm going to view the APIs, here's the parameters on that. And it's also about building a community for those developers. So having the forum with them to connect to each other, having competitions for them to compete against them, ways to promote their applications. So we basically want to, at this point, we want to hear what they're doing. We want to hear their feedback. And we want to, if they've got a great idea, we'd love to hear it and we'd love to promote it. We're obsessed about Net Promoter, so everything we do is Net Promoter. We're also obsessed about being honest and we put things out there. And sometimes we even say, we're putting this functionality out there. It's kind of half-fake, we might break it, but tell us if you like it and then we'll kind of push it through. So I mean, open source has cleared the way for the new model for software development and that's collaborative, open, out in the open. The minute you got to start body-swerving and head-faking, you're dead. Well, we do open source a bunch of stuff, by the way. So some of the UIs, we just open source. The work are infrastructure- It's about transparent value, right? If you have a platform, people want trust, they want credibility, they want to know that you're going to be around, right? They don't want to waste their time on a platform that's not sexy or cool. And Amazon's cool today, and HP's working on being cool with Helium. HP's cool, man. HP's cool. What's wrong with you? HP's wonderful. And cool. Of course. We want them to know that HP is going to take all this amazing functionality, serve them. We're going to be here. We're going to be here. It makes 10 years, 20 years. HP's not going away. So they can rely and build their business on it. That's all we want. And I tell you, I'd be in a blue mix, guys. I say the same thing, guys. I don't hear engineers saying, oh man, I can't wait to up and blood my app to blue mix. Just that's not the vibe. You got to earn it. You have to earn the receipt. We all have to earn our stripes. You got to earn the stripes, earn the respect. And I said, it's definitely doable. Why not? If you have distribution, I'd be in, right? You guys have- We do as well. You have technology you're sharing, right? So the gifts are all about, it's about the gifts. At the same time, make a good environment, cool people. The premium is there. They can get on with it. The transparency is there. They can engage for themselves. Absolutely. You guys have a good group. I'm very impressed with Robert Young-Johns. Obviously George Kedifa, big fan of him. He's now doing some other special projects. Chris Selland over there at Big Data. You guys are doing great work. Always been a big fan. I'd like to see that action happening. But I want to give you guys the final word on culture. Share with the people out there. One of my favorite questions. What's the culture like? You guys are cool. You're super geeky. You're polished. What's it like? What's it like inside the culture? What do you guys do? What do you guys stand for? What kind of people do you hire? What do you guys do for fun? Share with the folks out there. What's the... I'll have a go. So we're the kind of place where you build things like Idol on Demand and Vertical on Demand. We create new stuff. But we understand the legacy of what we do and the importance on our customer base. It's a fun place. We're very relaxed. We like people that are creative and smart. I wouldn't categorize it in any way. The top, the bottom. It's people that are smart, that like to create, that have respect for customers. We're very, very customer focused. But we're good fun. And you guys solve hard problems. And we're smart. We wear jeans and... Big mothership behind you. The aircraft carriers, Martin Miko said. But really, to me, I think the challenge has always been whether you're talking to any engineer, it's always about having fun, solving big problems and really being relevant. That to me is like, and you got all the action in front of you. You guys are like I said, the facial recognition and some of the confidential stuff you told me about operationally is fantastic. It's great, great. Well guys, thanks for coming on theCUBE. This has been a CTO conversation here inside theCUBE. It's been great talking big data. I love it. I could do another hour of it. Now it's going to get ready to wrap down day three. We'll be right back. This is theCUBE live in Barcelona. I'm John Furrier. We'll be right back here for this short break.