 Live from Munich, Germany. It's theCUBE, covering IBM FastTrack Your Data. Brought to you by IBM. Welcome everybody to Munich, Germany. This is FastTrack Your Data, but I brought to you by IBM and this is theCUBE, the leader in live tech coverage. We go out to the events, we extract the signal from the noise. My name is Dave Vellante and I'm here with my co-host Jim Kabilis. Rob Thomas is here, he's the general manager of IBM Analytics and longtime CUBE guest. Good to see you again, Rob. Dave, great to see you. Thanks for being here. You're welcome, thanks for having us. So we were, we talked, we missed each other last week at the Hortonworks DataWorks Summit. You came on theCUBE. You guys had a big announcement there. You're sort of getting out of doing a Hadoop distribution. Right, we, theCUBE gave up our Hadoop distribution several years ago, so that's good that you joined us. But talk, tongue in cheek. Talk about what's going on with Hortonworks. You guys are now going to be partnering with them for, to essentially replace big insights. You're going to continue to service those customers. But it's more than that. What's that announcement all about? We're really excited about that announcement, that relationship, just to kind of recap for those that didn't see it last week. We are making a huge partnership with Hortonworks where we are bringing data science and machine learning to the Hadoop community. So IBM will be adopting HDP as our distribution and that's what we will drive into the market from a Hadoop perspective. Hortonworks is adopting IBM data science experience and IBM machine learning to be a core part of their Hadoop platform. And I'd say this is a recognition. One is companies should do what they do best. We think we're great at data science and machine learning. Hortonworks is the best at Hadoop. Combine those two things. It would be great for clients. And we also talked about extending that to things like BigSQL, where they're partnering with us on BigSQL around modernizing data environments. And then third, which relates a little bit to what we're here in Munich talking about is governance, where we're partnering closely with them around unified governance, Apache Atlas, advancing Atlas in the enterprise. And so it's a lot of dimensions to the relationship, but I could tell you since I was on theCUBE a week ago with Rob Bearden, the client response has been amazing. Rob and I have done a number of client visits together and clients see the value of unlocking insights in their Hadoop data, and they love this, which is great. Yeah, I mean, the Hadoop distro, I mean, early on you got into that business just to, I mean, you had to do it. You kind of got to be relevant. You want to be part of the community. And, you know, a number of folks did that, but it's really sort of best left to a few guys who really want to do that and Apache open source is really, I think, the way to go there. Let's talk about Munich. You guys chose this venue. There's a lot of talk about GDPR. You've got some announcements around unified governance, but why Munich? So there's something interesting that I see happening in the market. So first of all, you look at the last five years. There's only 10 companies in the world that have outperformed the S&P 500 in each of those five years. And we started digging into who those companies are and what they do. They are all applying data science and machine learning at scale to drive their business. And so something's happening in the market. That's what leaders are doing. And I look at what's happening in Europe and I say, I don't see the European market being that aggressive yet around data science, machine learning, how you apply data for competitive advantage. So we wanted to come do this in Munich and it's a bit of a wake up call almost to say, hey, this is what's happening. We want to encourage clients across Europe to think about how do they start to do something now? Yeah, of course, GDPR is also a hook of the European Union and you guys have made some talk about that. You got some key notes today and some breakout sessions that are discussing that. But talk about the two announcements that you guys made. There's one on DB2, there's another one around unified governance, what do those mean for customers? Yeah, sure. So first of all, on GDPR, it's interesting to me, it's kind of the inverse of Y2K, which is there's very little hype, but there's huge ramifications. And Y2K was kind of the opposite. Look, it's coming, May 2018, clients have to be GDPR compliant and there's a misconception in the market that that only impacts companies in Europe. It actually impacts any company that does any type of business in Europe. So it impacts everybody. So we are announcing a platform for unified governance that makes sure clients are GDPR compliant. We've integrated software technology across analytics, IBM security, some of the assets from the promontory acquisition that IBM did last year. And we are delivering the only platform for unified governance. And that's what clients need to be GDPR compliant. The second piece is data has to become a lot simpler. As you think about my comment, who's leading the market today? Data's hard and so we're trying to make data dramatically simpler. And so for example, with DB2 what we're announcing is you can download and get started using DB2 in 15 minutes or less. And anybody can do it. Even you can do it, Dave, which is amazing. For the first time ever, you can still do it. We'll test that problem. Let's go test it. I would love to see you do it because I guarantee you can. Even my son can do it. I had my son do it this weekend before I came here because I wanted to see how simple it was. So that announcement is really about bringing or introducing a new era of simplicity to data and analytics. We call it download and go. We started with SPSS. We did that back in March. Now we're bringing download and go to DB2 and to our governance catalog. So the idea is make data really simple for enterprises. You had a community edition previous to this, correct? We did, but it wasn't this easy. Not anybody could do it. And I want to make it so anybody can do it. Is simplicity the only, greater simplicity the only differentiator of the latest edition? Or I believe you have Kubernetes support now with this new edition. Can you describe what that involves? Yeah, sure. So there's two main things that are new functionally-wise, Jim, to your point. So one is, look, we're big supporters of Kubernetes. And as we are helping clients build out private clouds, the best answer for that, in our mind, is Kubernetes. And so when we released data science experience for private cloud earlier this quarter, that was on Kubernetes, extending that now to other parts of the portfolio. The other thing we're doing with DB2 is we're extending JSON support for DB2. So think of it as, if you're working in a relational environment, now just through SQL, you can integrate with non-relational environments, JSON, documents, any type of no-SQL environment. So we're finally bringing to fruition this idea of a data fabric, which is I can access all my data from a single interface, and that's pretty powerful for clients. Yeah, more cloud-native development. Rob, I wonder if you can, we can go back to the machine learning, one of the core focuses of this particular event and the announcements you're making. Back in the fall, IBM made an announcement of Watson Machine Learning for IBM Cloud at World of Watson. In February, you made an announcement of IBM Machine Learning for the Z platform. What are the machine learning announcements at this particular event, and can you sort of connect the dots in terms of where you're going, in terms of what sort of innovations are you driving into your machine learning portfolio going forward? I have a fundamental belief that machine learning is best when it's brought to the data. So we started with, like you said, Watson Machine Learning on IBM Cloud, and then we said, well, what's the next big corpus of data in the world? That's an easy answer. It's the mainframe. That's where all the world's transactional data sits. So we did that. Last week, with the Hortonworks announcement, we said we're bringing machine learning to Hadoop. So we kind of covered all the landscape of where data is. Now, the next step is about how do we bring a community into this? And the way that you do that is we don't dictate a language. We don't dictate a framework. So if you wanna work with IBM on machine learning or in data science experience, you choose your language. Python, great. Scala, R, Java, you pick whatever language you want. You pick whatever machine learning framework you want. We're not trying to dictate that because there's different preferences in the market. So what we're really talking about here this week in Munich is this idea of an open platform for data science and machine learning. And we think that is going to bring a lot of people to the table. And with open platform in mind, one thing to me that's conspicuously missing from the announcement today, and correct me if I'm wrong, is any indication that you're bringing support for the deep learning frameworks like TensorFlow into this overall machine learning environment? Am I wrong? I know you have PowerAI. Is there a piece of PowerAI in these announcements today? So stay tuned on that. It takes some time to do that right, and we are doing that. But we want to optimize so that you can do machine learning with GPU acceleration on PowerAI. So stay tuned on that one. But we are supporting multiple frameworks. So if you want to use TensorFlow, that's great. If you want to use CAFE, that's great. If you want to use Thaendo, that's great. That is our approach is we're going to allow you to decide what's the best framework for you. So as you look forward, maybe it's a question for you, Jim, but Rob, I'd love you to chime in. What does that mean for businesses? I mean, is it just more automation, you know, more capabilities as you evolve that timeline without divulging any sort of secrets? What do you think, Jim? What do I think you're doing? When you ask about deep learning, like, okay, that's, I don't see that. Rob says, okay, stay tuned. What does it mean for a business? If I'm planning my roadmap, what does that mean for me in terms of how I should think about the capabilities going forward? Yeah, well, what it means for a business, first of all, is what they're using deep learning for is doing things like video analytics and speech analytics and more of the challenges involving convolutional neural networks to do pattern recognition on complex data objects for things like connected cars and so forth. Those are the kind of things that can be done with deep learning. Okay. And so Rob, you're talking about here in Europe how the uptake in some of the sort of data orientation has been a little bit slower. So I presume from your standpoint, you don't want to over rotate to some of these things. But what do you think? I mean, it sounds like there is a difference between certainly Europe and those top 10 companies in the S&P 500. What's the barrier? Is it just an understanding of how to take advantage of data? Is it cultural? What's your sense of this? So to some extent, data science is easy, data culture is really hard. And so I do think that culture is a big piece of it. And the reason we're kind of starting with the focus on machine learning, simplistic view, machine learning is a general purpose framework. And so it invites a lot of experimentation, a lot of engagement. We're trying to make it easier for people to onboard. As you get to things like deep learning as Jim's describing, that's where the market's going. There's no question. Those tend to be very domain specific, vertical type use cases. And to some extent, what I see clients struggle with, they say, well, I don't know what my use case is. So we're saying, look, okay, start with the basics, a general purpose framework, do some tasks, do some iteration, do some experiments. And once you find out what's hunting and what's working, then you can go to a deep learning type approach. And so I think you'll see an evolution towards that over time. It's not either or. It's more of a question of sequencing. One of the things we've talked to you about in theCUBE and the past you and others is that IBM obviously is a big services business. This big data is complicated, great for services, but one of the challenges that IBM and other companies have had is how do you take that service expertise, codify it into software and then scale it at large volumes and make it, you know, adoptable. I thought the Watson data platform announcement last fall, at the time you called the data works and then so the name evolved, was really a strong attempt to do that, to package a lot of expertise that you guys had developed over the years, maybe even some different software modules, but bring them together in a scalable software package. So is that the right interpretation? How's that going? What's the uptake been like? So it's going incredibly well. What's interesting to me is what everybody remembers from that announcement is the Watson data platform, which is a decomposable framework for doing these types of use cases on the IBM cloud. But there was another piece of that announcement that is just as critical, which was we introduced something called the data first method. And that is the recipe book to say to a client, so given where you are, how do you get to this future on the cloud? And that's the part that people clients struggle with is how do I get from step to step? So with data first we said, look, there's different approaches to this. You can start with governance, you can start with data science, you can start with data management, you can start with visualization, there's different entry points, you figure out the right one for you and then we help clients through that. And we've made data first method available to all of our business partners, so they can go do that. We work closely with our own consulting business on that, GBS, but that to me is actually the thing from that event that has had, I'd say the biggest impact in the market, is just helping clients map out an approach, a methodology to getting on this journey. So that was a catalyst, so it's not a sequential process. You can start, you can enter, I think you said, wherever you want, and then pick up the other pieces from a maturity model standpoint. Exactly, because everybody is at a different place in their own life cycle, and so we wanted to make that flexible. I have a question about the clients, customers use of Watson data platform in a DevOps context, so are more of your customers looking to use Watson data platform to automate more of the stages in the machine learning development and training and deployment pipeline? IBM, do you see yourselves taking the platform and evolving it into a more full-fledged, automated data science release pipelining tool, or am I misunderstanding your strategy? No, I think you got to write it. I would expand a little bit. So one is it's a very flexible way to manage data. When you look at the Watson data platform, we've got relational stores, we've got column stores, we've got in-memory stores, we've got the whole suite of open source databases under the compose, IO, umbrella, we've got cloud, so we delivered a very flexible data layer. Now, in terms of how you apply data science, we say, again, choose your model, choose your language, choose your framework, that's up to you, and we allow clients, many clients start by building models on their private cloud, then we say you can deploy those into the Watson data platform, so therefore then they're running on the data that you have as part of that data fabric. So we're continuing to deliver a very fluid data layer, which then you can apply data science, apply machine learning there, and there's a lot of data moving into the Watson data platform because clients see that flexibility. All right, Rob, we're out of time, but I want to kind of set up the day, we're doing cube interviews all morning here, and then we cut over to the main tent. You can get all this on ibmgo.com, you'll see the schedule. Rob, you've got, you're kicking off a session, we've got Hillary Mason, we've got a breakout session on GDPR, maybe set up the main tent for us. Yeah, main tent's going to be exciting. We're going to debunk a lot of misconceptions about data and about what's happening. Mark Altshuler's got a great segment on what he calls the depth of correlations. So we've got some pretty engaging stuff. Hillary's got a great piece that she was talking to me about this morning. It's going to be interesting. We think it's going to provoke some thought and ultimately provoke action, and that's the intent of this week. Excellent, well, Rob, thanks again for coming on theCUBE, it's always a pleasure to see you. Thanks guys, great to see you. You're welcome, all right, keep it right there, buddy. We'll be back with our next guest. This is theCUBE, we're live from Munich. Fast-track your data, right back.