 Okay, welcome back to Oracle Open World 2012. We are live, theCUBE coverage of Oracle Open 2012. This is our third day of extended blanket coverage. I'm John Furrier, the founder of SiliconANGLE.com. This is theCUBE, our flagship program. We go out to the events, extract the signal from the noise and share that with you. I'm joined with my co-host. I'm Dave Vellante of wikibon.org. We're here with our friend Pauline Nist. Always a great guest, Pauline's with Intel. Pauline, welcome back. Well, thank you. Thank you guys for having me. Always great to see you again. We have great memories and great sound bites, but we want to kind of delve into the technical kind of like big picture. Larry Ellison, obviously keynote Sunday. I had predicted going in last week as a preview. I said, we're going to hear all about big data, how he invented big data. And then I'm going to, because last week we heard about cloud. And I'm like, oh, he didn't say anything about big data on Sunday. Of course, yesterday he's like the king of big data. And also the role of hardware in this software company is all sun. So that's their theme, hardware and software, working together. So it's the Steve Jobs, I want to be Apple, the enterprise kind of ego going on there. So that being said, what's your take on Larry's vision around the hardware and software relative to what he's putting together? Well, I think the best example of Larry's vision is really what they've done with the Exadata boxes. And I have to say, as an old, long time, tried and true data center person, they really have done a lot of incredible engineering on the software side. I mean, don't get me wrong. All of the SSDs, the PCIe flashcards, all of the great Sandy Bridge chips in there all help. But what they have done with compression, what they have done with deduplication, what they've done with filtering data before they send it over to the CPUs, you know, I don't think that a lot of people thought it was going to be Oracle who broke new ground as fast in the database world when they came out with the Exadata boxes. But I think they show that there is nimble and clever about adapting to take advantage of the hardware and really making it run better together. So we've joked in the past that the Exa, the X in Exa stands for Xeon. We heard a big push this week around Spark. So what's the deal? Steve Mills told me over a year ago now, publicly, Spark is dead, mark my word, you can quote me on it. We're going to see him at IOD, John. We'll have him on theCUBE, so we'll ask him about that. What's your take on what Oracle's doing with Spark? I don't purport to understand the relationship between Oracle and Fujitsu. I will just tell you what I know from dealing with our partners in Japan and our Japan organization, and that is that the Japanese government let out an RFP several years ago now, I think actually before the recession, because they want a CPU design firm in Japan. They want a Japanese CPU. And at the time they let it out, both Fujitsu and NEC were bidding on it and NEC dropped out. And so Fujitsu picked up the contract. So Fujitsu is getting a very, very large check from the Japanese government to design CPUs in Japan with design teams, with evolution of the semiconductor technology. And that obviously is a very nice prepaid voucher to kind of had to keep the technology going. So obviously Fujitsu partners, whoever they might be, would be able to take advantage of that investment. Essentially, like I said, a farm subsidy, John. That is. I mean, there's no paper trail for that, but the paper trail that is there is everybody knows what the Japanese government's doing. I mean, that's very clear. And the question is going to be, are they going to re-up and do another generation of that in the near future? And I think that's what everybody's waiting to see. So let's talk about Oracle and the data center. Larry has always wanted to have his own HP, his own hardware. It's all coming together for him. This year his messaging is really tight. It's really solid. I like what he's doing. They're like, I agree with you. You know, as much as we kid on Oracle, we do got to give him props for being clever and nimble. I mean, they've done, they're not a laggard. I mean, they definitely are not innovating. They embrace the innovation of others and call it their own. We know that, but they're not sleeping. They're not asleep at the switch. So that being said, fully integrated hardware software, data center. Now in that marketplace, you can't have one vendor. So it's hard to become the apple of the enterprise. What's your commentary on that approach of being the apple of the enterprise for Oracle? Well, in my new job, I spent a lot of time with software partners and with the evolving sort of SaaS cloud past scene. And there's one big advantage that I think we're going to see emerge for the software providers. And I include Oracle, but I also include a lot of the partners that are here. If you're going to go to cloud, if you're going to actually run a database in the cloud, who are you going to trust to run it for you? I don't have them to think as good as they are that Amazon's the first place you go. If I want to run a database in the cloud and I want multi-tenancy and I want security, I'm going to go to Oracle. Just like if I'm a healthcare provider and I want to run my hospital in the cloud, I'm going to go to Cerner or Siemens or one of the guys who does that for a living. And so that's one of the advantages that I think by virtue of the acquisitions that Larry's got, which is that he's got now not only database, but applications presence to go after that big chunk of enterprise database. And I don't think we ought to underestimate that because I think that there's a set of services that people will find more comfortable buying from him than perhaps other places. We had an entrepreneur on the CEO of Nutanix who said Oracle sells to rich people and their customer base. They pay a lot of money for Oracle software and like you said, they don't want to go to the Amazon, which I've called in the past the junkyard of cloud. Yeah, you can go build your own. He said they don't care about TD Ameritrade, seven buck trades. They want the wealth manager, the Goldman Sachs. So their clientele is used to having that kind of reliability in a partner to be catered to and deliver. And they don't mind paying for it. So the question is, as long as it can work, it's okay. So what do you think about that? Well, I think that umbrella pricing is always very dangerous. This market is littered by people who find it very hard to come down market and get taken out by people who come up market. On the other hand, UI and everybody else will try to hold whatever pricing we have as long as we have it and we can do that in a market scenario. So I think that the challenge and the joke I always tell people is I've been coming to Oracle over the world probably for 20 years and you go to the bars and you listen to everybody sit in the bars and piss and moan about the license prices and the checks they write to Larry. But the funny part is they're there every year. And when you come back the next year or the year after and I've noticed no shortage of people in the bars still complaining about it. Now what I do hear out here on the floor like you is lots of people providing, I've seen a couple of the storage vendors give people lots of tricks for how you can do stuff on your storage device and not have to pay for as many licenses. So there's a lot of that that goes on. And the question is, when is somebody going to offer something that's an equivalent that is really deemed as being as good as for your transactions for the stuff that really counts and then they'll come in and they'll undercut the price. The switching costs are huge. What Larry does every year is he ups the table stakes and says, hey, oh, we checked all the boxes, we got cloud, we got multi-tests, now we got big data. So the headroom is there, but the switching costs are so high. And don't you think that Oracle in any software company is always going to figure out a way to preserve its license revenue, right? They've done so with virtualization. They've done it across the board. The question is, do you get to a point, for instance, where an Amazon decides that there's enough market for it to do a premium service? You know, a redundant fault tolerant cloud kind of service that starts resonating with corporate people because it offers more of what they're looking for, but at a price point above, it's kind of minimum entry price, but below the kind of premium pricing. And I mean, I always have faith in the capital market system. I mean, you know, high tech is, I used to work in Boston around 128, which doesn't exist anymore, you know? And the thing that I always found most amusing about that is what the mini computer guys patted themselves on the back for doing the main frames. Okay, Unix and the PCs did the VAM. I mean, it's a circle of life kind of thing, you know? It just goes around and goes around. And we're seeing some massive forces right now. So let's go to the circle of life because, you know, there are emerging trends that are kind of really on fire right now. And it's all in parallel. On the business model side, on the technical side, and also on the market, with Flash, it's changing the game. So storage is now a big part of that. So comments on the dynamics of Flash, when we talked last year, we had some good conversations. So talk about what's happened in one year in this whole Flash boom. Because a lot of people have changed their stories a little bit, but expanded their scope of their value propositions in the architecture level and the infrastructure. We've seen software-defined virtualization all the way up to sand, caches, all kinds of stuff. Well, I think the first inroad for Flash, which we talked about previously, was the easy one. It was a solid state device. You didn't have to do anything. You didn't have to change your software. You just had to pull out the hard drive. You had to put it in, and suddenly, boom, you got all this performance. At the beginning, you paid a little bit more, but the prices have sort of come down. And it had some interesting artifacts, which is less power and, oh, by the way, higher reliability, because it didn't have things that spun around. Which is never a good thing. Now I think you're seeing the next tier of sophistication, which is everybody out here on the floor with these PCIe cards. It's what Oracle's done with the X3 box. It's what Fusion I.O. is here talking about. I mean, everybody's got a version of how to put the Flash on there. But I think you're starting to see some hints in the industry for what the next big breakthrough is going to be, which is, yes, people are going to keep doing clever things in the storage stack with this generation, but you're already starting to see hints from people like HP and IBM and others about next generation Flash technology, which is not just going to be a shrink or a little bit denser, a little bit cheaper. It's going to be a new transistor type. And probably the people who've been most public about theirs is HP with Memrister, where they've come right out and said, we're on an aggressive schedule. And, oh, by the way, we're not going after storage, we're going after the DRAM business. This is going to be a memory tier device. Interestingly enough, IBM announced to deal with Hynix, another one of the Flash players about four months ago. They didn't say what they were doing with them. Obviously, Samsung would be involved in things like this, Intel would be involved in things like this. So I think you're seeing us start down a road where we can see major, major changes in what everything from the memory through the storage tier is going to look like in another few years. Well, if you had to pick a technology to go into the enterprise, you probably wouldn't have started with Flash. No, no. But now Apple, Anabit, the volumes of Flash are so huge because of the consumer market, do you think that things like Memrister and other alternative technologies will be able to compete on a cost standpoint? You absolutely need the client market. I mean, there is nothing being developed that people won't find very clever uses for a client. Because we say it all the time when we're Intel, we can't make a market with Intel servers. It doesn't matter how many Xeons we ship, we need those million PCs a day going out the door to drive a volume curve. We can't even drive, you know, a DRAM conversion. You really need it to happen on the client side to really load the fabs and get the numbers there. So how's Spark going to do it? No, nevermind. Well, I'll tell you, I had the same answer for Spark that I'd give you at IOD, which is, our guess today is that none of the risk guys are making any more than 150 to 200,000 chips a piece. But they can do a custom part that is totally optimized. You heard the fidgets you guys talk about Oracle numbers. They can put special features in like that because they're not really constrained by how big the die is and what complexity they put in. And generally, the companies who can afford it are making boatloads of money on the software side. I mean, you only have to go look at Oracle's software margins, IBM's software margins to understand that you can always have a boutique chip if you want a boutique chip. So we're talking about cloud earlier. I think it's actually pretty clever what Oracle has done. They seem to have sort of skirted the hypervisor war and now they're sort of burying it into the Oracle RedStack cloud. What do you think? I mean, what do you make of that? And how do you think the uptake will be? It gets back to what I said earlier about why people will trust them to do cloud databases. I think part of what they're selling is security and the real world hasn't solved those problems yet. Speaks to the CIO, isn't it? Exactly, exactly, that they are, by controlling the whole stack, they're going to be able to go out in between the database containers and owning the virtualization layer, say, we're going to make this safe for you, we're going to guarantee you that other people aren't going to get into your data. And I think in the real world, even our security strategies, the open source security strategies are still evolving in how people are going to support multi-tenancy and how they're going to give them the kind of security that people want to be in a cloud environment. Yeah, so one of the things that, I wonder if you could help us squint through this that Larry was talking about is that, he said last night, we don't believe that the way in which Salesforce and even NetSuite, when they started in the business, do multi-tenancy, i.e. at the application layer, is the right way to do it. We've never felt that we weren't ready with the database, like he didn't say that, but Oracle wasn't ready with doing it at the database level, so they just, you know, did separate VMs. Now they've got 12C, they've got multi-tenancy in the database. How real is that? I mean, one might observe, okay, well, of course, Larry's going to say, we want to do it in the database. And of course the application guys are going to say, no, the application is the place to do it. Can you help us understand technically what the trade-offs are there? Well, I think the trade-offs are where your crown jewels are. Yeah, bingo. Which is, if your healthcare provider, for instance, and you've got to deal with HIPAA, it's not the app that lets somebody log an order on the hospital floor. It's your medical health record that's got to be protected. And that's data, okay? So you can do what you want at the app day at the level, but that doesn't get you off the hook from having to guarantee to the HIPAA guys that you're protecting the data from ever being accessible to anybody who shouldn't see it. And I think a similar thing is true, even in corporations, for say, things like your financial data, where you're subject to Sarbanes-Oxley rules and audits, that's a little different than an app layer where you might be logging sales calls and you might be logging front log in. And it's not something you're going to get audited for on your quarterly reports. So I think your sensitivity to whether that is at the app layer or at the data layer kind of depends on, again, where your crown jewels are. And if your crown jewels are in the data, that's what you're going to care about. So let's talk about HIPAA and the other big whales out there, obviously SAP, HIPAA in the enterprise. You're in the partner side. You're talking a lot about partners. The big guys and the upstarts. What's your general take on the landscape right now? Obviously HIPAA, we all know what's going on there. Meg Whitman's turning around the ship there, cleaning house, taking her medicine and kind of retooling, which looks good actually on paper right now. Although the enterprise group's doing very, very well with Donna Telly. What's happening with those guys and what are you hearing on the trends? What trends do you see that are really exploding out, that are really have legs to them or at least some potential to have legs? Well, I think all of the big guys, and you're going to hear a message I think in many ways very similar to what you hear here when you go to IOD, they're figuring out how to take the hype and the sex of the Hadoops and the cloud and everything and pull it in and incorporate it and connect it to their tools so that somehow you can still do all of the stuff that was familiar to you at the enterprise level but incorporate all of this new stuff. The buzz though really is on how does the new stuff take over the world? I mean, you know, there are people trying to figure out how to put transactions in Hadoop. You know, there are people working on, you know, Google's got a huge drill and dremel thing on how to do analytics in Hadoop. You know, their theory is if we do this the right way, we don't need any of those other databases, you know, we're going to inherit the universe. Now personally, I'm thrilled to see that happen because database has been one of the sleepiest, most boring technologies. You know, you couldn't get a comp sign major to go into database five years ago to save their frigging soul. Now it's where the action is, you know, and to get a whole new generation of kids coming out of the Carnegie Mellons and the Stanford's who want to experiment and play with this stuff and to have an open source environment that lets that happen, I think it's great. It's great for competition. It's great for entrepreneurs. It's great for new stuff emerging. I mean, you know, we're going to be sitting here three years from now having this whole different conversation around what the lay of the land looks like, I think. So extend that out a little bit with some vision because we'll just speculate. Big day to ask you Larry Ellison kind of giving what I call the kindergarten definition of big data kind of demo was pretty trivial. It kind of made it sound like it was complicated. But again, not too hard to pull off, but what he did point out is the business case. You know, so not so much the Twitter demo he did yesterday. It's like, yeah, I could do that in a weekend. But it's the business case. A brand manager wanting to ask the question, who should be sponsoring the Lexus car? Great demo for the business value. So let's take that out. So there's no, I don't think there's any debate about big data having value. How does that transform with these new computer science majors and the data, the role of data scientists, create who's going to come from people? So what has to happen in your mind, given your experience, you've been through a lot of cycles. What are you watching evolve? How is it going to evolve in your mind? This big data technical and then business? Well, I think the interesting thing for these kids is they want to innovate something they can touch and play with. And the beauty now, I mean, when I was in school, we didn't have open source, we didn't have the open source foundation, we didn't have Linux. You know, you had to go work for corporate America if you wanted to write code at the operating system level. And the neat thing now, if you're in school is you've got Apache, you know, you got open source. I'm excited, I mean, Linux for a long time has really been frustrated by servers. Servers were a pain in there, but we just wanted more complicated things. But now you've got everybody looking at how do I connect these servers in parallel? How do I pull a dupe in to do stuff like that? And if you can touch it as a kid and play with it, you know, you and a group of grad student friends, you know, can get together and propose a project and do it, and that's the beauty of it, which is you don't really need some big grant, you know, from DARPA, you don't really need some corporate sponsor. You know, I worked at Carnegie Mellon when I first got out of school, you know, and there are crazy people down there doing all kinds of things in labs. Crazy good, let's define that. Yes, yes, no, crazy good things. Like radical ideas around configurations and systems. Who haven't been burdened by, you know, stored procedures when they should have made a basis. So let's get back to Flash. So I look at Flash as like addressable memory back in the days when, you know, we would program, I mean, back in the early 80s, you know, when I was cutting my teeth on code, we had 64K to work with, and that was a lot back then, you know, you had to load an editor, and you had to, it's hard to, now you've got RAM and you've got Flash. So Fusion.io has a software development kit, so the trend is making that non-spinning disk Flash programmable for developers. Not like the engineering hardware developers, but like software developers. So what's your take on what might come out of that revolution, that capability? I see what I think is going to come out of that revolution is it's just going to look like a second tier of memory. You're not going to need, you know, anything more than how to read and write to it. That's the end point we're going to get to. What a level store. Well, I actually think it'll be a 2LM. I mean, you guys probably aren't old enough to remember it, but back in the day, there was an IBM mainframe that actually had DRAM and Core, if you remember what real Core memory was. You know, the little lady strung together in Asia. And it was a 2LM, because DRAM was just so outrageously expensive that you couldn't, you kind of used it almost as a cache to the Core memory. And I think we're going to get to that kind of point with the next generation of Flash, which is you'll use DRAM as a much, much bigger kind of caching device, but you'll have this huge persistence store that you can also, you know, read, write, address. And it's not going to take magic to do it. It's going to give it to kids who want to write code and play with it, particularly when they can get the operating system code and the Hadoop code. I totally agree with you. I think we are on the beginning of a complete revolution in systems programming for dummies. I mean, you're talking about a complete software specific, software-led movement that takes the hardware equation out of it and gives the power of many, many mainframes to a kid. I mean, I actually think that was the beauty. The only thing I can liken it to in my lifetime is what the Vax did with virtual memory, so that you could get away from all the partitions and the overlays and all the crap that we all grew up with, and it was suddenly the illusion of, it was good enough that you really did have the illusion of having unlimited memory. And you suddenly had a whole generation of kids who didn't know what it was like to ever be memory constrained, and that gave birth to a whole raft of applications that got developed for those machines simply because you could. So let's talk about another trend that we talked briefly about at IDF on our, which we have on siliconangle.com. If you go to youtube.com, it's our Silicon Angle. We have Pauline's interview at Intel Developer Forum here in San Francisco a couple weeks ago. But there's a trend that we talked about, network virtualization, software defined networking, now being marketed as software defined data center, which I love. It's like a path, it's like, get there, it's nirvana, you're getting there, everyone's happy. But it's still far away. Share with the folks what's happening in that world at a big picture level and how partners are reacting to it. Is it the next big boom like Flash has become? There's a lot of hype with it because Nasir was bought by VMware. You can see Larry kind of teasing out, you know, that kind of direction. He didn't really see anything specific, but you can see the dots forming. Well, they actually did. Fowler talked about who, I always forget how to say Zego, is that who Oracle? Is that who Zego? Yeah, right. Fowler talked about it this morning. That's Oracle bought, they were kind of the number two SDN behind NYSERA. And Oracle snapped them up the week after VMware bought NYSERA. So Larry was obviously thinking and watching. So they positioned that as SDN. And yes. But it's sort of, to me, it was more IO virtualization, but maybe they're... No, well, you're right. I mean, I think they took Oracle today to kind of the first step in positioning it, but I would actually not leave from SDN to Software Defined Data Center. I think VMware wants you to do that. And that was the position you heard from VMware for the week that they were here. But I think there's more than enough to bite off in just the Software Defined networking space because it's a world that's owned by relatively few players today. And it's emerging. And yes, and I think that the world wants to go open. I mean, the virtualization guys want it more than anybody because they want to see and manage the network, which is why Oracle wants it, which is why VMware wants it. But I think we all benefit from a network that is more easily managed for performance. I mean, the problem you have today is, as you've seen here, you can add more processors, you can add more PCIe cards, you can add more SSDs. The one thing you can't add easily is more network bandwidth. I mean, you know, you're sort of locked in. You said at HP Discover, network is the bottleneck. Yep, exactly, exactly. And that's the last, I mean, you know, this game, this industry has always been a fine, the next bottleneck game. I mean, for years it was CPUs now, you got CPUs coming at your ears. Storage is now checked. Storage is now checked. Where do you go next? You go to the network and you've got to have- Hello, Converged Infrastructure, we're back. All the naysayers, we were talking about that earlier yesterday, where we're poo-pooing Converged Networks in a few years ago. We're here. What do you think of this notion of software-led infrastructure, though? Are you saying it's going to be confined to the network or? No, I just think we have to walk before we can run. We got to solve the network problem really well, and then we can move back to solve the data center problem. I just don't want to see people turn solving the network problem into somehow the data center problem. There are two very, very different sets of solutions, and when you solve the networking problem, you have to solve it at all tiers. You've got to solve it in your servers. You've got to solve it in the backbone. You've got to solve it in the phone companies and the comms infrastructure. Because if you don't solve it through the whole system, you'll just move up the bottleneck chain. I mean, the joke in the US these days is that the bulk of the US bandwidth between six and nine at night is consumed by Netflix, so we're all paying the penalty. And that's because the core backbone infrastructure is getting saturated. It's not because the Amazon servers or whatever, there's just so much being pumped out. And that's why I say it's a big string of things to change, because you can give IT guys the tools to make malleable configurations in their four walls, but if the stuff they connect to also doesn't change, they're going to hit a brick wall. So about IOD, the IBM on-demand event, information on demand from IBM coming up, we're going to have the cube there. What do you think about what's going on with IBM? And you mentioned they've got to showcase some of the big data stuff. But IBM obviously has been a great turnaround throughout their market caps, the largest of all the companies on the list that we've reviewed here at Oracle Open World. HP's still going to work their way up, but IBM, they're in a good position. What's your take on them right now relative to the world that's evolving around them? Well, I think people have talked about it for a long time, but the real flavor you get from IBM that's different than you get here is they're a business services company. It's a company that's really led by the services guys. And they're going in there to solve business problems, whether it's to run expense vouchers for a company or do analytics for a company or provide a solution, but they don't tend to couch the sale in terms of I have a database to sell you or I have this to sell you or I have that to sell you. It's what's your business problem and how can we help you solve the business problem? The whole smarter planet thing. And I think there are relatively few people who can get their hands around it as well as they can. And the question is can they do it in ways that are going to help us innovate and move forward? Because they certainly have the arsenal. I mean, they've got solutions in every dimension. I have, you can't look at their analytics portfolio and in these days of big data, not view that they at least in the near term have the mega advantage of being able to do something with big data that actually gets you dollars at the other end. And I think they'll be wise enough to play that, but that's a big deal right now. I mean, they've kind of gone, just like Oracle they've done lots of acquisitions, but rather than do it at the apps tier, they've done it at the analytics tier. And the question now is, can they print money because of big data? Do you see IBM as Oracle's biggest competitor? That's interesting because on the one hand, you know, if you did the Venn diagram, there'd certainly be a big chunk of overlap space in the middle where if you want an enterprise, high-end enterprise database, you know, you're looking at IBM or you're looking at Oracle, but then the question is, what do they bring in the rest of their circle? And that's the part where they're fairly different. But then the other one that we talked about earlier that I don't ever underestimate is, you know, the number one by volume database provider in the world is Microsoft, you know? And if you go with the old strategy of coming up from the bottom is always easier than coming down from the top, you've got to wonder over time if that doesn't also have its appeal because they're guys who are looking at cloud. You know, I think Microsoft today and Oracle today are the two big guys who are making, I just looked at the top 10 cloud providers that IDC put out. And of course, you got Salesforce, you got Adobe, and you got a bunch of the typical people you would see. The two corporate people you see are Microsoft and then Oracle, which is actually kind of interesting. Right, that is interesting. So, you know, we've been following that now for the last couple of years, but you know, what's interesting is that Larry's not specifically called out Microsoft, at least that I can remember, John, at Oracle Open World. He's certainly called out Salesforce, he's calling out EMC, I think he's taking potshots at IBM, but not Microsoft, and it's interesting. Maybe it's the pricing. You know, yeah, well, right. But your point is, I mean, they're everywhere still. A lot of the guys he's called out are guys that he wants to take customer dollars from. I mean, you know, I said to people when the ExoBoxes came out, be very, very careful, and it's not just going after the server companies, the HPs and the IBMs, it's going after every soft, every storage partner that he's got on this floor. You know, because you look at how many EMC or NetApp boxes are connected to Oracle databases, and if you buy an ExoData, you don't need those. Where does the revenue from that move to? You're still putting those bits somewhere. You're just paying Larry for them. Pauline, talk about Intel for a minute, because Intel has been going through some changes. IDF was a really refreshing conference this year. Although it did seem a little bit kind of smaller scale, not a lot of hype, a marketing hype, but like just meat and potatoes Intel, you know? And the role of storage has changed. And so talk about how that has affected us. We're talking about, you know, checking the boxes, right? You guys have always crushed it on the chips and the PCs and servers. Storage is second check in the industry and it's then got networking. But so how does that relate to Intel? Or check on the boxes, servers, storage, networking? Well, I think we are as interested in anybody in pursuing this NVRAM technology. We're just not talking publicly about it yet. I think a little bit of what you got at IDF was, quite frankly, we were out there very early and very loud about wanting to get into phones and tablets and other media and didn't really hit the dates and the marks we had. So now the rules are until you got a product, you don't stand up and say you got a product, until it's shipping with a partner, you don't stand up and make a lot of noise about it. So I think we've kind of learned from our past mistakes that we've got to go out there and sell the product and sign up the people who are going to ship it. And IDF was actually kind of interestingly timed because I think when it first went on the calendar, everybody had hoped Win8 was going to be shipping. And of course, when that slid out till the end of the month, we kind of ended up in this funny never, never land where we could show a lot of ultra books and we could talk, but the software is not our product. We couldn't launch it for our partner. So life is timing. Sometimes you win, sometimes you lose. But the messaging was awesome. I was on the Twitter stream and keep doing my normal tweet thing, live tweeting of the keynotes. If you were. Very awesome messaging, intelligence systems. You guys have more solution focus when we were talking about that briefly. Less ships, speeds and fees. I don't think Dottie had one speeds and fees slide in this whole presentation. Quite impressive. And now the role of storage group because David Tue, who I interviewed and also he's been on the queue multiple times. They seem to be taking on more of an expandable. They always seem to be not as popular as the data center guys that they didn't tell which you know, it's the cream of the crop over there, data center. But storage always gets, Dave and I always talk about the storage guys always kind of get the bad rap. But now storage is the hottest thing and it's leading the charge. And a lot of these with flash or anything else. I think the renaming of the group, the data center group, you know and particularly with Diane who's an XCIO coming in going to see a lot more focus on storage and on networking. I mean, look at what we've done in fabric. So we bought Fulcrum. We bought the Qlogic and Fitiband assets. We bought the Cray assets. So I think we didn't buy them just to put them on the shelf. You know, you're going to see us invest in those areas because we understand that as the silicon is getting faster, interconnecting the pieces is also important and not having those bottlenecks. So that, you know, as you've seen here the IO doesn't become a bottleneck. The network doesn't become a bottleneck. We're very, very interested in SDN as an open standard remaining, you know and I know that VMware has said good things and we're all going to hope that they deliver on those good things, but we're also going to take advantage of the standards bodies and everything because we don't want to go another round of a very close network subsystem because you need all three. You need, it's a three-legged stool and you know what happens to a stool if you cut off one of those legs you end up on the floor. I mentioned Infiniband. Larry of course loves Infiniband. I mean, it's crazy about it. Where the Qlogic booth of guys bought the Infiniband assets from Qlogic. This HPC business, John and I were out at, was at the DDN analyst meeting, was it last week? Yeah, it was last week. At DDNs, it's kind of a niche storage company but they're into HPC. And the premise that they put forth is that, you know, the HPC world is just set up perfectly to support big data. And I wonder if you could comment on that just generally. Well I think it's actually kind of interesting because with my history at Penguin, you know, for a long, long time, volume clusters with Ethernet, you know, was where all the money and all the heat was and there was this little, it took fringe playing around with Infiniband but you had to be, you know, a government agency who was doing something peculiar and you really were going to pop for, you know, all of those Infiniband cards. And as the CPUs have become faster and faster, you've seen more and more people on the HPC side moving, you know, to run MPI over Infiniband. And just as you've seen on this side, the data intensive guys like Oracle and like IBM moving to use Infiniband. So, you know, obviously there's something going on there and I think that the role of HPC, which has actually been pretty interesting and we talked a little bit earlier about the Japanese government, but what the US government has done is really used its resources to push HPC and to believe that if you push it to get to the high water mark in terms of either size of clusters or number of computes or interconnect, that will waterfall down to the rest of technology. And I mean, you know, it was an alternative to what Al Gore would like to have you think it really was DARPA who invented the internet, but nobody wants to talk about that because that means you'd have to acknowledge an agency that we don't want to talk about. But, you know, the government's going to do that. The government's got Exascale and, you know, Petascale computing going and there are certainly large elements of that that look very similar. I mean, Hadoop hasn't gotten there yet. Hadoop, everybody wants to build with cheap parts, you know, I just want to put these cheap servers and these cheap interconnects together. But when you get really serious about doing something with the data and running analytics on it, you're going to want computes. And the minute you amp up the computes, then you're going to want to amp up the IO. And then you're going to say, well, maybe an instead of band cluster really is what I want. But obviously that's not going to come out of open source. Open source is going to sort of stay with the volume, you know, the volume of the solution. Yeah, exactly. Our industry standard hardware, people always smack me around, but it's both, right? It's the commodity low, low end and industry standard like Dell and HP. So you think it's hand and glove with the HPC? Yeah, I do, I do. I think there are pieces that HPC will drive it and refine it, but it takes the rest of the world to adopt it and sometimes drive the, I mean, Melanox would not have been a huge success if all they had done was sold to the lunatic fringe of HPC. I mean, we wouldn't, nobody would be talking about a low load, low load Larry by 5% of them or something. So let's talk about like the systems, because we bring up the, let's talk about the Vaxes and core memory back in the day. A lot of the Hadoop big data stuff that you see out there are built on commodity gear. And so that's, you know, racks and stacks. And at some point an engineer is going to say, hey, you know what? I'm going to do a billion calls a second someday on an API and I'm going to have to build my own mainframe or mainframes, plural, parallel systems. How would you look at that right now? Because a lot of companies, and we face this as we talk about our products that we build on our data center, not today, but like at some point it's not just hiring the software guys. You don't have to get some hardware engineers to say, hey, let's do an infinite cluster. Let's look at the compute IO. What is, what is, how does that evolve in your mind's eye, right? From today, let's just take our case study, for example. Like, you know, we have a couple servers. We're going to grow. And a lot of people who are doing MongoDB, they're all doing the same thing, right? They may be racking and stacking on premise and their own private data center because it's too expensive to do IO in Amazon. They may be do a little Amazon out here, but only they have their own data center. What's that system look like if they have a clean sheet of paper? Well, you have a bunch of things going on simultaneously. You've got the guys like Google who are so big that they design their own motherboards to rack and stack and pack and, you know, get them in and cool them with ambient air, you know, and all that kind of neat stuff because they're trying to bring the cost down. And I think you've always had the guys and you see it even more so today because the rise of the ODMs that, you know, the offshore manufacturers and the tier two guys have always given people like yourself who want to play the ability to go get a board and rack and stack them and put them together yourself. I mean, you know, at Penguin, we used to do that all the time. And so there's some of that playing around that goes on. I think the challenge that you get- We call it ghetto servers, a ghetto data center. That's actually even worse than commodity. And I didn't think I'd find something that was, I'm not sure I appreciate being called ghetto servers. They're like boards, Michelle's just like, what is that? You know? It's below commodities, like above plankton, you know? You have to bring fans in and sit the fan next to it to keep it going. Plankton, ghetto, commodity. And that's always going to happen where people want to play around. The question is at what point, like I know guys who are hosting out of their garages, but suddenly the minute that they start selling the service, even if it's to some, you know, lawyer down the street or some local business, suddenly now they care about a little bit of UPS protection and now they care about, you know, having parity on stuff and making sure it's really going to run because they want the guy to send them the check every month. So it's kind of where you are in the food chain. You personally will adapt to fit whatever you can afford to fit, just like you do with the car you drive. You're having this conversation with the MongoDB president, his name's Max. And Mongo scales well for lamp developers, but all of a sudden the breaking point is a little bit higher up now. They can scale up to the point where they got to re-engineer. So that's well documented. There's a hardware failure model, right? It's like, okay, I'm hosting on Amazon. Everyone, Zing is a great case there. They did Amazon, then they went private. Where's that scale point for hardware in your mind? For those kinds of rapidly growing companies, you know, crazy game company goes from zero to, you know, drawing a scale, for example, huge growth company went from zero to 50 million people like literally in weeks. Well, I think, I mean, I think Netflix is a perfect example. They're just announced they're totally getting out of running any more data centers on their own. They're going to go to Amazon. I mean, they announced it two weeks ago, I think. And you would think if anybody could have the scale, it would be somebody like Netflix. But the flip side of it is, what's your business? Is your business the content business or is your business running data centers? And where suddenly have you changed so that now you got this focus of I got to worry about pipes, I got to worry about networks, I got to worry about, you know, peak load. I mean, that's an art and a science of its own. And you want to, you know, find people who are going to do that for you and make some mistakes the hard way so you're not online some night between seven and nine or you're going to hire somebody who already is perfected it. And I think you see a lot of that going on. Well, you think of the big switch, right? Yeah. But at the same time, not a lot of people are, you know, big scale guys are running the business on Amazon. I mean, Netflix is a clear example, but Zynga's gone the other way. Right, right. So which way do you see the trend going? It's going to be just like computing as the day. I mean, you know, you've got, you've got corporations, some of them still think mainframes are the answer. You have corporations who are totally running private clouds on, you know, Xeon boxes. That's the beauty of this technology. I mean, you know, there's a solution for everybody and I, you know, are there going to be some people that do a little better and some people that do a little worse? It's, you know, it's where do you want to put your cycles and who your core, I think a lot of it has to do with your core manpower and whether you've got people you know, did you hire a bunch of CMU grads along the way and you don't want to lose them or the Stanford kids and they still want to keep doing this for you or do you have to hire them as you grow? At which point you say, maybe that's not what I want to do. But a couple of years ago, it looked like the cloud service providers were really outpacing that, you know, enterprise IT. It feels like the gap has closed a little bit. I mean, I know CSPs are still growing faster, they're innovating faster, they're investing, but the enterprise guys have really got their act together with private cloud, haven't they? Has that gap closed? Well, I think what the enterprise guys have had to do, there's nothing like your CFO being a force of function and say, bring me a make buy. I mean, Diane did it when she was RCIO. Should we be buying cloud outside or should we run it in our own cloud? You know, and she went in and showed him what she could do for the same or less than going outside and still have the protection and everything else. And that's why we've got our own infrastructure cloud. So you have to do that make buy analysis, but I'll tell you who loves to buy. I mean, when we were at Penguin, when you're in a startup, you know, you got 80, 50, you know, 120, 500 people, pay as you go is a godsend, you know? You have a bad quarter, you just crank down those Salesforce seats, you know? You can cut your bill in half. I mean, where else can you say that? And that's, I think, the piece of the market that really, really responds well to that kind of a service offering, which is you have an enormous amount of flexibility. You don't have to, you can take advantage of having Salesforce or SAP or anything else you can buy as a service and really literally not have the capital and not have the physical plant, not have the people just, you know, pay for it as a expense budget. What was that? You mentioned Diane a couple of times, Diane, right? What was her message in the keynote this morning? I presume you watched? I think her big number one message was that IT is going through a major transformation and that these guys, exactly what I just said to you, which is the competition is the cloud and you've got to show that you can be as efficient as agile, you know, at moving as they are and you've got to demonstrate it to your management, but then she got into doing it as the three components we've been talking about, which is it's not just servers anymore, it's servers and storage, it's networking and infrastructure and that you've really got to use all of the technology you can and, you know, then she talked about Exa as a great example of working with a partner like Oracle where we've helped put all of those components together to demonstrate that, but that you as an IT person have got to do that in your own shop through a variety of ways, whether you decide to go cloud, whether you decide to build a private cloud, whether you decide to buy an appliance, you know, the efficiency demands and the growth demands and everything else aren't going to go away. Do you see that, I'll call it a hybrid cloud, but by that I mean specifically a federated hybrid cloud, you know, where you're federating applications. Do you see that as a trend that's taking hold and what's your update on that? I still don't see it yet. I'm waiting like you guys are, but I don't think we have all the pieces. What's missing, you think? I think what's missing is probably a combination of three things. There are some technology problems we still haven't solved in terms of the security and the protection. I think there are some standards that we've got to get a critical mass to buy into and they're starting to emerge, though obviously some players will support standards and some players won't support standards. That's going to happen. And then I think the last one just has to be that you have the confidence and the ecosystem to solve your problem and put your business there. Do you think that Oracle made some inroads in some of those three this week? Yeah, I do think Oracle made some inroads. But I also think that I think, getting back to what I said earlier, the biggest tension in this system is going to be open versus, I'm only going to say proprietary, semi-open. And where does that land? And I haven't seen the answer yet. But Oracle's value proposition is, hey, if you go all Oracle, you can do that. Oh, exactly, and that's exactly what Microsoft's proposition is to be and exactly what I'm here for. And VMware is going to say the same thing with this part of itself. So you've got these stovepipes of hybrid cloths. I said tension. Beautiful. Let me ask you a question about big data. When we're running out of time, you've got to go, but I'm going to wrap out. We've got plenty of time in the day here, but big data. So the big debate that everyone has, I mean, we talked about Larry's presentation and we were kind of critical of it, but there's a big confusion around what's the hard tech involved, right? So let's talk about the big data demos you see. You've seen our little tool there, vFinder that we're building. And Dave and I were just talking about this. It looks easy. Get some Twitter data, get some little data, make it work story, bring it in real-time. What is the hard tech about real-time analytics? In your mind, from your personal geek perspective, thinking about the complexity, help people understand where is it easy and where is it difficult? What's the- And he means geek is a compliment, by the way. Okay. Well, I'll tell you what I think is the hardest question. Tech athlete. No, that's okay. Geek is okay. The hardest one in my mind is what I'll call real-real time because you now have got near real-time and relevant real-time and all these other things that people put in that say, well, I used to do it in 30 days and now I do it in four hours and that's real-time. Real-time, to me, from the strongest arbiters of real-time, and I give you an Intel example, is when you want to use big data to instrument a production line. And you literally want to keep that production line within operating margins 100% of the time. I mean, at Intel, the way we measure our plant managers is that we calculate the maximum number of chips that should be able to be produced by that plant if it was working 24 hours a day, 365 days a year and was yielding at whatever the highest possible yield number was. And anything less than that is a failure, okay? So when you- Boy, I was lying for that job. If you're a fab guy, it's the most challenging fab in the universe. But this is also complicated today that you don't have people who watch it. You keep people out. People are contaminants. I mean, you know, one skin particle from you is not something we want in the room. But what you have to have is this all monitored electronically and the minute any piece of production equipment starts to vary out of parameters, you need to fix it instantly or what you're going to get is some chunk of chips in the middle that were processed during that period that are bad. It's the same thing with people who instrument wells. I mean, I had an energy guy tell me that he told instruments all of their wells these days because if they can keep a well in the ground four extra days or five extra days or a week before the bottom collapses, they will more than pay for any computer system you can sell them to do that analytics. What they used to do is used to put their best people out there and listen and make guesses about what was going on at the bottom of the well and now they don't do that anymore. They totally instrument it and it's down to minutes of precision. That is real, real time to me. But the consequences to them are what? If they don't have it. They're going to lose huge amounts of money. So that's where you see that. So then you move from that to what I call near real time, which was I was at the SAS conference. Macy's wanted to do one very big thing this year. Every retail establishment in the world up to this point in time has had internet inventory, web inventory and store inventory and they were like two different countries. And if you sold out on the web your only recourse was to march a little button store and find it in the store. And what they realized was they had stuff they were selling out of on the web that wasn't moving in some stores. It was actually being marked down when they could sell it for full price over here. So their goal was unify my inventory so that my inventory is one big picture and then I'm going to create 240 or some number of superstores that will actually ship directly to a web customer. So if you find it and it's in the store it goes from the store directly to you. It doesn't go to a merchandising center and they wanted the ability to reprice the entire inventory in four hours. So let's talk about the tech involved. How hard is this? What's the hard thing? Is it the database? Is it the compute? Is it the everything? Well, I mean they would tell, yeah, I mean the first hardest problem was just unifying the data because I'm sure they had two separate systems that did it and it's a big volume of data. But the second one is I think the analytics and the timeframe. What do I mean? Because it's not just reprice it, it's intelligently reprice it. Okay, notice that it's sold out on the web so I don't want to mark it down in the store or notice that it's the Friday after Thanksgiving and by God I got this one thing that ain't moving for crap and I better lower the price by 50% before tomorrow morning is the kind of intelligent pricing that you're looking for, that kind of stuff. So that takes a lot more smarts and a lot more analytics than just saying, I'm going to, you know, I'm going to, you unify the inventory and unilaterally reprice everything 20%. That's the dumb way to do it. What you want to do is selective markdowns that keep stuff moving but maximize your revenue. Okay, well, final question that always gets for you here is what's on your mind going forward? Share with the folks your vision for the next year between this Oracle Open World and the next time we sit down. That's an interesting one. I think as we've been talking about, you're going to see a lot more move to use SSD or to use Flash to continue to speed up that processing. I think you're going to hear certainly from Oracle next year, hopefully a lot more everywhere about software defined networking. I mean, that's one that's just getting kicked off now and I think everybody wants to jump into because it's kind of the next bottleneck that people see that they've got to get their hands around and a lot of people working on it. So that'll be an interesting piece to see what it takes to actually make progress and how quickly people are willing to adopt it after they start coming out with some solutions. I mean, I think, I don't know about you but I think the Flash stuff has moved pretty quickly. It took a while for people to kind of jump into SSDs and believe the price points were okay, they were a little nervous about it but now it's like everywhere. I mean, it's just everywhere and hopefully we'll get software to find networking to that stage. Well, I think in the case of Flash, it's about the business case too and it's starting to become clear, isn't it? Yeah, exactly, exactly. Pauline, fantastic having you on again. Thanks very much for all your time. Pauline is executive at Intel, senior executive, business person and also she's got a geek technical side of her, wouldn't be comfortable in the tree way. It's always good to have technical chops and being executive at a big company like Intel. So thanks for coming on theCUBE. This is Silicon Angles theCUBE. We'll be right back with Oracle, more Oracle Open World coverage right after this break.