 Okay, we're back here live in New York City for Hadoop World of Strata plus Hadoop World. This is where the big action is. Big data week in New York City, a lot of action. It's our day two of Silicon Angles, exclusive coverage of O'Reilly and Cloudera's Strata plus Hadoop World. We're excited to be here. And my two guests joining me is Steve Wacheef, Hadoop Technologist at Eula Packard, HP, and Amra Awadala, the co-founder and now CTO, not the VP of Engineering, CTO of Cloudera, CUBE alumni, good friend. And it was on the CUBE that Amra said that about Hadoop three years ago now, three and a half years ago, I saw the future and started Cloudera. So great to have you back on again, Amra. Thanks for having me. You're a rock star in the community and obviously great technical leader at Cloudera, co-founding is doing great. So welcome to the CUBE. Thanks, glad to be here. So Steve, obviously we love HP, we follow HP closely. Obviously there's a lot of stuff orbiting around HP as a company, but within the group that you're in, Enterprise Services or ESSN, which is what's called, they're in the convergent networking space, infrastructure, networking storage servers and have led that game for many, many years. Right. And now with Flash, Solid State, converging infrastructure is morphing to a whole other IO-centric infrastructure where Hadoop becomes important when you have industry standard hardware like what HP offers. So tell us one before you get into the whole big data Hadoop, Cloudera relationship, what are you guys doing with Hadoop right now? What's the internal view and roadmap of big data within HP? So absolutely, you know I think when we look at Hadoop we sort of have two perspectives and two areas that we want to address which is that one there's a lot of platform complexity. So customers struggle to understand what hardware do I buy and how do I configure that hardware and manage that hardware for Hadoop. And then the other aspect is application complexity which is how do I apply Hadoop once I have it to solve my business problems, right? So we're focusing on both those areas. My group, HP as a whole, my group specifically focuses on the platform complexity. We have reference architectures for Hadoop as well as an appliance for Hadoop. So obviously within HP they had that billions of dollars they spent on autonomy which is essentially the software side now. But there's really two problems happening in this ecosystem. One is obviously the software and the big data analytics, the application side. But also there's a lot of technical challenges on compute. A lot of energy and making it all work. Is that where you're now focused or are you also transitioning? Do you trade over to the autonomy groups? So we tend to go in with Hadoop, autonomy and Vertica when we go and talk, you know, have a customer conversation about data management. They sort of all do different things. You know, you sort of look at if it's a scale you have autonomy on the far left with unstructured document level analysis indexing. Vertica on the far right which is structured data, real time access and then you have Hadoop which is sort of on the slider in the middle that can sort of move around. And so we tend to just take a consultative approach and figure out what works best for the customer. So Amar I want to ask you a couple of questions because since we last talked you transitioned as VP of engineering at Cloudera, obviously the co-founder to CTO. And as the market moves more to the business value conversation which has been a big part of our topics this week. You're now talking to partners now. So is your new role, you're stepping out of the day-to-day engineering which you now have a VP of engineering. Explain the transition of your role and what you're working on now. It's standard, that happens with many, many enterprises and you need both functions as companies grows. You need them to be separate. The VP engineering is about the tackling and the hiring. Right, it's about hiring the engineers about executing on the protocol maps. Shipping code. Shipping the releases all the time. So it's a very tactical role. Which you did very well for many years. I started the company at the beginning and built the team to about 100 engineers and that's when I handed it off. The CTO role is a more strategic role about where the company should be going with this technology, how to stay aligned with what our customers and our partners are asking for and what they're building and how to make sure it all fits in the long term. So that's kind of what the CTO role entails. That said, it's sometimes very hard to define what a CTO actually does. So when I was in the process of kind of crafting the new CTO role for myself, I talked to a number of famous CTOs in the valley. One of them is Greg Papadopoulos, who was the ex-CTO for son. And he said a very interesting, he gave me a very interesting quote actually. He said, a CTO is like the CFO in this sense. The CFO is not responsible for making the revenues. The sales team makes the revenues. However, if the CFO mispredicts the revenues, you fire the CFO. So in the same sense, the CTO is not responsible for building the product. The engineering team builds the product. However, if the CTO mispredicts a key trend that affects where the product should be going, you fire the CTO. Yeah, so let's talk about those trends, because by the way, that's a brilliant suggestion because you have to scour the landscape. So I know you've been doing a lot of travel and I tried to, I was in the office, I should see if you're around. Yeah, we say sometimes the CTO is chief travel officer. Chief travel officer. You got me on that. I was trying to sneak that one in there. Okay, so you're actually in China and with Mike and Kurt, so you guys are, it's a global landscape now, so the market's changing. So what are you observing on the market side? Obviously, Hadoop is now mainstream to the point where Big Day is no longer just Hadoop, it's other things. Actually, I wouldn't say it's mainstream here. We're not mainstream, we're still at the beginning. We're still at the beginning of this movement. It's going to be a massive movement. It's going to take years to reach this maturity. So I wouldn't say we are mainstream. I mean, it's definitely growing very quickly. This event now is, what, 2500 people showing up at this event? Yeah. Four years ago it was just 500, so you can see at the graph. Okay, so I'd be gonna say more diverse with the more actors and business involved. So, I would agree with you. But I define mainstream to be more about where VMworld is. Like VMworld now is like 30,000 people go to VMworld. Oracle world is what, like 90,000 people. Virtualization. Yeah. So we're still like a few years away from that. Just don't sell for 600 million dollars to EMC. Ha ha ha. Then a $40 billion dollar market cap. Oh, that's it for the first thing. I know you know the VMware guys. I love them. Shout out to EMC and VMware. I love Pat Gelsinger who's been on theCUBE many times. Genius move. So, always good entertainment on theCUBE. Always fun. Okay, so that's good. So the trend-wise, you know, as you look at the landscape, HP has a lot of moving parts. So how do you look at partnering with HP, for example? Because like I said, they have a lot of moving parts. Huge conglomerate company. They got autonomy, they got Vertica, and they got ESSN. All huge opportunities. And ESC ends the muscle. Yeah, I mean, at first it's not a question that HP is the quintessential Silicon Valley company. So, I mean, we are born in Silicon Valley. We are part of the newer kind of generation of companies in there. It would be a shame on us if we don't partner with somebody like HP. Now they have many, many products that we need to integrate with. And right now the most obvious ones is as Steve highlighted earlier, was to get up and running a Hadoop cluster with all the hardware on the server side, on the network side, it is a complex proposition. It's not something that you can do. We were just at IBM IOD and they're now positioning big data. And they are not forking Hadoop. They are embracing Hadoop into their entire offerings. But they may have some top-level solutions that are more built for... Same thing, same thing. Like on the IBM side, you have NITISA. On the HP side, you have Vertica. On the IBM, what's the equivalent of IBM for search for autonomy? I don't know. I don't. Infosphere? Yeah, so, yeah, the big, it's, you know, sort of, you know, we're effectively trying to, you know, there's a couple of problems first. You know, I think you've got to get the, you know, the horse in front of the cart, right? So when customers talk to us about Hadoop, right? You've got to be careful about coming in straight away to the line of business, I think. Because they can go off and sort of start to figure out how to decompose things that are business problems into things that run as MapReduce. But there's a huge learning curve that's happening in the data center where these guys just don't grow Hadoop, right? And so, you know, we're really trying to make this simple and basically get them the platform first. So that's the horse in front of the cart. And then you come in and layer those solutions on top of that. Let the use cases drive more of an a la cart. Yes, yes. I mean, so we say, like we frequently would say there's so much buzz around big data, which is nice in everything because it starts the conversation. But at the end of the day, you need to be solving a real business problem. Absolutely, yeah. So we just look, we love of course all of the buzz and so on and so on. But I say like this big data movement is very similar to the e-business movement that happened a decade ago, right? It's just business. Just business. We're doing it with the technology. No one talks about e-business anymore. Yeah, exactly. Are you an e-business? No, no, everyone's e-business. It's 100% penetration. We think the same thing would happen with data. Like, you know, they did it right now a few years ago. In fact, we made that call just two days ago. We said, big data will be no longer, just like e-business. Are you in big data business? Yeah, I mean, what's next? So I think solutions are driving it, right? So like, no one talks about client server anymore. Yeah, exactly. You being a client server? Yeah. It really is, I mean, it's sort of, to quote an adage, you know, everything's bigger in Texas. It's just where I live, you know? We don't have big data. We just have data. Right? All right, so, Amar, I want to talk about partnerships. So you guys have been very successful. I talked to Mike Olson on the yesterday and looking at your briefings. Obviously, Impala, huge deal, great, great success. But as you roll out Impala, the big data platform, you have new logos and everything, you guys have been successful with your partnership, Milestones, looking at the business performance. The partnerships have been impressive and they're growing significantly. What are you seeing when you talk to, like, say, HP and other partners around, what's working, what's their mindset? Where's their, what's the psychology right now in the marketplace? I'll see platforms, you touched on a little bit of it. Can you expand more on that relationship? As the things cross the chasm, people who are running business, like HP and other companies, they have to integrate in, what's the mindset, what's some of the conversations that you're having? I mean, the theme is the same one we just discussed earlier, which is the buzz is great and everything, but we are about providing solutions. And when you're providing a solution, there isn't just one fixed formula that solves everything. It's a mix of things that will lead to solving the business case that the business is trying to achieve. And that's why these partnerships are very important. And that's why it's important to distinguish in the market between partnerships, which is what we call Barney partnerships, which means talk, hey, shake my hand. Yeah, we love you. Yeah, we love you, you're right. Barney, yeah, Barney, Barney deal. This is two solid partnerships where we're together delivering a solution Well, let's talk about the Barney partnership versus the real delivery because Barney, I love you, I love you, I've got that. But one of the dangers you mentioned was you get salespeople selling something too early and you can have some missed delivery there. So where are we on the delivery side? HP's a great example because when they sell, Donna Telly, I know him, I talk to him, I know him personally, he's a sales machine. He's not gonna, well, if we had this, no, it's gotta be baked. So what is it? So I've got a comment there, right? So our strategic relationship with Cloudera grew out of our natural business partnering where we started meeting in the market. It started with our sales force going, hey, customers are asking for Cloudera solutions on our hardware, let's strategically align this so we can deliver more efficiently. So I just think it's a natural organic evolution. That's a good response, I think that's legit because that's really what was happening. You guys make the industry standard hardware. If I said commodity hardware, we'd be getting phone calls from the server guys that Contier would be like, what are you doing? So like, no, industry's. You should give her a chance to counter them. Yeah, yeah, well, no, so I think it's a great point, right? So what is commodity? You know, it's a personal frustration of mine. You know, there's a lot of folklore out there about how you configure Hadoop clusters. And there's a perception that commodity means cheap. Commodity to me means that you're getting those same resources more, the price performance is much better by scaling out and scaling up. That's what commodity means. It doesn't mean cheap. Yeah, I know. I mean, we have a term inside SiliconANG, we call it ghetto servers, because actually we build our own servers. Yeah, yeah. We build our own servers and we call that ghetto servers. Because it's cheap, and you know what, there's a lot of risk points but we're starting up. That's the fun about Hadoop. But if you're a business, you put in blades and that's commodity hard, it's low cost, not cheap, right? So I totally want to just get that out there because I always joke industry standard, what the hell does that mean? But that's what's fun about Hadoop, right? I mean, I've had Hadoop clusters with laptops in my cluster. You know, if I'm trying to solve a particular, do some data set analysis, it's you use what you got, right? It's easier to manage if it's a margin of course. I was showing on our H-Base app last night and I'm embarrassed by how cheap or low cost the hardware is. And I think that's the general consensus of Hadoop. People say, hey, you know what? I can implement solutions in a very low cost way and that actually is tailor made for an HP because now you can then bring in other elements but the risk is, if it goes down, if you don't have the right support, it crashes. But you know, Hadoop is great. But thank you for your question, which is very important to want about the partnerships and a big company like HP, a big company like IBM or Oracle and many of our other partners, they don't move into a new market unless the customers are giving them the strong signal that we're ready for this, right? So this is not jumping ahead of putting the horse in front of the horse, it's not. Yeah, and also the other thing I would point out, just my observation in doing our CUBE interviews and talking to a lot of those players the business is so massive at the scale of an HP is that Hadoop is not cannibalizing it, it's just shifting the solutions and there's a lot of mash-ups so above Hadoop there's other things, you got graph database time series like IBM has an informix. There's all kinds of different stuff. There's not a mutually exclusive environment, right? So Hadoop works well and then we'll grow. So that's a big misconception that I hear from people is they, oh, Hadoop has to take away from something else. Actually, we're seeing a structured RDM relational databases actually expand. Those scope of projects are actually expanding. So, okay, great guys, thanks for coming inside the CUBE. Final question, we always do this. One year out to five years out, shoot the arrow forward. We're in the Hadoop big data, data-cosm ecosystem, whatever you want to call it. It's early, it's crossing the chasm on its way, I think, to a mainstream revolution. As I posted on SiliconANGLE today, I called it the South by Southwest for Geeks, Hadoop World. So, shoot forward. What's happening in the market relative to overall growth? How's it going to evolve? Just your vision. So I want to talk about one thing, which is it's not necessarily even that visionary, but it'll be one year from now, which is HP Moonshot Platform, right? Works great for Hadoop. Hadoop's a natural platform where you can partition the computation. And we've got our system-on-a-chip Moonshot Platform that's coming out. We plan to provide reference to architectures on it next year. And I think that's going to be quite revolutionary. We have Armour and Cloudera revolutionizing the data management middleware stack, and we're going to come in and revolutionize the data center. I just want to interrupt you there because I want to share with the audience out there. Moonshot is, I mean, HP's new hardware platform based on low energy consumption. We had a whole CUBE segment at HP Labs. So if you search HP Moonshot SiliconANGLE on YouTube, you'll find a slew of great videos on there. This is absolutely revolutionary hardware. I mean, we were popping our eyes out, can we get a box and it was only the prototype? So I'm sure people haven't heard about it yet, but you will check out Moonshot. So that was a good vision there, Armour. So a year out, I would say Impala will turn this industry upside down in more than one way and expect to see more things from us along these lines, along the Impala lines, and that meaning low latency. So low latency is definitely a very big theme for 2013. And then five years out, Hadoop World will be as big as VMworld. It's going to be 20,000 people for this event. Okay. That's when we can say it's mainstream. Okay, mainstream Hadoop in five years, I think maybe sooner, who knows, with this kind of pace. Exciting, Steve Wacheef, Hadoop Technologies at HP and Armour are all co-founder and CTO of Cloudera inside the CUBE. We'll be right back with our next guest right after the short break. Thank you. Thank you. All right, that was good. It was fun. Oh, that Moonshot, I've had more time. We looked at all the programs out there and identified a gap in tech news coverage. There are plenty of tech shows.