 Okay, we're here at Intel Forecast 2012. I'm John Furrier with siliconangle.com and this is a CUBE conversation with Jason Lacksman who's the general manager, cloud infrastructure group at Intel. Welcome to CUBE conversations. Have a nice dinner last night with your team. I'm always impressed with Intel. Just overall having their nose in the right place at the right time early and then the investment into R&D and then coming into the markets and doing that. You guys certainly had paid off for the company with the cloud investment years ago. You guys had made as a company and it's been well-documented in the news of late. But I want to talk to you about cloud infrastructure. So let's first talk a little about your role at Intel. Intel's changing, obviously. Always growing, the market's changing around computing. You've got storage and flash growing like crazy enabling cores on SSD cards, right? You know, for engine IO or violin memory systems, all kinds of cool new things. Storage has been very successful when you get the cloud group of the data center. They're all kind of coming together. So talk about the converged infrastructure equation right now inside Intel and how that relates to the marketplace. Yeah, so I think, you know, let me start with the customer first. I can hear what the customer's looking to do for a cloud is they really want to be able to take pools of compute network and storage and then be able to apply it to an application. And that fundamental way for what I think about a cloud and even before we started calling it cloud, we used to call it resource pools and that's what it was about. And so what we're thinking about and at the end of the day, we're not going to go sell a solution. We're an ingredient. We're a technology company. But we're always thinking about what are the underlying technologies? What do we have to bring the market that's going to help be a game changer that allow people to innovate and deliver solutions based off of it? And so, you know, when I look at compute, one of the big things that we're looking at right now is not just how do we keep driving more cores and make it more power efficient within the Xeon product line, but how do we come up with interesting compute that addresses a range of different needs? And so as an example, we are coming up on introducing some of our mic based products, which is the many integrated core and this is for HPC types of applications. So companies that have very big analytic types of problems can leverage IA compatibility, but bring it to something that's got 50 cores or more. We also are looking at introducing our first atom based SOC. So on the opposite side of the spectrum, you've got these ultra lightweight, many, you know, scaled out types of servers. And also looking at how you can put application specific enhancements into it. So one of the big trends we see right now is media acceleration in servers themselves. And we've got things and instructions and software development kits to go make that happen. On the storage front, it's a lot of cases, it's about scale out storage. So what people are doing is they've just taken a server, they're adding drives to it and the introduction of nonvolatile memory, interesting game changer there. So instead of having to continue to scale out just to drive IO, now you can introduce a solid state drives based on Intel technology, drive that IO, drive the number of servers down to get more performance, lower cost out of it. What's the biggest thing that you guys are enabling within the cloud? Obviously it's an ingredient, you know, Intel inside, you know, the famous, you know, inside the PC. You know, really creating that hardened technology that grew that whole application revolution. So for cloud, what are the key enablers that you guys are providing in there? For the, on the application side, that more on your OEMs are both. Yeah, so I think that there's sort of three basic buckets that I would say are important. I think number one is security right now. People want to know that the underlying infrastructure that they're deploying the cloud on is secure and they have the ability to audit those security requirements. So we are putting in fundamental technologies like our trusted execution technology to go make that happen. Second, it's all around performance. As much as people say performance doesn't matter, it really does. And yet to continue to drive more as law, we gotta drive, you know, more performance per core, more cores, and then make the systems themselves more efficient and deliver more performance. And then the third element is about efficiency. It's amazing how much cost goes into just the power consumption and that doesn't really do anybody any good. And so we got to go make sure that the performance divided by the energy efficiency is sort of the equation that we're driving. Those are the three probably most important ingredients, if you will, that are enabling a cloud solution today. So you guys always operate the cadence of Moore's law and how does that apply to cloud in your mind? Not necessarily from Intel. So if you take the, you know, the Intel perspective of Moore's law, doubling every, what, six months was it? Yeah, every 18 months. Doubling, so now the cloud sitting out there as a resource and we're just talking with Capgeminus. The view is consistent, the dust hasn't settled, these things are still in the air evolving. Where does that kind of Moore's law mentality really need to be focused on right now in the cloud? Is it at the past? Is it infrastructure as a service? What areas of the stack, if you will, needs the most work? You know, honestly, it's all of them. So let's take infrastructure as a service for a second. If I could tell you that the cost of your infrastructure was gonna be effectively cut in half because you can get twice as much performance every 18 months, that's a pretty compelling, you know, thing to go ahead and drive from the infrastructure as a service and your cost per VM. If you look at it on the flip side, software as a service, people that are developing software as a service on scale, they're not doing it at three servers, they're doing it at thousands of servers. And so if we can find a way to get twice as much capability out of that, it's a huge difference. And I'll just give you like one anecdotal example. We worked with two very large cloud service providers to figure out how we could accelerate their time to deployment to be ready with our new technology. So if we could- In terms of data center or the ability to preface in services? To be able to deploy, say, the latest technology in the server infrastructure. So let's assume they're deploying thousands of servers every month. If I can take the new technology and then accelerate it at six months, over a two-year period across just those two service providers, we're sitting $750 million to companies. That's the value of that infrastructure. That's the value in CapEx. And actually, I didn't even calculate the OpEx on top of that. Probably even more. Probably even more, yeah. Maybe you're pushing a billion dollars, right? But that's the type of volume economics. This is not- The people who are delivering cloud services, they're not the ones who are doing it through 1G2Z. They're doing it on a scale. And that's really the value and the importance of more laws. Delivering more capability faster, sooner at volume economics. Let's talk about the open data center of lives. There's been some criticism, not public criticism, people are polite. There've been some kind of, you know, backroom conversations. Well, you guys have the relevance of this. The market's changing. Is it really relevant? I saw a little Marvin about this earlier. Share with the folks why the open data center of lives is relevant and how you guys are handling the change and why you guys are behind it. Yeah. So the open data center of lives for the people that don't know, right, is a group of 300 end users. And so it's the customers who are deploying their own clouds or they're using cloud-based services. And from my standpoint, that's hugely relevant because, you know, there was a forecast. You know, people always had bigger forecasts, right? But there was a forecast that I saw earlier today that said that almost a quarter of a trillion dollars per year is going to be driven by cloud by 2020. I mean, that's an incredible amount of money to go capitalize on. And the reality is there's so many barriers to people actually moving to cloud still today. You're going to have to be able to talk about this. So if you can get 300 companies to have a common vision that say, you know what, we want to be able to adapt to cloud, but you've got to solve my top five problems and make those top five problems clear. Well, now I've got a target that I can go solve. And what's in it for Intel? I mean, what we do is we help to go create industries. I mean, that's what we've been doing. That's what we do the PC industry. That's what we're trying to go do here with cloud. And that's why we're behind the open data center alliance. Listen to the customer, let them tell us what the problems are. Don't second guess them. Let's go solve their problems. And then the other hand. Kind of enable the ecosystem. One thing Intel's done really well over the years, just being close to the company is when you get behind something, there's a lot of ecosystem around it that develops. Exactly. Can you talk about that in particular from, because you're an interesting spot. You're not kind of, you're kind of in the front lines, but you're not in the front lines. But at the same time, you're enabling an ecosystem of service provider partners, people that work with directly with large enterprises. What is the ecosystem like right now? What's the sentiment within the ecosystem? Yeah, so I mean, with this whole event that we're at today is really about is bringing together the users of cloud and the service providers, that ecosystem together to go solve the problems. And I think the ecosystem, it's changing. It's changing and there's also some conflict, right? On one hand, you have people wanting to deliver very proprietary vertical solutions. There's a benefit to that. It's fast, you've got one throat to choke is good. The other hand is that there's the risk of lock in or inflexibility and also customers get paralyzed. They're just not sure, like I gotta go make a bet and that bet may be irrevocable for some period of time. And the result is they sit on the sidelines. That's not all you want, right? We want customers to be able to feel like they've got choices, that these things will work together, that they're going to have some flexibility. They start to go use cloud services and the overall industry grows. And so what we're trying to do is find the right balance between how do you enable standards, which is good, it's like paving the road and still allow enough flexibility that people can go different speeds, different directions. And that's really, I think, the challenge from a vendor perspective. You've got a bunch of creating industries, slowly that get your take on big data. Yeah. Because, you know, obviously the storage group now is bolted on from what I heard with an Intel, my name ain't that be public, I don't know, but, you know, that goes hand in glove, right? Storage and servers. Big data seems to be, by our estimation, much more disruptive than, say, cloud is right next. Cloud is more an extension of infrastructure, as you said, resource pools. There's no real upside down market there. It doesn't mean really, at the same time, the big data market that talks about instrumenting, whether it's data, manufacturing data, military data, down to the consumer, retail, across the board, data and applications will change the business models of companies. That's like the PC revolution, putting productivity in the hands of the user, big data is putting data in the hands of, you know, business people and analysts, not just geeks. So we're seeing a massive rise up of kind of a new type of industry. How do you guys look at that? Because you have to enable that because it's part of your infrastructure and group data has to be dealt with in the cloud. You can't just spin down data. If it's a retail bursting situation, I can grab that data, I got to park it somewhere. It creates more opportunity, essentially. But it's challenging. So what's your view there and how do you see that unfolding? So first of all, I wholeheartedly agree with you. I think this is going to be far bigger than cloud. I'll just give a couple of numbers. If you look at the entire IT market, it's about half a trillion dollars. And so if you took all the costs out of it, everything was free, that's your upside. That's the most essentially you're saving from cloud infrastructure or infrastructure as a service and the efficiencies of it. On the big data side, you're talking about revolutionizing problems that are trillions of dollars problems. Government, you know. Waste. Waste, thank you for you said it, not me. But you know, government excess, right? But innovations, manufacturing data, new ways to go market and deliver products, healthcare, huge opportunity for big data, education, right? So you look at all these things and you say it's trillions and trillions of dollars. Every single vertical. And people want to monetize. And I don't think by the way, any of the forecasts that I've ever seen really do a good job of getting their arms wrapped around it because it's just so, so amazing. The thing that I am trying to go do and kind of back to Intel is again, we want to be able to create an industry or a big data. So this goes everything from how can we make more intelligent devices? So people talk about an internet of things or they talk about sensor data. We want to be able to have a better way of tapping into and having consistency across intelligent platforms, whether those are cars, smart signs, and embedded machines and manufacturing video cameras, right? How can you make those devices more intelligent so that they become kind of that network? And the second is how do you on the data center side enable and foster the use of things such as a dupe or other types of frameworks that allow people to cost effectively tap into that information? And so if we provide those building blocks and I think by the way the other things there's a huge enabling effort here. You talk about ecosystems. We need more data scientists, right? We got to go do work with education. We got to do work in research. We have to kind of do for big data what Excel did for small data, right? We got to make it easy and accessible to use. If we can do those three things, we're talking about an industry that's got so much growth that again, no one knows what that looks like. It's like what you guys do. You guys abstract the way the complexity hardened the tech so it's easy to use or turn key. We hope so. If we can do that with the high-end roles, because right now if you're a data scientist you got to be a quant shock or a PhD to configure H-Base, right? Yeah. That's not the way it's supposed to be, right? But I think that's ultimately an opportunity for the market. So with that, let's talk about some of the innovations that are going to enable big data. What are your top three? Is it flash? Is it the storage? I mean honestly, the compute, the course are key. So for you guys it's cores and flash, right? What else do you have? Networking is huge. One of the big issues when especially if you're running a MapReduce type of job is you've got so many different servers that are connected together and the latency to be able to go run a job effectively, the latency is important and bandwidth is important. And so we're seeing they need to be able to move to 10 gigabit as an example. That's one of the key technologies to accelerate that and do it cost effectively. Another thing that's really important is solid state storage as you pointed out, having fast access to the data that is stored and caching tiers in that regard. And then the third is delivering, you know, actually the software. Making sure that the software is tuned and optimized. Because again, this is another one of the problems people- What kind of software? Actually, I'll just use Hadoop as an example. I was working with a company that is a world class e-commerce company. And we were looking at their Hadoop infrastructure and at large scale they were getting 4% efficiency out of that cluster. That is not good. Not good for anybody. In terms of the hardware. In terms of the hardware. The processor and everything else. You know, when you hear a problem like that, the first thing you gotta go do is you gotta look at the software and say something could be tuned more effectively. And I don't think people know that Intel is a huge software group to go work on those types of applications. How we make them more efficient. And that's a great example. I mean, Hadoop is so early. It's nascent. It's great stuff. Final question for Jason. In the future, what are you guys looking at that you see as important? That may not be being discussed right now on the radar of mainstream and or industry journalists and analysts. But what are you watching to see that's really going to be an important piece that you just see the puzzle? Yeah. Well, you know, it's a good question. You know, right now, big data. I know it's kind of now on the radar screen, but I think just that here's the part that I think people aren't spending a lot of time on. I'll give you two things here. One is I think a lot of the focus on big data is web data. A lot of the energy, the money is going into how do I go mine the information coming off of the web? How do I figure to use it to monetize advertising placement? And that's certainly a great business model. I think we still need to take a lot of the smart data scientists and apply them to healthcare problems, manufacturing problems, logistics problems, retail problems, that to me is, I think, an area where it's under real problems. And the reality is we're so focused on the web. I think that's one. I think the other thing, which I think there's a huge opportunity for us when you look at the number of screens. Today, everyone's so focused on the phone and maybe the tablets and they're focused on the PCs. But the reality is a lot of screens in the future are going to be cars. They're going to be smart science. They're going to be TVs and kiosks. They're going to be surveillance. All these sort of embedded applications. We've been doing some work in some retail centers to go make some of these shopping experiences more interactive. Getting more developers to think about the fact that not only are they developing for a platform, but they're developing for this range of platforms, I think is another thing that hasn't really come to the fore. I think the more developers can be educated that the distribution of their work can be pushed out to a different edge device whereas retail will be very refreshing to them. I think most can't get their arms around that. Okay, final question then, I guess final, final question is what's your on your agenda this year at Intel? Looking ahead to the next year, what's your agenda for the group? Yeah, so I mean, I'll give you sort of two higher level objectives and then I'll talk about some of my personal ones. I want to be able to see that we have helped really move the agenda on standards-based cloud computing forward. I think the work that's being done here through the Open Data Center Alliance is tremendous. I really want to make sure that we're starting to see that cycle come full circle and users are stating their requirements, solutions providers delivering to it and the proof of concepts to actually show that it all works together. That's one, and the second is is that I want to see the market start to mature around big data, that there are solutions to go tap into that huge amount of data that people aren't tapping into today. Big data is about real time and you guys help provide with great cores and the embedded systems. Jason Westwood of Intel will be right back with our next guest here at Intel, forecast 2012. This is CUBE Conversations. Thank you.