 Live from Dublin, Ireland, it's theCUBE. Covering Hadoop Summit Europe 2016. Brought to you by Hortonworks. Now your host, Dave Vellante. Welcome back to Hadoop Summit in Dublin, everybody. This is theCUBE, my name is Dave Vellante and we're here live. This is day two of Hadoop Summit. It's the first time we've done a Hadoop Summit in Europe. We've done a number of them, of course, in San Jose. It's my pleasure to greet Tug. Grawl is here. He's the Chief Technical Evangelist at MapR and he's joined by Ted Dunning, who's the Chief Application Architect, also at MapR. Gentlemen, welcome. So we're right here next to the MapR booth. We've been seeing it through our middle camera all days. This is a very interesting ecosystem, right? Look at the Gartner Magic Quadrant. I see Cloudera's on there. I see MapR's on there. I see Hortonworks on there. I say, okay, these are competitors. Get your all at each other's shows. It's really an interesting time that we live in. What's going on in our world, Ted? I think that there's a very interesting distinction in this marketplace, where we actually have a lot of interoperability, a lot of commonality. And if you contrast this marketplace with the NoSQL marketplace, where there's very little interoperability, by having this common basis, we've built a much bigger market, as opposed to just having tiny islands of functionality. And so that commonality makes a common market, which we can all compete on technical advantages, especially weekend. And that means that the overall customer experience is far better, because they can try alternatives and they can have more confidence that there will be at least one out of three who will be big successes. When you say especially weekend, because you have differentiation in the tech. Yeah, MapR's strategy is a little bit different. We are builders of revolutionary technology. We incorporate that into products, and so we sell products. That's different from a company that says they have no IP other than logos and trademarks, and then largely drive services. It's just a different product model. Ours is a bit more traditional, a bit more conservative. And so our differentiation is the technology. Well, you just described the spectrum of MapR to Hortonworks. All right, subscription services. And the cloud era is kind of in the middle. Yeah, yeah. Is that fair? Okay, so now, Tug, you're a technical evangelist to do a lot of outreach to the developer community. How does that message resonate to developers and what's your pitch to the developer community? Yes, it resonates, because what they want, they want to be sure that they will be using an API on a framework, independently of the platform, so they can choose between the three or the pure Apache product. Then they can focus when they use one or the other tools from the Kafka for streaming in real time or using SQL on one of the distribution. They can then work with the ops guide to see what can be the best distributions to run in production with different challenges about complex topologies in multiple data centers and so on. So for the developer, what is important, they have access to all the open source API. So the Hadoop market, if I can even call it that anymore, it's evolved to big data market, whatever we want to sort of, we know what we're talking about, not old data, new data and big data. It's very fragmented. I think Doug Cutting was saying that every time a new project comes out, they got to go support it and it's really difficult to support those new projects. 50% of their engineering resources go toward supporting those new projects and it just seems to be growing at a very rapid pace. But it's also, so it's a very highly fragmented space. We were talking this morning in the Cube and trying to come up with analogies. Is it like the old Unix market, you know, where you had DEC, UX and IBM and HP and Sun Solaris and then Linux came along and so you unified things. Although of course there were different Linux distros as well. Are there similarities that we can make, Ted, between the previous markets like that and this market? Can we learn anything from that? What's same, what's different? Yes, I think we can learn. Although of course the only real lesson from history is don't get involved in a land war in Asia and don't invade Russia. But beyond that, I think we can have some lessons. But I think that it's a very different world. It is not like Unix because in Unix there was very limited interoperability. You couldn't easily take a program from one Unix and run it on another. Yeah, that was their dream but it never happened. That's right, there was no binary compatibility. Unix was far better than the previous state of affairs where you didn't even have consistent file system semantics. So Unix was much better. But Linux then did provide, as you say, a much more unified operating stance. Now you can take programs and much more easily move them from one distribution to another. And that's much more like what we have in the Hadoop world. And I would contrast that with the relational database wars starting in the early 80s and say the non-sequel marketplace where we see far, far less commonality than we have in Hadoop. And I think far, far less aggregate value for customers and therefore for vendors. Because vendors are nearly getting value from customers in proportion to the value they give to customers. Because of that lack of interoperability. Then no sequel's much less valuable. Okay, so ODPI claims that that's sort of their mission is to create that interoperability. You have a different take on that where we'll all give me your take. Yeah, actually I think what I'll do is I'll quote Mike Olson, uncharacteristically. He says there are no barriers to interoperability now and I'm using Hadoop in a very broad sense. It's not officially correct. But we need a word for all of this stuff. So even though I'm in Apache and very, very aware that Hadoop as a specific technical meaning, let's talk about Hadoop ecosystem of the broad range of things including Spark. There are effectively no barriers to interoperability of basic programs. And in fact, as far as I'm aware, no map our customers had to change a line of code moving from our competitors to us of basic Hadoop programming. Now they may have to change some of their administrative procedures and things like that, but that's astonishing. That could not be true in the relational world. That could barely be true in the, well it could be true in Linux but certainly couldn't have been true on the UNIX world. You would have had to change headers and things like that. So we have far more interoperability than we've ever seen in a major replatforming. That's a big deal. So consensus comes through Apache? Certainly it has in the past. And I think that it's done a very, very good job of building consensus. Apache doesn't build anything. Apache doesn't build consensus. Apache provides a set of good practices and a place for people to come build community and build consensus. And certainly I've been involved in open source now for roughly four decades. And Apache is doing the best job I've seen of providing place and guidance for how to build community and consensus. It's working. Yeah, it's working, absolutely. In a way that it never has worked before. In our industry, that's true. And I think that was part of Mike's point as well. He drew comparisons to some of the really mid-computer efforts and a lot of those were posturing. So it's interesting to hear both sides. Horton works on earlier. And obviously you guys have different takes. That's fine, you know? If we all agree to be a boring world. But so, Tug, coming back to the developer angle, what are they asking for generally and specifically from MapR? So what they are asking is to be able to develop the new type of application on a single platform to make it easy to do the, I will say, historical big data with analytics that you have used on machine learning, but also be able to use on a single platform the real-time data. Some will talk about data in motion. You've heard this term before, but it's really grabbing the data in real-time, processing the data, storing the data in an easy way on a single platform with any APIs they want to use from Apache most of the time because you want, as we just discussed, this concerns things on the API. So this is what we see most of the time. Be able to build new and modern services on a single platform. Okay, and then so your simplification strategy, Ted, is convergence, right? Can you talk about what that means and how you're putting that into action, and then I want to end with the business impact of all the signals. Yeah, so Tug was talking about APIs and common APIs are, we think, critical because they are the interface to which developers code. And there's a lot of them. There are some Hadoopish-centric APIs. That's HDFS, that's Yarn, there's HBase API, there's the Kafka API. These are all emergence consensus sort of standards, but there are also traditional APIs like file orientation or document database. And those are the key things that developers absolutely need. And what we're doing is converging all of these APIs into a single platform, a single coherent experience. So if I go to my home directory, which happens to be on a Mount Markluster, who would guess, I can do LS, a normal Linux thing. I can see via the Linux APIs of files, directories, files. But now I also see tables and streams. All three modes of persistence and communication between programs in a single platform. Some people say that these need to be divided into specialists, but we feel that that leads to cluster sprawl and individual small clusters for single purposes. If you take a legacy situation with five silos and you bring it to a situation where you have 12 clusters, you haven't made things better. We think you need to take it to a single platform that has good multi-tenancy and multi-valency in terms of multiple APIs. Revolution comes from implementation of good APIs, but standardization comes from APIs not dictating exactly which bits implement those APIs. Okay, so that's pretty clear as to how you guys are approaching this. Not how much time left. We got a half day today. So I want to talk about the business impact of all this cool tech. It looks like the initial phase of big data Hadoop, whatever we want to call it. We joke sometimes. The ROI was reduction on investment. My enterprise data warehouse, which never lived up to its promises and I was able to cut the cost there. So the phase one and that's still growing, but not that interesting. Where's the real action in terms of business value that you guys are seeing? Yeah, and bottom line cost reductions always have this limit of zero cost. You can never go beyond that. So this is kind of a finite universe. Stock goes to zero, you're done. Yeah, and reducing things always gives you a blinkered viewpoint and you really have a little bit more of a depressive attitude toward the whole world because all you're doing is decreasing things. What's much more exciting. But there's good news. Yeah, the good news is and the exciting thing is that big data actually attacks the top line as well. Of course it does decrease the cost of persistence and making data available and processable, but it also attacks the top line because it presents new opportunities. There's a fundamental structural change in the economics of computing where we have linear scalability where before we had non-linear scalability and this changes the shape and the geometry of the essence of what drives the economics of this. That change then creates new opportunities and that's why we're seeing essentially a universal drive toward this new re-platforming of the enterprise. I'm not saying universal re-platforming of all applications but of a large amount of the new opportunities in computing are being platformed by big data. Some of the inappropriate uses of conventional technology are being re-platformed to this and this is where the big and exciting growth is both for computing and for the users of computing. New opportunities to do business in ways that were impossible before. And there's a lack of packaged applications, you know. So, I mean, is that, again, the developer issue? Is it a market maturity problem? It's a market maturity. I had an Apple One, there were no apps for it. The Apple Two, there were very linear apps for that. And Macintosh, we began to start seeing quite some applications. I had a Lisa, not many apps for that either. No, but we've gone from that homebrew style, which, I'm sorry, Hadoop 10 years ago was definitely homebrew style. Cast-off computers, cast-off software from Google and let's build this up and let's all make it ourselves. We can do it. Yeah, see, it's that way there. But the world has changed and it is becoming much more mature. There are beginning to be some very serious applications available on these big data platforms. That's exciting. All right, last question. MapR, like the business, what's going on? Right, you've got Cloudera staying private, Fidelity just took its numbers down. Hortonworks did a secondary. What customers say to you, how you doing? You're going public. Do you do what you need to do? What do you tell them? Of course I refer them to our CEO for all fundamental questions like that. We're kicking ass, is the technical term for it. We're doing great. We have huge numbers, very big increases year over year. A very large fraction of our vast majority of our business is very high margin product sales. So from a business standpoint, we are really doing well. And regarding whether we go public, again, talk to our CEO, CFO about those sorts of things. I definitely don't project in public on that sort of thing. All right, we'll leave it there that great. Ted and Tug, thanks very much for coming to theCUBE. It was a pleasure having you on. All right, keep it right there. We'll be back with our next guest right after this short break. This is theCUBE, we're live from Dublin, Duke Summit 2016. Right back.