 Live from the FIA Barcelona Grand Villa Compensator in Barcelona, Spain, it's The Cube at HP Discover Barcelona 2014 brought to you by headline sponsor HP. Welcome back to Barcelona everybody, this is Dave Vellante, Alan Andreoli is here, he's the Vice President of General Manager, we're HP servers, Alan, great to see you, welcome to The Cube. Dave, good morning, how are you? Good, thank you, feeling good in Barcelona, right, always a lot of action at HP Discover. We start dinner late, we don't sleep much, but somehow we're energized. This was a trolled night, and you know, it's all on the train only. That's right, we'll sleep later. So lots going on in your world, the world of compute has changed dramatically, cloud, scale out, big data, new types of workloads, so give us the update, what's happening in your world? It's an amazing change, you know, for years, I mean servers have been basically running back of these functions, you know, the ERPs, accounting, CRM, these kind of, you know, traditional tools. And now what you see is that there is an explosion of new workloads with a new economy, and any enterprise is being challenged, you know, in building new businesses to find growth, to find, you know, serve a new wave, and compute is the foundation of that. What you see is that HPC, performance computing, which used to be a bit of a niche market, suddenly is becoming mainstream. Many people want to crunch the big data, so you know, this is making it like a new life for HPC. What you see is that the cloud is driving an all new model for compute. One single customer may use several millions of servers, and there is one customer who has several millions of servers for one single workload, doing only one thing. So this is driving like an old paradigm for compute. And our strategy, what we are building is now syncing compute, reimagining compute in the perspective of all these new workloads, and this creates an old new world, and it's extremely exciting. You mentioned HPC and big data sort of colliding, and a lot of the HPC work historically was long projects, lots of simulations, sort of, you know, in the dark, we're not going to talk about what we're doing type of thing, and now it's coming to real time with whether it's, you know, Hadoop or other big data applications or analytics applications bringing transactions together. And it seems like the techniques are very similar. It's a lot of scale out, sometimes shared nothing. Talk about that workload shift a little bit. So there is a lot of scale out and there is a lot of blending between what used to be silos. So, you know, I've been in the computer industry for 30 years. I know it doesn't look like it. And well, you had, you know, you had the servers and then you have, you had the storage bay and then you had the networking gig connecting to the network. This is all evaporating right now. What you see is that on the same rack, you can have compute, you can have storage, object storage, you know, like a server, you just decide, do I put compute nodes or do I put, you know, disks or SSDs? And you have network integrated for the whole rack. So it's all coming together to go very, very fast, very high performance. And then the next frontier becomes this is generating a lot of bits. This is taking a lot of energy. So how do you generate the maximum output per watt? And how do you deal with space and density in this context? So we came with a new technology, for instance, for cooling, which is amazing, where we enable extreme performance with water cooling but without the water around the electronics. So that's kind of, you know, one example. Then, you know, we look at the cloud providers. How can we provide a solution that is very cost effective? We made a John Venture with Foxconn, the world leader in manufacturing, you know, cost effective electronics. And we did a kind of a joint model where we offer now cloud line to the big service providers. And then, you know, soon and later in the year, there will be even more to come. So we are surfing on a huge wave of innovation. Yeah, I mean, I've been around a long time, too. And I remember it's like it's all coming back now. Water-cooled mainframes. And then the HPC, I remember there was a renaissance in super computing companies. Kendo Square Research and Convex and Danny Hillis' Thinking Machines, all those guys. And they talked about scale out and this is the way to go. This is the future of the promise. They ran no applications, you know. So what's changed? Has Hadoop has changed all this? I mean, why now? Is it economics? Is it application availability? No, it's a business demand, it's business outcome. It's, you know, they need to crunch much, much, much more data because you used to have data coming from your ERP. Now you have data coming from your ERP, coming from all your machines and from your customers to social networks and all kind of different sources who are generating so much data using, you know, search engine, using mobility devices and whatever. This huge amount of information, the more information there are, the more you've got to find a way to crunch these information to be able to drive business outcomes and hopefully revenue and money. That's why HPC is becoming at the center of the agenda for many enterprises. And so-called big data in this scenario becomes the dominant workload. It's a dominant workload for a number of enterprises. But still, you know, the plain vanilla kind of base server is huge and is a core of the business today. Yeah. And with Gen 9, which is our new generations of servers, we've got a fabulous competitive advantage. We've, for the same what, between Gen 9 and Gen 9 and Gen 8, Gen 8 and Gen 9, we have three times the performance that we have. So it's, you know, even the base servers are going to quantum leap in performance and so we, you know, we remain very, very strong in the core and now we go into all these optimized workloads and, you know, driven by the business outcomes of our customers and our businesses. And people are bringing those workloads together. I mean, we're seeing transaction workloads and analytics workloads come together so that machines can make decisions in real time, right? Right. And look at the applications that are now driving the world, right? I mean, you have your mobile phone and you can buy, so suddenly you get into a financial application, security, validate your code and whatever, and then you get, you're going to pick into an inventory and you're going to ship it and you're going to build it and invoice it. And I mean, this is like, you know, everything is getting integrated into very sophisticated platforms running millions of customers. I mean, we had on stage yesterday, the customer, Qi Hu in China, who just, you know, started to buy our cloud line products. You know how many customers they have? 640 million customers online, more than Americans or more than Europeans. They're on their way to a billion, right? It's Facebook. And it's a company growing like this. You've seen Alibaba, right? Yeah, sure. When you see what's happening in China, it's like, you know, with a multiplicating factor to what's happening in the West, and in the West you and I see what's happening, right? So it's like an all new world and this is why compute is so important, because compute is the foundation of all of that. Can you talk more about the Foxconn deal and what that brings to HP, what the motivation was there and how you guys are working together and what it means for customers? You know, this is a great, great innovative idea. We just, you know, I think it's not invented in such a complex world. Not invented here is a very, very, very, very risky mindset. So we felt, whoa, there is this huge wave coming of people who want extreme scale, extreme cost, and you know, they don't want much engineering. They want just open design and, you know, like totally optimized, totally bare, because this is being managed at the software level and in that context, you know, the hardware is more like the slave and the software is the driver. And so we felt, you know, do we have all the competencies for that? What are the competencies required? It's manufacturing, it's, you know, cost of confidence, it's capability to design for extreme cost and it's extreme scale. And so we felt we are the largest in what we do and Foxconn is the largest in what they do. And the loyalty of the relationship between the two companies for 30 years led us to be John Venture only to do that. Service providers, gear, high volume, high quantity, and open design. And this is going great. So essentially that's a big part of your hyperscale strategy. Is that fair to say? It's a key workflow. It's not the dominant part. Our dominant business is to serve the enterprises. But then there is the service provider track, which, you know, you go on the cloud, if you go on the public cloud, you have these big players who need this type of gear. If you go in the enterprise, I don't think many enterprises will want to use that because enterprises want ease of use, manageability, tailor to their own workload. And that's what we do with the whole ProLiant line, right? So what we do with Cloudline is more very, very simplistic, bare iron optimized for the very high volume service provider. Now some enterprises may use it because you always have exceptions to the rule, but, you know, the intent is to serve the service providers with that. Well, we've had some discussions. I'd love to get your thoughts on this with large banks, for example, who are looking at that type of solution, trying to replicate, let's say, what the public cloud guys are doing. I presume you're getting calls from those guys. But it's a very small number of enterprise customers are going to do that, right? It's really a service provider space. So nobody can predict the future. You can be conservative. You can be aggressive about what everybody can predict, but how quickly the change is going to go. Right? I don't know what the future will be, but I want to be ready for the future, right? And I want my customers to be. So if OCP, right, or OCS, or Scorpio in China become, which are the open designs you are talking about, right, become dominant, we are ready. We are working with the key players there. We have an offering. We're ready to go. Customers? Thanks. Besides, then, do I want ProLiant or do I, you know, want to test and pilot and look at what OCP can offer to me? And, you know, that's where I'm going for part of my workloads in the future. If they want to do that, we are there. We are an OCP Gold member. We are working very closely with Microsoft for OCS. We are developing with Scorpio, with Alibaba and August. We are there. Yeah, whatever the market does, you don't care. I mean, that's always been your sort of attitude strategy. But we can offer the customer choice. We can say, if you want that, that's the plus, that's the minus. If you want that, that's the plus, that's the minus. With HP, you can have this choice and whatever you select will be with you, will offer you global support, global cell support, global service and consulting if needed. So, in a way, for this type of space, it's offering the OAM experience at the ODM cost. And that's unique. Nobody else is doing that. The OEM experience at the ODM cost, right? Okay. And think about what it is to work with an ODM, right? You are on your own. Oh, yeah. You have an outlet. You are on your own. It doesn't work. They gave you some more spare parts to fix it by yourself because you have to get somebody, you have to wait for the morning or the evening to get somebody in China to help you, right? This is not what we do. That's what Fox is going to do over there. But then, HP is in the front end helping the customers to make it work. That's our value proposition for that space. So, I guess there's certain hyperscale guys, Google, for example. Are we okay with that model? But many of the guys you're selling to, I don't know, maybe you're selling to Google, if you already probably can't tell me. But many of the guys you're selling to want that experience, that OEM experience. So, this service provider space is like a kaleidoscope. You know, like these instruments where you turn it and you see a lot of different colors. Oh, yeah, kaleidoscope. Yeah, sorry. There are a lot of different flavors to it. Right. So, guys that are doing specialization. So, there's always a coming from, you know, using traditionally blades. Yes. You know, core billing and, you know, network applications. And then they're going to offer cloud. So, they're thinking, are we using the same kind of architecture? Or are we going to do like the pure play cloud type of, you know, product? The Googles, the Facebook of the world are saying, phew, I'm so big. I can do it myself, right? And that's another approach. And some people are saying, I'm waiting to see what the Googles of this world are going to do. There is going to be some generic product coming out of that and we're going to use that. And some of the people are saying, well, I want an enterprise class product that works all the time. That it can be service, that can be managed. And, you know, I want to get this at the most affordable cost and insert this in my data center. There is different flavors right now. There is not one size fits all. And I think this is a big change in compute. There has been one size fits all for many years. And now you need to be able to offer the versatility of an optimized portfolio that matches the different business outcomes of different customers. Yeah, and that's our strategy. Thank you. You pretty much stand unique in that category as far as companies that sell. I mean, there aren't many left, big server ventures, right? But guys that have that full portfolio. I mean, IBM is out of the X86 business, right? I guess, I guess Dell in a way. So they are not, I never criticized, or whatever companies. No, I'm just trying to compare them all. No, they all have the successes, right? The point is that for many people, infrastructure has become trivial. Strategically. Ah, so from a vendor strategy standpoint. We see the opposite. We are doubling down in infrastructure. We want to reinvent infrastructure. And infrastructure is centered around compute. Compute is a foundation. We are the world leader in compute. And what you see at the horizon with the machine, you've probably heard before. Oh, sure, yeah. We're talking about the machine. Look at where this is taking us. We are building this very strong position at the foundation of the new style of IT with compute, store network, and the convergence of the tree. And then we piloting the emergence of a new engine, the machine behind all of that. That's where we're going. It's very powerful and we are very excited and our customers are about this direction. Nobody else is doubling down in infrastructure right now. Well, we've been in the business around the same amount of time and I'm sure you'll agree. Ever since I've been in the business, I've heard hardware is dead. Yeah. And it's not dead. There's a lot of money to be made in hardware. The way you make money has to change. And all these changes we've been talking are pushing for the reinvention of hardware. Yeah, for sure. Because software has got to run on something. Yes, hardware has got to be invisible because your user interface is software-based. But at the end of the day, the platform down below is what makes everything work. And we're going to make this better and better and better and more optimized to the direct business outcomes of our customers. Yeah, and you mentioned the ODM experience before. Their business model is not to provide that high-tech service. Their business model is the opposite. Yes. It's commoditized. Yeah. It makes it simple. It makes it very, very, very, very basic. And that's okay. We're very useful for the segment of the market. The segment of the market is very useful. We're doing this as well with our general manager, Chris Foxman. But that's one of the facets of what we offer. We reimagine compute in a very broad context. Context of the business of our customers. And we look at these different workloads and look at them right now as different options that the customers can have. Now, then comes an expression. Is this going to stay like, is there also a way of a re-ification of convergence at the horizon? And it's like an ecosystem. It's like, you know, you may think over time you're going to have a self-governing body that will allow these applications to self-configure themselves in a way that they optimize themselves and you don't need to buy different servers for different types of workloads. That's what we are working on as well for the next horizon led by the machine. Well, that's interesting because I was going to ask you about that because in the early days of hyperscale and everybody thought this would continue forever, I just went off the shelf, commodity components, ODMs, ascended, done quite well. But now you're seeing the big hyperscale guys pushing for customization and they understand there are unique workloads that I can't meet the requirements with just this off-the-shelf stuff. So that's come around, but now you're saying the vision of the machine is that homogeneity. Homogeneity. So you can have the same kind of toolkit, Lego blocks or the differentiation of different workloads or whatever. Or you can manage the convergence with systems that are going to kind of self-adapt the workloads that you are building. And this gives you scalability, agility and performance in your data center. I love the machine. And I love when Martin Fink talks as I just stop and listen, I write. It's so exciting. Very exciting, really innovative. I have no idea. I think I'm sure it's HP's going to make it work technically. It's so many. No, I know. I believe that. He showed the simulation machines from a business model standpoint. It's really printed. Yeah, yeah, yeah. From a business model standpoint, there's a lot of work to be done. You're developing the Linux variant. You've got the SDK coming out. It's all about the software that runs in the machine. If you can figure that out, you've got a winner and it will change the industry. To be in such a business, you've got to be innovating. And that's the blood of HP. Well. And to make it's even more. I think she's saying that we are increasing R&D by 10% year over year. I have been saying for five years, HP has to get back to its roots. Invent. That's what we do. And that is what HP is doing. You guys are talking about innovation. Invent. It used to be in the logo. That is the heart and soul of this company. So, Elad, great to see you again. Thank you so much. Thank you. Thank you. All right, keep right there. We'll be back with our next guest right after this. This is theCUBE. We're live from HP Discover.