 DataWorks Summit Europe 2017, brought to you by Hortonworks. Okay, welcome back everyone. We're here live in Germany, in Munich for the DataWorks Summit 2017, formerly Hadoop Summit, I'm John Furrier. My co-host Dave Vellante, our next guest is Mike Merritt Holmes as the Senior Vice President of Global Services Strategy, a Think Big and Teradata company, formerly the co-founder of the Big Data Partnership, merged in with those Think Big and Teradata. Mike, welcome to theCUBE. Thanks for having me. Great to have an entrepreneur on you, the co-founder, which means you got that entrepreneurial blood. And I got to ask you, you know, I'll see you're in the Big Data space. You got to be pretty pumped by all the hype right now around AI, because that certainly gives a lot of extra, extra steroid recognition. People love AI, and it just gives a face to it, and certainly IoT is booming as well and it have things, but Big Data is cruising along. I mean, it's a great place to be. I mean, the training is certainly going very, very quickly right now, but you know, the thing for us is we've been doing data science and AI and trying to build business outcomes and value for businesses for a long time. It's just great now to see this really, the data science and AI buzzword really starts and take effect, and so companies starting to understand it and really starting to really want to embrace it, which is amazing. It's inspirational too. I mean, I have a bunch of kids in my family, some are in college and some are in high school. Even the younger generation are getting jazzed up on just software, right? But the Big Data stuff's been cruising along now. I mean, it's been a decade now of really solid DevOps, culture, cloud, now accelerating, but now the customers are forcing the vendors to be very deliberate in delivering great product because the demand for real time, the demand for more stuff is at a halftime high. Can you elaborate your thoughts on your reaction to what customers are doing? Because they're the ones driving everyone, not to create friction, create simplicity. Yeah, and our customers are global organizations trying to leverage this kind of technology and they are doing an awesome amount of stuff right now to try to move them from effectively a step change in their business. Whether it's kind of shipping companies doing preventive asset maintenance or whether it's retailers looking to target customers in a more personalized way or really understand who their customers are, where they come from, they're leveraging all those technologies and really what they're doing is pushing the boundaries of all of them and putting more demands on all the vendors in the space to say we want to do this quicker, faster, but more easily as well. And then the things that you're talking about, I want to get your thoughts on because this is the conversation that you're having with customers. I want to extract is have those kind of data-driven mindset questions have come out of the hype of the Hadoop. So, I mean, we've been in a hype cycle for a while, but now it's back to reality. Where are we with the customer conversations? And from your standpoint, what are they working on? I mean, is it mostly IT conversation? Is it a front office conversation? Is it a blend of both? Because, you know, data science kind of threads both sides of the fence there. Yeah, I mean, certainly you can't do big data without IT being involved, but for it's since the start, I mean, we've been always engaged with the business. It's always been about business outcome because you bring data into a platform, you provide all this data science capability, but unless you actually find ROI from that, then there's no point because you want to be moving a business forward. So it's always been about business engagement, but part of that has always been also about helping them to change their mindset. I don't want to report. I want to understand why you look at that report and what's the thing you're looking for so we can start to identify that for you quicker. What's the coolest conversation you've been in over the past year? I mean, I can't go into too much details, but I've had some, you know, it's some amazing conversations with companies like Lego, for instance, you know, just they're an awesome company to work with. But when you start to see some of the things we're doing, we're doing some amazing like object recognition with deep learning in Japan. We're doing some fraud analytics in the Nordics with deep learning. We're doing some amazing stuff that's really pushing the boundaries. And when you start to put those deep learning aspects into real world applications, and you start to see customers clambering over to want to be part of that, it's a really exciting place to be. Let me just double click on that for a second because a lot of the question I get a lot on theCUBE and certainly off camera is, I want to do deep learning. I want to do AI. I love machine learning, I hear all, it's kind of coming to reality. So people see it forming. How do they get started? What are some of the best practices of getting involved in deep learning? Is it using open source, obviously, is one avenue, but what advice would you give customers? From a deep learning perspective. So I think first of all, I mean, you know, a lot of the greatest deep learning technologies are in open source, as you rightly said. But I think actually there's a lot of tutorials and stuff on there, but really what you need is someone who's done it before who knows where the pitfalls are, but also know when to use the right technology at the right time. And also to know around some of the aspects about whether using a deep learning methodology is going to be the right approach for your business problem. Because a lot of companies are like, we want to use this deep learning thing. It's amazing. But actually it's not appropriate necessarily for the use case you're trying to drive on. It's a classic holy grail. Where is it? You don't know what you're looking for. It's hard to know when to apply it. And also you've got to have enough data to utilize those methods as well. So you hear a lot about the technical complexity associated with Hadoop specifically, but just all big data generally. I wonder if you could address that in terms of what you're seeing, how people are dealing with that technical complexity. But what other headwinds are there in terms of adopting these new capabilities? Yeah, absolutely. So I mean, one of the challenges that we still see is that customers are struggling to leverage value from their platform. And normally that's because of the technical complexities. So we introduced to the open source world last month Kylo, something you can download free of charge. It's completely open source on the Apache license. And that really was about making it easier for customers to start to leverage the data on the platform to self serve ingestion onto that. And for data scientists to wrangle the data better. So I think there's a real push right now about that next level up, if you like in the technology stack to start to enable non-technical users to start to do interesting things on the platform directly rather than asking someone to do it for them. And that, you know, we've had technologies in the BI space like Tableau and obviously the best debris data warehouse solutions like Teradata that have been giving customers something before and previously, but actually now they're asking for more. Not just that, but more as well. And that's where we started to see the increases. So that's sort of operationalizing analytics as an example. What are some of the business complexities and challenges of actually doing that? That's a very good question, because I think when you find out great insight and you go, wow, you've built this algorithm, I've seen things I've never seen before, then the business wants to have that always on. They want to know that it's that insight all the time. Is it changing? Is it going up? Is it going down? Do I need to change my business decisions? And doing that and making that operational means not only just deploying it, but also monitoring those models, being able to keep them up to date regularly, understanding whether those things are still accurate or not, because you don't want to be making business decisions on algorithms that are now a bit stale. So actually operationalizing it is about building out an entire capability that's keeping these things accurate, online, and therefore there's still a bit of work to, I think, actually in the marketplace still around building out an operational capability. So you kind of got bottom up, top down. Bottom up is the Hadoop experiments, and then top down is CXO saying we need to do big data. Have those two constituencies come together now? Who's driving the bus? Are they aligned? Is there still sort of a mess organizationally? Yeah, I mean, generally in the organization is someone playing the chief data officer. Whether they have that as a title or a role, ultimately someone is in charge of generating value from the data they have in the organization. But they can't do that with IT, and I think where we've seen companies struggle is where they've driven it from the bottom up. And where they've succeeded is where they drive it from the top down, because by driving it from the top down you really align what you're doing with the business and strategy that you have. So the company strategy and what you're trying to achieve. But ultimately they both need to meet in the middle, and you can't do one without the other. One of our practitioner friends was describing this situation in our office in Palo Alto a couple of weeks ago. He said, you know, the challenge we have as an organization is you got the top people saying, all right, we're moving. And they start moving, the train goes, and then you got kind of middle management sort of behind them, and you got the doers that are far behind. And aligning those is a huge challenge for this particular organization. How do you recommend organizations to address that alignment challenge? Does ThinkBig have sort of, you know, capabilities to help them through that, or is that sort of, you know, you got to call Accenture? In essence, our reason for being is to help with those kind of things. And you know, whether it's right from the start, so, oh my God, my chief data officer or my CEO is saying, we need to be doing this thing right now, come on, let's get on with it. And we help them to understand what does that mean? What are the use cases? How, where's the value going to come from? What's that architecture to look like? Or whether it's them helping them to build out capability in terms of data science or building out the cluster itself. And then managing that and providing training for staff. A whole reason for being is supporting that transformation as a business from, oh my God, what do I do about this thing too? I'm fully embracing it. I know what's going on. I'm enabling my business and I'm completely comfortable with that world. You know, there was a lot of talk three or four or five years ago about the ROI of so-called big data initiatives, not being really, you know, there were edge cases, which were, you know, huge ROI, but there was a lot of talk about, you know, not a lot of return. My question is, you know, has that, well, first question, has that changed? Are you starting to see much bigger, you know, phone numbers coming back where the executives are saying, yeah, let's double down on this? Definitely, I'm definitely seeing that. I mean, I think it's fair to say that companies are a bit nervous about reporting their ROI around this stuff, in some cases. So there's more ROI out there than you necessarily see out in the public place. But we're certainly seeing- Why is that, because they don't want to expose to the competition? Well, yeah, I think- They don't want to front-run their earnings or whatever it is. They're trying to get a competitive edge. Many of you start saying we're doing this. Their competitors have an opportunity to catch up. So- Very secretive. Yeah, and I think it's not necessarily about what they're doing, it's about keeping the edge over their competitors, really, over their competitors. So, but what we are seeing is that many customers are getting a lot of ROI more recently because they're able to execute better rather than being struggling with the IT problems. And even just recently, for instance, we had a customer of ours, the CEO phones us up and says, you know what? We've got this problem with our sales. We don't really know why this is going down. In this country, this part of the world is going up and this country is going down. We don't know why, and that's making us very nervous. Can you come in and just get the data, get a workout, why it's happening so that we can understand what it is? And we came in and within weeks, we were able to give them a very good insight into exactly why that is. And they changed their strategy moving forward for next year to focus on addressing that problem. And that's really amazing ROI for a company to be able to get that insight. And now we're working with them to operationalize that so that that particular insight is always available to them. And that's an example of how companies are now starting to see that ROI come through. And a lot of it is about being able to articulate the right business question. Rather than trying to worry about reports, what is the business question I'm trying to solve or answer? And that's when you can start to see the ROI come through. Can you talk about the customer orientation when they get to that insight? Because you mentioned earlier that they got used to the reports and you mentioned visualization, Tableau. They become table states. Once you get addicted to the visualization, you want to extract more insights. So the pressure seems to be getting more insight. So two questions, process gap around what they need to do process-wise, and then just organizational behavior. Are they there mentally? What are some of the criteria's in your mind in your experiments with customers around the processes that they go through and then organizational mindset? Yeah, so what I would say is first of all, from an organization mindset perspective, it's very important to start educating not just the analysis team, but the entire business on what this whole machine learning, big data things all about, and how to ask the right questions. So really starting to think about the opportunities you have to move your business forward rather than what you already know and think forward rather than retrospective. So the other thing we often have to teach people as well is that this isn't about what you can get from your data warehouse or replacing your data warehouse or anything like that. It's about answering the right questions with the right tools. And here's a whole set of tools that allow you to answer different questions that you couldn't before to leverage them. And so that's very important. And so that mindset requires time actually to transform a business into that mindset. And a lot of commitment from the business to make that happen. Then you look at the process, then you get to the product. Yep, and so basically once you have that mindset you need to set up an engine that's going to run and start to drive the ROI out. And the engine includes your technical folk, but also your business users. And that engine will then start to build up momentum. The momentum builds more interest and over time you start to get your entire business into using these tools. And it comes first to teach it. So this kind of makes sense, it's kind of riffing in real time here. So the product gap conversation should probably come after you lay that out first, right? Totally, yeah, yeah. I mean, you don't choose a product before you know what you need to do with it. So, but actually often companies don't know what they need to do with it because they've got the wrong mindset in the first place. And so part of the roadmap stuff that we do that we have a roadmap offering is about changing that mindset and helping them to get through that first stage where they start to put the articulate the right use cases. And that really is driving a lot of value for our customers. Because sometimes we hear stories like the product kind of gives them a blind spot because they tend to go into with a product mindset first. And that kind of gives them some baggage, if you will. Well, yeah, because you end up with a situation where you go and you know, you get a product in and then you say, right, what can we do with it? Or in fact, what happens is the vendor will say, these are the things you could do. And they just give you use cases. It constrains the kind of constraints. It constrains things. Exactly, yeah, exactly. It closes tons of opportunities. Totally, yeah. Because you're stuck within a product mindset. Yeah, exactly that. And you're not, you know, you don't want to be constrained. And that's why open source and the kind of, you know, the ecosystem that we have within the big data space is so powerful. Because there's so many different tools for different things. But don't choose your tool until you know what you're trying to achieve. We have a market question. And maybe you just give us your opinion. Caveat, if you like, it's sort of a global macro view. But when we started first looking at the big data market, we noticed right away that was the dominant portion of revenue was coming from services. Yeah, hardware was commodity. So, you know, sort of maybe less than you would obviously in a mainframe world and open source software has a smaller contribution. So services dominated and frankly has continued to dominate since the early days. Do you see that changing? Or do you think those percentages, if you will, will stay relatively constant? Well, I think it will change over time, but not in the near future for sure. You know, there's too much advancement in the technology landscape for that to stop. So if you had a set of tools that weren't really evolving that were kind of very mature, and that's what tools you had, ultimately the skill sets around them start to grow and it becomes much easier to develop stuff and then companies start to build out industry or solution specific stuff on top and it makes it very easy to build products. When you have an ecosystem that's evolving and growing and speedy with the speed it is, you're constantly trying to keep up with that technology and therefore services have to play an awful big part in making sure that you are using the right technology at the right time. And so for the near future, for certain, that won't change. Complexity is your friend. Yeah, absolutely. Well, you know, we live in a complex world, but we live and breathe this stuff. So what's complex to some is not to us and that's why we add value, I guess. Mike Merritt-Holmes here inside theCUBE with Teradata. Think big. Thanks for spending the time sharing your insights. Thank you for having me. Understand the organizational mindset, identify the process, then figure out the products. That's the insight here in theCUBE. More coverage of DataWorks Summit 2017 here in Germany after this short break.