 Live from Las Vegas, Nevada. It's theCUBE, covering EMC World 2015. Brought to you by EMC, Brocade and VCE. Hi, we're back at EMC World. This is Dave Vellante. Ryan Peterson here is the Chief Strategist for Solutions, EMC. Focused on the big data space. Ryan, great to see you again. Great to see you today. So, I want to show, where do we start? Let's focus on what's going on in analytics and big data. It's been an area that we've focused on, as you know, many, many years. You guys really hopped in, leaned in, as Paul Moritz would say, full force the market, sort of came right to you. So, give us the update. What are you seeing? What are customers telling you? Yeah, I think the big change is that the distribution battle has been going, you know, obviously ODP has made that battle even maybe bigger, to some extent. You know, I go out to see customers a lot. I've seen probably a few hundred customers in the last year or two. And what I think is shifting is the thinking that one distribution's going to win this. I used to walk in the door and see people, they'd all be wearing the same shirt. Maybe they're a Clodera shop, or a Hortonworks shop, or a Pivotal shop, whatever it might be. That's shifting. That whole world is now going to, well, hold on, the application providers that are choosing their distro to work with. You know, for example, Microsoft and Hortonworks and Clodera with VAE systems, we heard from them the other day, these application vendors are creating solutions on top of distributions. And just like Oracle Database and Microsoft SQL and MySQL and all these different databases have existed inside of the same infrastructure, they all do SQL. Customers are starting to have to think about how they're going to handle multiple distributions in their environment. So that's an interesting premise, let's talk about that. So in the early days, you had obviously Clodera, one distro. The world was really simple. And then everybody started introducing the distro. Well, MapR was pretty early on, but Fujitsu, Intel, WinDisco, Pivotal, I'm sure there's three or four, oh, there's SiliconANGLE, I think had a distribution. And so now you've seen consolidation there, which everybody sees, okay, that's a sign that the world's going to a winner-take-all world, but you don't see that. No, I think that there's not going to be a winner-take-all. And historically, we see this, right? Databases, certainly there was never a winner-take-all. I mean, certainly Oracle did really well in that space. Microsoft did great in that space. Open source products like MySQL done really well in that space. But Red Hat, Susie, look at all the Unix distributions. Certainly there was some that didn't do well and went out, but yeah, you know, Icelon are one of our best products. One's on BSD now. I mean, different Unix distributions for different reasons and requirements. And I don't think Hadoop's going to be any different. So I'm fond of saying that it's not a winner-take-all, but the first guy, number one, makes the most dough. Number two, makes some good dough. Number three, makes a little bit of dough, kind of breaks the even, hangs out, and then everybody else, you know, so loses money and can't justify the business case and eventually gets out. But most markets like this can support three. Yeah. So that feels about right. Yeah, I think so. You've got Cloudera, Hortonworks, and MapR. Those are kind of the big three. IBM, maybe if IBM wants to fund its own distro at 10, it can afford to do that, but who cares? I think some of the unique consolidation on space is obviously Intel consolidating into Cloudera, ODPs, saying basically we're going to standardize the backend of Hortonworks, Pivotal, and IBM, and some others. But that created another territory, and then you've got MapR, kind of on their own, but in any case, you've got those three different major distribution platforms now. And they all do things a little bit differently, and as a result, they have different SQL engines, for example. Yeah, so the ODP piece is interesting. I mean, you know, Pivotal, I always felt like, okay, Pivotal is getting into the distro business to learn, and see what it can do. Maybe it'll hit a homerun. It seemed to me it'd always be kind of a Hail Mary approach, but it made sense. It's okay, and so then you get to the point of, okay, it no longer makes sense. The money to be made here is up the stack. Let's consolidate with Hortonworks. Obviously, Rob Bearden was on the other day. Obviously, he's thrilled. We had Mike Olson on the other day. He's like, who needs this? So, my feeling is that in the middle of those guys go down there. Let the market decide, right? I mean, that's just kind of where you are. Yeah, the market's gonna arms. Yeah, and my position is, how do we maintain flexibility so that all these different distributions can have the same argument they can fight, but at the same time, a customer doesn't have to replicate their content for this data lake and that data lake, and the third and fourth or fifth data lake. We want to create a centralized infrastructure that allows all of those different distributions to access that content so that any application vendor who may have chosen their horse, for example, maybe they chose the horse of Impala. Maybe they chose the horse of Embari. One of those different components is going to be interesting for an application vendor. And if you want to buy that application vendor's application, they're going to want to go with their distribution. Well, that's an interesting example. I mean, there's room for Impala. There's room for Embari. People are going in each direction. You look at what MapR has done. They've obviously got some traction in the marketplace. So I think your scenario is right on. I mean, you use Unix, Linux, very good examples. History tells the future, right? Yeah, it pretty much does. Do you feel like this market is larger and can more easily, let's say, sustain three? That my scenario, the third guy kind of maybe doesn't make so much, actually could in this world. I'll say something somewhat provocative. And I know that some of the people at EMCF, I was just talking to my good friend, Matt Cowher, and he was disagreeing with me on this. I feel like Hadoop is going past the analytics business. We all talk about Hadoop equals analytics, Hadoop equals. I really look at Hadoop as a major platform, maybe even the data center operating system of the future. Everything's used to be divisible. You take an application, you divide it in two, divide it in three, divide it in four. And that's how, you know, NetApp and VMAX and VMAX and all these other kind of block or file systems had scaled. It's been open to the volume and the volume and the volume in the same with systems. Exchange server's a great example. After some number of users, some number of emails, I got to break that up and create a second one, a third one, a fourth one, and it breaks it into divisible format. As we've done is scale out platforms now, that has completely changed the architecture. As a result, customers are redefining their entire data center based on those requirements. So looking at scale out storage, scale out compute with Hadoop, scale out memory, Spark, and now scale out applications, also part of the Hadoop ecosystem in many cases, but then look at like Cloud Foundry being that application development framework that's a scale out platform. All of those kind of components in a stack, make it so it's extremely flexible. And think about things like SQL on Spark with the latency metrics you're seeing on memory and DSSD's capabilities around the space. Do we think that maybe Oracle's going to get run for their money? The scale up database will actually go to a scale out database. The whole world's changing around data center. Well, so how are people using Hadoop? I mean, our sort of research indicates that the initial foray into Hadoop was the sort of reduction on investment in the traditional data warehouse. Even though you ask anybody what's going on in your big data, the two things that they absolutely point to that support the big data initiatives are the traditional EDW and ETL. Yeah, I think frankly the EDWs came out with a lot of cost. Certainly the benefits were there to justify that cost. But as customers started realizing, you know, I don't need to move the data into the EDW to do an ETL job, thus saving myself a lot of cost per terabyte. I can just put it in Hadoop and let it do the work and then spit out the results CDW. That became a really good justification for starting out Hadoop. But again, I think all this stuff was just justifications. It was the way to get in the door and get things started. Right, so that's my point. So now what? So that's sort of, I think you agree, right? Initially it was like, let's save some dough. This is great. Take some of the air out of our EDW because we're chasing ships and it's a bloody nightmare. Now what? How do we get more return on this? What are you seeing? You know, I had a great conversation with Doug Cutting. I've had some great conversations on the Hortonworks site as well. And it's been, I think this is what's happening next. ETL got started. Analytics grew it out further because now I can do some of the things that may be iterated or in a teaser we're doing before on Hadoop directly. But the biggest thing is scale out applications. Applications are being built that consume common data sets. Great example might be taking a Telco tower, Wi-Fi tower, or a wireless tower. And I'm getting all this CDR and XDR content. Well, you know what? That data looks the same. They're the same towers and across multiple different providers. The information all looks the same. I'm consuming it the same. So why can't I build a common application? EMC's actually been building some applications. BAE systems built an application around this. It sits on top of Hadoop. Starts automatically processing that data. It gives you a result. And that I think is the next step is we're seeing application providers build scale out Hadoop-based applications that are going to automatically consume that content and we're going to see more and more and more of that over the next few years. I couldn't agree more. I mean, it's taken a while, right? Mike Olson, I think at Hadoop World 2011, said this is the year of the application. Right. And, you know, It's been a couple of years. We're still talking about it. And he said, look, we had to build out the infrastructure. And maybe I was a little bit over my skis on that prediction, but ultimately it's happening. And that's where, so we've just come out with a new scenario on our big data forecast. I'd love to get your opinion on this. You look at big data and where the spend is, it's all services, like 40% of the market's services. And that's not scale. Services don't scale. So we sort of started to think about this and forecast it out and said, look, the future software. And software today is quite a small piece of the pie relative to the rest of the business, largely because of open source, but we think that's going to flip. We think that software is going to overtake services. Now maybe it's going to take five, seven, 10 years, but software is the growth engine. It's the only way this thing's going to scale. So let's go back on history again. Windows 1.0 came out. I remember that. I was alive. I'm still pretty young, but it was alive. And I remember it being really flat. There wasn't anything really there. And you played with visualization things. A lot of programming involved. A lot of services involved in those early adopter customers. They had to figure out how do I use Windows? Same thing happened with your iPhone. Same thing happened with Android. New opportunities just comes out. You've got to play with it and a lot of services to build your own apps. But when somebody figures out, hey, you know what? That data set's the same as that data set. The customers are all looking for the same thing. The software company's going to build a common software that can handle multiple different applications. You're going to see a massive shift of spend away from the services side into the software side. Scale-up applications are going to start controlling the data center. And as a result, you're going to see things like Hadoop and Cloud Foundry really as a combination I think, are going to scale out architecturally speaking across the entire environment to help handle applications and custom development against those systems. Ryan, your role strategy. What's your scope? Is it, I've been focused heavily on big data, but I've got all the vertical solution categories. So I've got media and entertainment, healthcare, life sciences. As you know, Hadoop has really been prominent in those kind of early vertical spaces. We've got architects who spend time thinking about how can I really change healthcare? How can I change life sciences? How can I change surveillance, for example? And we really look at how, the things we're saving people money, we're making people new money, we really think that people need to change their thinking and mentality. And our guys really think about that process every single day, how can I make, save, or save lives? And we love, of course, the saving lives one is the most important in our mind, but at the same time, we got to make the company some money so that they can afford. Making money is good too. You got to make some money so you can afford to save lives. You enjoy your life. Yeah, exactly, we're going to save. It's funny, we have this term life, liberty, and happiness. And pretty much all of these life, liberty, and happiness things can get tied back to, whether it's life, I'm saving lives. Liberty, we've actually had countries liberated by the use of things that could do. Data usage to get into things like Arab Spring, incredible changes in shifting economies. And then you have happiness. Well, you know what, we build major movies on top of our platforms, and we use analytic tools to make that process faster. Of course, we wouldn't have a life if it wasn't fun and enjoyable. So we think that what we're involved in is life, liberty, and happiness across all of the solution categories that me and my team support. Yeah, and I think, you know, the famous quote by Jeff Hammabocker that the best minds of my generation are trying to figure out how to get people to click on ads. That's the early Hadoop days. That's really changed. Somebody asked me, it was even last summer at a conference, you know, when are we going to see, you know, this big data affect our lives? It's happening. It's happening all around us. You guys talking about the information generation, which is good marketing, but there's some really interesting tidbits in there and sort of a futurist look at what the world's going to look like. I've got a session just after this. I'm going to speak to hopefully the IT leadership track. It's a bunch of CIOs and my pitch is chief information officers need to start thinking like chief innovation officers. And if they just simply do what the company tells them to do, then they're not going to allow for the process of change and complete conversion of their business. Every business out there is at risk because of what's available for their big data processes. Well, you know, it's an interesting topic and innovation's not going to come from doing a bunch of non differentiated heavy lifting. It's going to be combining technologies and creating new business models and changing the world, as you say. Well, data sets are everywhere. I mean, to me, big data is not about your data and your company. You might have a lot of data, but it's when you start adding an external sensor, start bringing in Twitter feeds, bringing LinkedIn information. I mean, the World Wide Web has tons more data than you've got in your environment. That's big data. How do you utilize those things to provide a new answer for your organization? And some of those things might save money. They might make money. They might save the lives. And you've got solutions in your title. So now, based on what you just mentioned, let's talk governance and you got so many opportunities for you all. Right, and that's the next big thing for us is to really collapse all of those capabilities and requirements, make sure we're providing the strongest, most stable, most governed cataloging processes that as you're bringing data into the data lake, we can make sure that you know where it is. I equate it to Legos. Memorial Day, you get a Lego box. You pour it out on the ground. It's just start building something. Well, and that's kind of what Hadoop is, right? You just kind of build something. You figure out what's going to come out of it. No schema. Yeah, but really what you need to do is take all of those pieces and say, you know, I've got red one by twos. I've got some blue two by sixes. Oh, I've got a couple of Lego wings. Hey, maybe I can make an airplane because I've got wings. Start building those things up. But if you don't have wings, you're not going to build an airplane. So you catalog what you have in your environment and also what's outside of your environment. How do you know what you can build and what you can create? And we need a chief innovation officer to start thinking, what can I create? Not, what do I have to create? Yeah, right. And then connecting to APIs and so many sources of public data now that you can leverage. It's just living in amazing times, as Joe Chucci says, interesting times. Brian Peterson, really appreciate you coming to theCUBE. Great session, good ideas, thought provoking. As always, thank you. Thanks, I appreciate it. All right, keep right there, everybody. This is Dave Vellante, we'll be back. We're live at EMC World 2015. This is theCUBE.