 The Cube at Hadoop Summit 2014 is brought to you by Anchor Sponsor, Hortonworks. We do Hadoop. And headline sponsor, WAN Disco. We make Hadoop invincible. Okay, welcome back everyone. We're here live, this is The Cube at Hadoop Summit, our flagship program. We go out to the events, and strike a signal from the noise. I'm here, I'm John Furrier with Jeff Frick from SiliconANGLE in The Cube. Our next guest is Oliver Ratzberger, an SVP software teradata, welcome to The Cube. Thank you very much. So obviously the big business and big data is all about outcomes, right? And we were just talking on our intro, and you've got three actors in this ecosystem. Startups, jockeying for position, rubbing nickels together, trying to make things happen. You've got the series seat, the pre-IPO, then you have the big whales, right? We have existing legacy businesses and customers. And all three are working together, but at the end of the day it's about business outcomes. So the question is, how serious is Hadoop in the mind of the customer? And what's the reality? I mean, you guys have an approach to teradata that you've been serving customers for years, but now Hadoop pops in, and it's now talking business outcomes for the first year you're hearing that at Hadoop. Last year was SQL for Hadoop, now it's like business outcomes. So comment on that trend and what you see. So I think it's very important to focus on business outcomes. In fact, I spent the last 10 years of my career integrating technologies like Hadoop and Teradata and others to drive business outcome because ultimately it needs to make a difference in a company and you need to build a platform that is agile with data and that produces outcome that actually changes what companies are doing, right? If you just create cool visualizations, if you just run cool queries, that's not really meaningful, right? The executives need something that helps them to make a change, to make a shift, to adopt to a very fast changing market. The consumer market is moving so fast, right? With new devices, with new capabilities, everything is real-time, everything is self-service, and if you are not good with data and with analytics, you can miss the ball very quickly on that. You know, I was talking yesterday about the iOS announcement, it was great, everyone was talking about Apple and they do an event, it's the pomp and circumstance, but iOS 8 is very much a consumerization of IT enterprise, so that teases out certainly the vector we've all been seeing in mobile, but I got to ask you, what pressure does that other forces put on the CIOs and infrastructure? Because now, it seems to me, and I want to get your comment on this, that the pressure to move faster is more than ever, and I want to, what does that mean? The pressure to move fast, but a dupe and the big data infrastructure from the software, what does that mean to move faster? I mean, is it like weeks, months? I mean, what are some of the timetables and characteristics of that? No, this is great, so talking from personal experience, moving fast means hours and days, right? Weeks is already, yeah, you are too slow, and we heard it earlier today on stage, it's like agile is not fast enough, right? Now, you have to see, yeah, you have to be careful about that though, right? Because it very quickly turns into wild, wild west. If you just say, oh, I'm going to do big data, I'm going to throw away governance, I'm going to throw away integration, I'm just going to throw data out there and see what happens. That works very quickly, very shortly, but six months into it, 12 months into it, you are all of a sudden presented with a real problem, right? So, building an organization for agility, I think is the biggest thing that CIOs and C-level executives really have to do. And so, you have to start with an agile environment first and introduce a big data on top of that. You can't assume that I do big data and all of a sudden I'm agile. It's usually the opposite, right? And designing an organization for agility is actually very, very hard work, right? And in many companies that I've been talking with over the last years, I find that they want to be agile, but they fall into this waterfall methodology all the time, right? They want to see roadmaps. They want to see, you know, when can I release this product? And they haven't realized that there is a place for agile and there's a place for waterfall and they need to live right next to each other and it's no different with big data than it was with small data, but you need to really focus on that and you need to train your people, you need to shape your organization, you also need to establish funding models that support that. Yeah, the process is also great, but you also can slow you down but you want the agility or super agile model to coin that new term. But one of the things that's interesting is living in Silicon Valley and certainly in theCUBE, we've been here from the beginning so we love to separate the hype and BS from the reality. That's kind of, and it's fun to see the hype because that means people are hustling and trying to make things happen, certainly in the startups. The one telltale sign this year that happened was at the Facebook Developer Conference. Mark Zuckerberg changed his flagship mantra, move fast, break stuff. Now, the DevOps culture certainly influenced Cloud and Hadoop, right? It's clearly that e-glass, spit nails kind of mentality. But now you have Facebook move fast, break stuff to the new slogan, move fast with stable infrastructure. That to me is an admission. You have to. This is serious. So comment on that and how that translates, because that's motor stop horrendous for the enterprise, isn't it? And see, you need to find that healthy mix because move fast, when you are very early startup, right? When you're very new into a market, you can break things very early on but once you're established, people will not accept it. Facebook users won't accept if their profiles are broken or if their security is broken or whatnot, right? So you can't make mistakes. Same thing with executives. If the financial numbers of the companies change every day because somebody's agile with financial numbers, that's not good, right? And so you need to realize that you don't build a nuclear power plant in agile. You don't build an airplane in agile but you build certain new products in agile, right? And certain new features. And I think companies that master that, that realize that I need governance and stability for certain parts, it's a spectrum. Agility is a spectrum. It's not an on-off switch. It's not like I'm going to jump and now everything is agile, right? It goes from well-defined to very agile and you need to build the capabilities for that. You need to build platforms for that. You mentioned DevOps. I'm a very strong believer that DevOps needs to come much, much more into big data and into the platforms. In fact, when we look at the unified data architecture, how we call it a third data, DevOps is a big topic for us because right now analytical applications are lacking the kind of DevOps that the world has developed around cloud solutions, right? And I think that's the next big thing that will allow even more agility with data because it can't be that operations teams fight development or business user teams because they want to be agile. And today that happens way too often. So talk a little bit. You were on the customer side before you moved to Teradata. So now you've got that perspective. You're fresh on that other side. Yes. Talk about two things. One is how high up the command chain does this push for agility as well as stability need to come? I mean, are CIOs getting it? Are the old school guys figuring it out or does it have to come from even higher up the chain to push that? And my second question is, you said you worked at Sears before. eBay obviously is a data company. Sears, you know, it's still a retailer. How are you finding it with your customer base that they're figuring out that they are actually all information companies that just happen to wrap it around a particular product or service? So to your first point, I think that the push for agility needs to come even higher than the CIO. The CIO is one of the enablers, right? And so is the CMO, but ultimately it needs to be a goal of the CEO and even sometimes of the board of a company, right? To really drive for rapid iteration. Now, they also need to realize that it's a balancing act, right? That you need to have good governance where it's important, right? You can't close the books of a public trade company on just some Excel spreadsheets or some data leak data. You need to have good integration for certain things. On the other side, you need to be able to prototype new ideas very quickly. One of the biggest problems that business communities, the CMOs, the pricing guys, the supply chain guys always have is when they have an idea and they go to IT, what does IT pull out? That try that requirements document. Go right down what you need. And at that point in time, the executives of the business units, they really don't know really what they want. They have an idea, they have a vision of where they want to go. Now they have to write a vision down into a binding contract, right? And so what happens? Big projects happen. I need all data in all possible combinations with all reports possible with all algorithms. And you end up going in this 18, 24 month cycles. And by the time they're done, two things usually happen. The first thing is like, well, that was interesting last summer. This summer I have a completely different macroeconomic environment. I need something completely different. Or they just go swipe in Amazon account. Yeah, exactly. Or they look at the data and they say, well, I didn't expect that. Now I have 50 new questions, right? And so this for product development around data, that hasn't worked. And I think that it's very important that the CIO is a big driver in that, but it goes beyond that, right? And what I've seen, and you talked about this years, what we have done from within the analytics side is we pushed very hard for agility. But we also realized we need to do it with a purpose and with a methodology. You don't be agile by just throwing out documentation or by just landing data somewhere and hope that the person who landed it there is still there when you need that data. And so what we did is we actually took the entire development community and got them through scrum training. So we said, you know, might as well pick one methodology, we picked scrum, we did that, but we also included business units into this, right? And we picked a couple of Lighthouse customers on the marketing side, for example, that were willing to go with us through that training, that were also willing to co-locate their people with ours. So they were sitting in the same conference rooms in the same cubicles right next to each other to really work next to each other and not have that traditional separation of, here's IT and here's the business unit and there's this scrum and stocking firewall in between, right? That slows everything down. So how did that work? What were some of the interesting outcomes of that process for your own internal development? So what we've learned over the years, that started many years ago at eBay. So I was seven years at eBay and I think as early as 2005, we started realizing that agility with data is paramount. We need to design for making that happen. And so one of the concepts very early on that we developed was we realized we need to train people on Agile, but if you train somebody on Agile and then they need to open a ticket and wait 14 days to get a response, it falls apart right there, right? So self-service. And so this is where the consumerization comes into it. This is what we learn from the Apples and the Facebooks. It's like self-service, self-service, self-service. And so the first thing that we built at eBay was the concept of virtual sandboxes where you would come to our website and we would have it as an internal product and you would say, I want my own sandbox and we had an SLA of less than five minutes, you need to have your sandbox and you need to have access to all the data that you're allowed to see, right? It needs to have security, it needs to have governance wrapped around it. But we made it very easy for people to try something out right there on the production data, right? And so that was an important start but we also started realizing that governance and time boxing are very important principles. So over the years we realized that to be agile, you need a methodology, you need concepts like time boxing. If something takes longer than 30, 60 or 90 days, it's too big, right? And so we have done some of the largest projects at eBay as well as at Sears, where we said, you know what? Let's do them in 90 days. And at first it's like, oh my God, that is crazy, right? You can't do this in 90 days. You can even commission the boxes in 90 days. But the point here is that it forces you into a different mindset. It forces you into saying, you know what? I can only spend 20 days off the 90 days on performance tuning because I don't have more time. I don't have the luxury to go for a year and re-architect things because I won't get it done. But it allows you to move very fast and very rapid iterations, learn every step, fail fast, right? Put something together and say, you know what? That's not working. Park it somewhere or throw it away. Because when projects get too big, the hurdle of accepting failure with data is just too big. Once you have sunk in $3 million, it's very hard to walk away from that and say, you know, that just didn't work out. And so, very important to bring these things together with methodologies, with platforms, with capabilities, like sandboxes, like self-service tools, and with online collaboration. I mean, one of the things that we also learned and we first built at eBay was, we built what we called the data hub. An internal crowdsourced social website, think of it as LinkedIn for analytics where people would follow analysts, would follow data sets, would follow visualizations, right? And we would present to them what other data sets are there, right? What are other people doing with data within the organization? To really take out the structure of a company and connect people at different levels. And we had like marketing people, all of a sudden realizing that finance people weren't working on the same problem, but they used different data sets because nobody ever told them that they were looking at the same thing, right? So, there's a lot more that needs to be built and the technology choice of, you know, how you land data and whatnot, it's just one part of being really agile and making that happen. Let's talk about the reality of the customer and what you guys are doing. We had Merv Adrian on yesterday, we were talking about, Merv, what benchmarks do you say are, do you need to see when you say this is a mature industry? We always talk about that, right? And he said revenue. So, we hear a lot of fun from the startups and from everyone else that throw in fear and certainty and doubt about who's better than what. But you guys have been criticized and from a flood standpoint as, you're out of touch, you're too slow, you're a big player, had dupes about nimble and agile. And others have been called that too, by the way. How do you answer that? Because at the end of the day, you guys are producing revenue. I mean, MAPR's been criticized, oh, they're going to go out of business. They have hundreds and 500 customers that get more employees than most of the startups. They're just executing, right? So, at the end of the day, when you look at the execution, talk about what Teradata is doing from a standpoint of relevancy and share with the folks out and clear the air on that whole flood feature. Well, so, there's a couple of things. As you said, it's all about execution, right? It's ultimately to a CEO, to a CIO. It matters if you can produce tangible outcomes and results, right? Playing around with data is one thing, but you really need to bring some business results together. What we do at Teradata, I mean, and I'm to some extent a testament of that, right? Teradata hired me last year being one of their most influential biggest customers for the last 10 years to bring that perspective of we need Hadoop, we need Teradata, and we need other technologies. And it's a best of breed integration. It's not a one-size-pony that fits everything here, right? No lock-in. And bring that... Bring your solution and you drop it in in a multi-vendor environment. Yes, and so, we're driving a lot of that new thinking into Teradata, right? And we are constantly thinking about what is the next step that we need to solve for our customers? Also, what are the capabilities that we see that come out of some of these communities that are good, but that are just not enterprise-ready? How can we turn them into something that a CIO can turn on and within a couple of weeks can have it in production, right? Let's be specific. Give an example, because we know that it's not enterprise-ready yet. You saw the acquisitions on Hortonworks and then Cloudera followed with their security M&A deals and that's clear. They're not filling a hole, it's just evolution, right? They're actually moving as fast. They're peddling as fast as they can, as you said. So that's clear. So Hadoop is going mainstream, across the board. Give some examples of what you guys are doing with customers to kind of set the stage in, you know the name names, but like use cases and kind of some specific examples. So what we are seeing a lot is the integration between the traditional data that you find in something like an enterprise data warehouse and new types of data. Whether it is image processing, whether it's natural language processing, whether it's sensor data, log file processing, right? Where it is very powerful to have a processing platforms like Hadoop, right? To, for example, do image processing. To give you an example, eBay at any given point in time has probably, I'm guessing right now, more than a billion images up live on their site, right? Now there's very interesting use cases around that, right? There is use cases around fraud, there's use cases around better search, right? If you look for a red pair of shoes, I want to see similar ones, right? And so you need to process images in a way that traditionally you haven't done with an environment like a data warehouse, right? Sensor data is another great example, right? I mean, new experiences, right? These are not like, these are net new outcomes. That some companies like eBay have already done over the years, of course, right? There's other examples where, for example, right now we're working with a train operator in Europe and we're looking at sensor data to help them predict before their locomotives break down because that's a big problem, right? If a locomotive breaks down somewhere on the track, that's a big cost and delays and whatnot, right? And so it turns out if you take the sensor data and if you align it with the service records and with everything else that you're doing, you can not only do predictive maintenance, you can actually predict with a very, very high percentage within the next 12 to 24 hours when something is going to break down. So I want to get your perspective on something. We talked about the Wikibon survey that's coming out. Jeff Kelly was teasing it out, it's going to come out in the next week or so. Look for that survey, it's really awesome. 800,000 sample down to 300 direct inquiries with targeted customers. One of the questions was this, what is the primary driver of big data analytics projects in your organization? 20.13% said to save money, quote, by improving operational efficiency. 23.43% said to increase revenue, 54.79% said both. Do you see that and what does that mean from an adoption standpoint? Is that a mindset? Is that part of the evolution? Can you comment on what that means from a customer standpoint? Obviously they want revenue, you talk about that. They're like, but they want both. It has to be both. I think it has to be both. And again, my example from my experience at eBay, it was always both, right? We built solutions looking at sensor data on how to make our data centers more efficient because we knew any dollar that we could save in a data center would drop right to the bottom line of the business. At the same point in time, we built solutions to see how do we attract or convert more customer traffic, right? How do we experiment? How do we make it very easy for a product manager to have a new idea and try it out? So at eBay, we built an experimentation platform that allowed us to say, you know what? I'm not gonna just do AB testing and we heard that earlier today on stage as well. We do a through C testing. We try multiple different variations, right? And we pick the ones that work the best for the customers. And so I think it has to be both. It has to be top line and bottom line because there's so much opportunity out there, right? Whether it is data centers for somebody like eBay, whether it's supply chain for some of the very big retailers or whether it is, it's systems like turbines or whatnot from the likes of Siemens that really need to optimize that at a global scale, right? There's millions of dollars in OpEx, in cost, but also in revenue attached to some of these opportunities. And we're clearly seeing this going in both directions. We saw the true car guy on yesterday, amazing company, they're doing big data, they're doing all kinds of new insights that they're getting that they've never seen before, right? So this notion of operational efficiency. Yeah, we've seen that before. Okay, we know what that means. Consolidation, do more glass, blah, blah, blah. Plastic enterprise, you know, talking point. Driving revenue, okay, that's cool. Business units who want some demands, but doing both means a discipline. What technologies are out there now that allow customers to do that and what is needed to get the CIO and then now the CEO, as you mentioned, Rajao, driving both of those agendas? So I think this is where this unified data architecture really comes into place. You need on one side that the scalable, manageable enterprise data warehouse where financial data, where service data, where whatever can really come together. You need on the other side, very tight integration with an unstructured platform like Hadoop, right? To do the different types of processing, to land the different types of new data that are coming in, but combine them. And when you use that, we just recently released the query grid, right? Which allows us to now do push-down processing. If you come into Teri, you can run queries and without you even knowing, half or two thirds of the processing can actually happen on a Hadoop cluster because that's where some of that data sets, right? We bring that in memory. We make that very easy. What used to be data movement, ETL tools, copy data from one system to another with tickets and firewalls and slow, right? Is now on the spot, write a query and it goes and reaches out. And it's not just Hadoop, it's other technologies. It's SaaS, it's R, it's Storm. I mean, there's a whole ecosystem of technologies. It's a reconstruction. Yes. I mean, the way it did it warehouse is not that fenced out organization anymore. It has to be central. But the thing I think is so powerful is really this everyone always wants to know, how are we more innovative? How can we drive more innovation in an organization? But really what you're talking about and what a lot of CUBE guests talk about is by enabling a broader set of people to do more experimentation with data quicker, it's failing fast. You just basically, by pure percentage, you've got that much more in, you're going to get that many more successes. You need to open the funnel, right? You need to open the funnel for new ideas. If your funnel of ideas is like three wide because that's the number of IT projects you can do in parallel, then you're stuck, right? You can't be fast. You need to get to a point where you can have tens of thousands of ideas within a company, and you quickly get to a bottom. What are the most meaningful ones? What are the ones that have the best chance to success that really make a difference? Whether that is for somebody like eBay, paid search advertising, or whether that is email campaigns, or whether that's a loyalty program for a retailer, right? It's really how to iterate and try out new ideas, because one idea might work today, it might not work today. And if you take too much time to figure that out, you're already left in the dust, right? I think you've already enabled a lot of you. Thanks for coming on theCUBE. We really appreciate it. We got time for a break here, but I got to say what you've done at eBay, what Yahoo has done, these early web properties were really genuine of web scale. Certainly, Facebook came in and established the DevOps modelist, then it was quickly, I was already in the web scale companies. That's going to the enterprise. So, we're excited, congratulations for all the great stuff. I want to give you the final word. Share with the folks out there in your own words the relevance of Teradata in context to this perfect storm of innovation. That's a great question. I think that Teradata, with its experience of integrating data at various levels, and especially now with new technologies that we integrate, is at the forefront of somebody that needs to put something into production that really makes it happen, right? It is what we specialize and focus on is like, we build and integrate the technologies so that you can bring them into your company, whether it is Hadoop, whether it is Teradata, whether it's AsterData, or the other surrounding technologies, and really turn them loose on your problem rather than dealing with the low levels of how do I install that, what patch levels do I need, what interconnects, what network cards should I have, how much memory, how many cores. There's so many choices to be made, and good choices, and if you are a very sophisticated company, you want to make those choices. But a lot of companies out there, a lot of the Fortune 5000 companies cannot spend time on engineering the same platforms over and over. And I think what Teradata brings the most to the table here is we do that engineering for our customers, we bring a pre-packaged solution into that, but we make it open to integrate with other technologies so that it's not a standalone island that is locked in, but it's an integrated ecosystem that allows it to move very, very fast. Oliver, thanks for coming on theCUBE. Teradata, one of the big players out there. Again, all three areas. Startups, growing growth companies, and public companies all doing some great damage here, and great stuff. Business outcomes, serious discussions here at Hadoop Summit. This is theCUBE, of course, we're extracting that signal, sharing that with you. We'll be right back after this short break.