 Live from San Jose, California, in the heart of Silicon Valley, it's theCUBE, covering Hadoop Summit 2016. Brought to you by Hortonworks. Here's your host, John Furrier. Hello, and welcome back to Silicon Valley Live, it's theCUBE here at Hortonworks, Hadoop Summit 2016. This is theCUBE's SiliconANGLE's flagship program, where we go out to the events and extract the signal from noise. We have two guests back on theCUBE, Wei Wang, Senior Director of Product Marketing, Gordon, and Matt Morgan. VP of Products and Alliances. Welcome back, good to see you guys. Yep, nice to see you. The last time we saw you guys in Dublin, turned out the update. Alliances, Matt, is really, really hot. We talked about the data in flight in Dublin. These two things, again, the center of the show here, more technology here in this event than some of the stuff we saw in Dublin, looking more solution oriented, but here it's about technology. Silicon Valley, Matt, the cloud and the cloud, hybrid cloud. It's the center of all the conversations. Can you just talk about the Hortonworks position on the cloud, Hadoop in the cloud, how it's being rendered, how these apps being generated? Yeah, I appreciate it, John. So first off, when I look at this event, it's very different than a year ago. The conversations have moved largely to the business layer where people are having conversations of how modern data apps are fitting in with their overall agenda. Some of the conversations that are on main stage are just blow away, right? You've seen the curing cancer conversation this morning. You saw the conversation about ensuring jet engines remain at full thrust with lower maintenance windows from General Electric. These are all really exciting threads. We're going back to your question about the cloud. If I look at the evolution of the data conversation, we started with, obviously, the structured data. We then moved away from databases to, hey, how do I augment that to have all types of data with unstructured semi-structured? Then we moved to the idea, oh my gosh, we got to have a data lake. And there's been some vendors that have come along and said, hey, the right answer here is some sort of converged infrastructure where you have everything stored in one place. Well, I would argue that the customers have led us away from that thought process. If you look at all of these case studies, they have a very common thread. They are now deploying what we see are these connected data platforms. The idea that data must be local to where it's generated and the analytics of why it matters is also local to where it's generated. That's a very different conversation. So decentralized? It's decentralized. Think of it as a data plane instead of a data lake. This could be data that's within the firewall in your data center. It's data that sits in the cloud, all of which has related meaning and getting a complete view of it's very, very important, but you're not going to do that with one single platform. You're going to do that with a connected platform architecture. This idea of a data plane that can sit across all of this and it's for data in motion, data at rest, data on the wire, all of it and everything in between. Some people are saying the opposite though, I've heard counter argument on that, which is centralize the data and move the management or governance, centralize the governance, the management and then let the data kind of sit out there. Is that the same thing or am I mixing up the metaphors there or the architecture? Okay, so this is a complex conversation. Let me try to break it down the way we see it. There is still a need to have centralized management and control of data. That problem hasn't gone away. What has evolved is the idea that everything can go into that central lake without exception and reality is we now understand that with these modern data apps reaching out to the edge, the final, the bleeding edge if you will, the need to have cloud connectivity and analyzing the data that comes off those devices is critically important. If we wait for that data to make its way through the cloud across your VPN into your data lake before we analyze it, we've already missed the value. So the idea of a connected data platform strategy is to think about different sets of data where it needs to be for the application need at that particular time. And I think that that is a very different point of view. So let me give you a couple of examples, right? Let's take connected car. People talk about connected car. It's a common thread. It was Herb that talked about more LTE connections go live on automobiles, the new net new subscribers on AT&T cell phones. That's a big deal. The reason why that's important. I mean, Mervage from Gartner. Was that Merv that had quoted that? Was it Merv? It could have been Merv. It was Herb who quoted it on Merv. Herb, not Merv, okay. So the idea on the connected car though is the cars are the ultimate endpoint. They're generating enormous data that is important to the car. So if these cars are self-driving, that data is being interpreted real time of what's going on. At the same time, there's machine learning that is often facilitated close to the car's connection point, which is the cloud that helps that car understand and cars behind it understand exactly what's coming on an upcoming curve. Inside the firewall, you could have oceans of data representing millions of miles driven that's being analyzed about how we can best equip the next generation cars that are going to come off the line. That's a connected platform strategy. We also laid out on main stage a retail version of that. I've got people walking into my retail stores. They have smartphones. These are connected to ID cards. We are pushing offers to them. There's data associated with the interaction at the end point. There's also real time computations that happen for the collection of different offers you're going to push that often can be in the cloud. At the same time, before I prepare my next proposals for my next generation of offers, I could have a whole marketing initiative analyzing that data over months and months. Right? This is a huge deal. I want to just lay this out just for the audience out there because there's so much confusion because depending on how you're viewing the elephant in the room, which is, you know, pun intended, data-driven applications. You could look at, say, Uber and say Uber's a app. That's right. There's app developers who work on Uber and they use data and some will argue they may or may not use Hadoop or whatever. Just use that as a random example. They probably do it most likely. But that is a different construct than saying there's an evolution of the data platforms out there, meaning cars have apps in them naturally that are run by machines. Right. That's a game changer. Businesses have processes that are essentially their own little data platform. So you can almost say that there's an emergence of a new set of data platforms that into themselves are enabling apps. That's right. And so the connecting is the key. That's what you're saying. That's absolutely right. So the word connecting is commonly used in this new world of internet of things and pervasive computing. Things are connected. They talk to each other. There's no difference whatsoever when it comes to data. So Hortonworks strategy around connected data platforms is a way of unleashing control to the point where you can build the right data architecture for your specific modern data app. We're gonna look at all of those platforms as a data plane. That's very, very different than let's pick one of the previous gen approaches that says, hey, we wanna have a converge platform. We wanna have everything associated with one single lake and everything's gonna come to that. That is not really feasible with these modern data apps. So would you agree or say, would you agree with me if I said the statement, there are app developers out there that are doing data-driven apps as well as the emergence of platform developers? Right. Because if you think about it, IoT and these things, they have these architectural issues, but it's not about architecture. There are people actually building platforms. That's right. So do you think there's an emergence of that category? Yeah, so I think that there is an emergence of that category. I also believe that if you look at what a modern data app has finally shown itself to be, it's like nothing we thought it was going to be. So years ago, people thought that a modern data app would look like your classic SaaS app with forms and analytics, and there are some of those, right? But what we're seeing today is that with modern medicine and precision medicine, connected cars, oil and gas, energy exploration, all the showcased types of apps that are happening in our business track, they're completely different. Some of them have many different user experiences depending who's interfacing it, and they're more of a connected system. So Uber's a good example, but I think they're even bigger than that. So the question I've been getting certainly online to get a lot of feedback on is like, hey, we're all the apps. John asked the question, we're all the apps. And is there an app track? So people like to put things in buckets after thinking of an app developer. There's not so much Hadoop apps per se in terms of like pointing at things like that. The growth of a number of Hadoop apps is X versus the growth of platforms. That's what you're suggesting. Analytics being a platform for it's not an app per se. So if you think about, let's talk about a modern app, right? A lot of people think that BI is the ultimate modern data app. They think, oh my gosh, here I can get access to my data. I can do my analytics. I would argue that that is a modern data app, but it's only a piece of what these modern data at platforms have become. If you think about what, that going back to the connected car, going back to manufacturing lines of the future or farming of the future, these types of applications are much, much more sophisticated, more impactful, and in many cases localized to their business because people are using them as competitive advantages. Cause if I'm a, let's say I'm in the farming industry, I'm in agriculture, and my job is to move chickens across the table. I wanna be able to optimize my yields in ways my competitors are. That's gonna be something I'm gonna localize, right? That's a process that's being automated. So I'd ask the question, since you're in an alliance, there's gonna be a way in on this. You manage the ecosystem. If the market has changed to a direction that people didn't see or moving in this new growth area, you mentioned the football just platform development, just as a placeholder, that would imply that there's a lot of pivoting going on. That's right. In the ecosystem. So I gotta ask the question, and this is something that I haven't seen yet, but in 2009, the Sequoia Capital of the Memo, RIP, stopped funding startups. They stopped literally funding. But we haven't seen that here, but there's been some talk about consolidation, which we seeing some pivoting, there's still funding apps and companies. What is the pivot strategy for the ecosystem and alliances? Are they vectoring into a new capability? What's your thoughts? So how do you, because people are questioning, is there gonna be a slowdown? We don't see a slowdown, but at the same time, there's still getting funding. Where's the pivot action going? Okay, so I think it's a segmentation conversation, right? And if I had to back up and look at what's happening in our industry, there's a couple of different segments. First off, we announced our cloud strategy. We are now looking aggressively at MSPs to augment that strategy. Obviously, we announced Microsoft as our premier partner, but in addition to that, we have a whole program around MSPs. And that's to serve the infrastructure needs of this connected data platform layer. And many organizations wanna localize the approach, verticalize the approach. We have a whole nother segment of partners that are really designed around building out, we call them platforms, we call them modern data apps. This idea that there are applications that will be built on data in motion, data at rest, and a connected data architecture that generate value. We have a couple of announcements that Wade's gonna walk through around those applications, but this stuff is early on the curve, right? These are packaged applications. Do you think it's a growth opportunity? It's an absolute growth opportunity. Wait, give us the update on the announcements that what you guys announced. It has the HTTP 2.5. Yeah, so. Other cool things. Do you see how it works? Certainly, there were speed up the pace of we delivering extended services offerings, right? We extended services release strategy. We talk about 2.4 with you at Dublin. And this time, we're announcing HTTP 2.5. It's a point release. It's a point release, but it's quite important. Let me give you a big point. They say odd numbers are always much better than even numbers. That's probably true. That's obviously true. So let me give you a couple of examples. Titan, actually, the partners we work here on this floor. The first one is dynamic security tax and metadata governance issue. And you can see if you're going to answer metadata governance session yesterday, it's full house, standing room only. And we talk about this dynamic security and governance initiative in HTTP 2.5 as a GA functionality. As part of that, there are three companies actually are working to integrate their technology to build modern data applications with us, specifically around Apache hours. For example, Waterline, Trifecta, that's your BI company there, and also a TVO, all integrated their applications on top of our platform-specific on governance and metadata. So that's important. The second one, I think that we talk about modern data application. If you have gone to our room keynote yesterday, we actually talk about building the next generation modern data application on top of Yang and also everything is dark rice. So Apache Metron as a cybersecurity modern data application or fraud detection, all are built integrated in the house on this platform, converged with the traditional, as you said, DevOps applications in the container. Is there an app track here, because I've been looking, I didn't see it and I was going to get some questions on Twitter around this. There are almost no really, apps with the actions right now, certainly Docker, Con, last week, containers really highlights that the DevOps ethos has won and application developers in the enterprise is just a boom right now. People are really excited and there's a lot of movement. Has that relate to your modern apps and what's the traction point? Yeah, so two comments on that. And one is, if you look at the people I talked to in this conference, that a lot of the people I have interacted with is not those traditional IT developers anymore. They are actually application developers, right? People would talk to me about specific use cases. I want to build this kind of thing. Can you tell me how to do it, rather than I have to go to my IP department and try to build something on top of it? So that's one comment there. We do have application track, we don't label it, but our business track now, if you go to every of those sessions, it's truly, they talk about the applications they build on top of it. Right, that's- On top of the data platform, not so much they do. On top of primarily Hadoop platform, but also they talk about other platforms to build on, and they use, again, we are GAR notebook on Spark, that's Spark, right, a lot of Spark applications that is highlighted in this conference. So Hadoop is a major element of it. So Hadoop is a major- But it's not mutually exclusive to other- It's not mutually exclusive. Data stores and other things. Not at all. And that's what customers are saying. And also in the cloud, right? So how about the alliances and partners? Let's hold on to the ecosystem. What's the exciting big news here this week? So I have mentioned that we're working very aggressively or actively with people like, I think, trifecta on the floor, a waterline and also a TVU. And the other thing, we did announce a resell agreement with ASCEL. So we talk about that because our customers, the goal here as a platform vendor is to provide the most choices available to our customers. Our customer wants certain relationships, certain type of solutions. We're announcing sort, if you remember, and as well as pivotal. So in this, to round it up, to run it up, continue down that journey, we're announcing a resell relationship with ASCEL. It's interesting. I see what HPE's doing, HPE Enterprise was talking about the pivotal guys yesterday. I really think there's an emerging trend that no one has teased out yet around platform engineering and platform development. I mean, pivotal calls it data engineering, which is not data scientists. These are engineers building platforms. So you're a platformer platformer. Would that be accurate? Could I say that? Absolutely. Well, if you look at the category, it has an S on the end connected data platforms. We look at this data problem is solved by this interconnection, right? And you look at the work we're doing with AppScale. It's a great example of apps that sit on top of these connected data platforms. In that particular case, it's all to do. But the idea of a single vendor being able to provide a BI tool for cubing directly to a business analyst gives you a complete solution. You know, in technology development circles, there's always the word religion tossed around, you know, certain religions. Religious war is going on when they talk about platforms. So you guys are suggesting your religion would be not converge, but connect. That's correct. That's it, that's all set. That's a good summary. And converge means throwing into one data lake, define converge versus connect real quick. So, okay, so I'll make a comment about data lakes. Still important, you're still gonna have data lakes. Okay, when you may have a cybersecurity data lake, you may have a marketing data lake, you may have a data lake in the cloud in the connected car strategy. Our argument is it doesn't have to be a single lake. Our argument is we understand these names. There's a power law of lakes. Exactly. And I think- Niche lakes, big lakes, little lakes. And data gravity. Finger lakes. And you need to have a platform strategy that's gonna be very related to the idea that there will be more than one, right? And that's what a connected strategy is. That's where the difference becomes a converge versus connected. And the best way to pay that off is to look at HPE. HPE announced today that they're transforming their entire IT stack around the Hortonworks connected data platforms approach at the data layer. And they're actually replacing some of that converged approach. So they had that stuff installed. They had tried it, they hadn't looked at it, but their strategy is moving much more towards Hortonworks. Their movies are composable. That's their big platform. That's the first thing. Essentially, you could argue is platform building. That's right. Versus converged hardware. Boxing. Yeah, that's right. Box moving. That's their channel. It's app moving in the cloud. Awesome. Way, Matt, thanks so much for sharing the insight. Congratulations on the announcement. We'll be watching the platform of platforms converged versus connected. The religious wars of platforms war. Of course, in the press, you always want to always kind of create a war out of something. But thanks so much for sharing. It's theCUBE here live in Silicon Valley 4. I do summer 2016 right back with more at this short break.