 Okay, we're back here live at HP Discover in Germany. We were in Las Vegas a couple months ago with HP, documenting and extracting the signal from the noise about HP's comeback turnaround, whatever you want to call it. Certainly they got a lot of action, a lot of great products here. We'll go right into it. I'm John Furrier with Dave Vellante at Wikibon.org and we are here with Pauline Nis, CUBE Alumni Pauline. Great to have you on the CUBE again. I see servers are doing very well on the Intel side. So, and the server business is exploding. Although we were saying software led infrastructure, which is, it's about the software, not only the hardware. We didn't say software, not the hardware anymore. Hardware is still expanding very rapidly. That they put more cores on. So, my first question to you is, explain to the folks out there the dynamic going on between software and hardware. Relatives who come in like HP, because a lot of HP customers know them as hardware vendor. They sell a lot of hardware. It's a delivery mechanism. They still need more cores. But talk about the role of the software. Well, I think a really great example of the role of the software is the new storage box that HP announced with the autonomy software built in. And they've also got the store ones with the deduplication. You know, our core message is, that's a perfect example of why you need the horsepower of Xeon. Because if you're going to put all of that calculation on the storage side, whether it's deduplication or whether it's the autonomy search capability, you want the CPU horsepower to back that up so that people are getting the performance that they want to get out of it. So, we think that there's still a pretty unending need for performance across that space, particularly as the software and the hardware come together. Now, you guys also talk about software, but Intel never puts themselves out there as a software company, because you harden the top, you put it in the silicon and chips. But the big focus of dynamic workloads like Gen A, I just retweeted my post from February, is a dynamic workload kind of model. That's the future. You're seeing purpose-built components around like Moonshot, that's workload dynamic. How does it affect your side? Well, I think what you're seeing from us, and you're going to see more in the next couple of weeks, is we're also giving people the ability to choose what matches the right workload. We just announced that Supercomputer, our Xeon 5 processor, the many integrated core architecture, which is really geared at high-performance computing. We are now going to have atom servers that are going to be coming out pretty shortly. So, we understand that based on the workload that people want, the price point they want, really the application they're trying to solve, that they're going to want a choice. And we think that's great. I mean, we think that they'll get me wrong, a great place for general-purpose computers out there. I think the storage boxes from HP are a perfect example of where you need the I-O in a storage box, you need the compute cycles, as you put more capability in the storage box. But we also understand that you've got everything from web-serving to high-performance computing in the marketplace. So, can we talk more about this whole software-defined, software-led thing? I mean, for years I've kind of observed, there's always software choices, you know, software raid, or I think what Veritas was doing in the early days. But it was always, you know, the tight coupling of the software and the hardware that gave you the most business value for the most demanding applications. Is that changing with software-defined? It is, because you have to sort of follow the progression. I think one of the things that's happened as a result of virtualization is you see people want to be able to tune that configuration. And right now, it's not always easy to tune it because you've got these black boxes often that are providing storage, black boxes that are providing networking, and the virtualization engines and the operating system would really, really like to be able to do service-level agreements that demand doing that. And I think that's where we work with HP a lot on the whole issue of open. We'd like to see open standards, whether it's for this software-defined networking. We'd like to see more open storage configurations where storage is really a server, and you can go in there and you can look at the Linux or whatever operating system is running. And I think you're going to continue to see a push for that across the board. And I don't think it's going to be easy. I think that people run a lot of legacy stuff that's going to be around for a long time, particularly where they're bringing up new applications. There's going to be pressure for more open, more standards, more software definition. Well, we might have talked about this a little bit at Oracle Open World, but I want to get your angle on this. So, given that, don't you see or do you see the server as becoming increasingly the control point? In other words, we're talking about storage, the last 15 years we've seen function move from the host out to the sand. Copy services and migration, whatever, replication. A lot of that function with Flash is now moving closer to the server. And it seems like you can't manage fast servers from slow disk. How is that changing? Where does metadata fit? What does that mean for your business? Well, I think two things. I think you were going to see, and we've talked about this before, that whole continuum from memory to storage changed dramatically in the next five years. And I do think that those slow disks are really going to become archival devices, that you're going to see people, you know, you're going to see this continuum from nonvolatile RAM that sits at the memory tier to SSDs that sit at the storage tier, and then, oh by the way, you know, there's going to be where you store that stuff, which isn't tape anymore, but which is rotating device. Somebody said to me the other day, it was down at Strata that Hadoop was the new tape. You know, so. I know, yeah. But the flip side, it actually is very funny because I was telling somebody a story that I heard from a customer. It all has to work together, and I think in the U.S. we've still got some backbone bandwidth issues to deal with. I had a customer, a medical customer, tell me that they ship data to the National Cancer Institute cancer database every week. And you know how they do it? The cheapest way to do it? They put hard drives in a box, and they give them to FedEx. So who knows? The hard drive may survive as the interchange media of choice across long distances. You know, the old Chevy Truck Access message. Love them up, baby. That's right, that's right, that's right. Pauline, one of the things that people are talking about is like redundancy, automation. Self-healing's kind of been around for a while. What's your take on this whole self-healing flash trend? I think we're going to see new algorithms and new, you know, just the example that I would give you is main memory today. You actually don't care about the fact that those ECC bits are fixing what are constant failures in memory. I mean, you can't build the density of memory devices we have today without failures, but the user never sees them. All you ever care about is if the failure rate ever gets too high, somebody comes and replaces the device, but the ECC codes are hiding that from you. And I think you're going to see more of that crossover into the storage domain, that you're going to be able to see better algorithms, you're going to be able to see things like erasure codes. I mean, when you just look at the fact that you can put huge computational resources on everything from encryption to some security features to the kind of erasure code features. You talk about computes. There's currently going to be a need for computes there, but it's going to be to make the solution overall a lot better. Well, there's a erasure code trend that's kind of interesting, right? Dispersing data, very math intensive. You must hate that. It's, hey, just like ECC, we love it, you know? People always say, you know, because there is a lot of focus now on lower end processors and what's going on in some of the competitive world. And I don't see the need for computes going away. I mean, I see these workloads that are going to be optimized for different configurations, but just like we've seen to expand to fill all available storage and Hadoop is doing a fine job of that, I think the same thing happens with computes. I mean, I think there are new algorithms, there are new solutions, there are much cleverer things that people come up with when they have those cycles. So remember back in the web app days when you had accelerators, application acceleration. So, you know, we're kind of hitting this new era where workloads and all the purpose-built stuff we were talking earlier with GTA and Moonshot, all this transformation. So, IO, we've always been fixated on IO. David Floyd, IO-centric infrastructure is a big focus area. What are you seeing with IO, the IO issues at the chip level and within the servers because everyone wants real time right now and real time in analytics is really exposes the scabs to a lot of solutions out there because, you know, now with big data and some of the stuff we're seeing with autonomy and other apps, you know, walking the talk is now measurable. You know, say, if I can't get the data out, the app sees it, it's failure. It's like, I know is the key to that. I think you're going to see, you're going to see continued use of SSDs. You're going to see continued use of other PCIe cards that put the non-volatile RAM there in the short term. I think in the long term, you're going to see a next generation of NV RAM that virtually everybody around the world is investing in trying to get to, that's really going to blow open the memory tier so that you're just going to have huge, much lower cost NV RAM available at the memory tier level. And that's the only, you know, IO really up to this point has not kept pace with Moors Law because it hasn't been a semiconductor device. You have got to get it over, you know, from not just main memory but to having the next chunk behind it be semiconductor technology that's accessible via memory controller to really give you, you know, an order of magnitude bigger main memory and then still the SSD technology behind that and then, you know, some backing story that'll be your archival device. So Jim Gontier talks about this notion of separating the control plane from the management plane and software defined, which I love the fact that they're getting behind that trend. I really like it. I think it's relevant and it's good for HP to jump on something relevant rather than making up a crazy word for it. And in Donna Telly's group, they're doing a good job there. The data center operating system. I mean, software defined data center is basically the data center operating system. So obviously you guys know a little bit of the operating systems, you know, working with the kind of components you guys build on contain whether it's a PC or servers. What do you see that, how do you see that vision playing out? He's got one architecture except we're in the control plane versus others are taking a different approach. Well, I think you've got all the major players that you would expect to see there, whether you call it a software defined data center, which is actually a term VMWare uses in addition to HP, or whether you call it a cloud operating system. I mean, there are, or whether you look at a lot of the open stack stuff that's going on, people are trying to figure out how you move that level of orchestration up a tier. So, you know, it's not just really running in the rack. It's running across the data center. And I think it's going to take a little while to get there. I mean, I don't think it's easy because the reality is it's taken us a long time to get to operating systems being reliable, dependable, you know, fault-tolerant, bulletproof. And people are going to be looking for that at the next tier too. And so you're going to see a variety of people try to jump into the deep end of the pool there. And I think it's going to take a few years to settle down. Yeah, we got to get there with you guys doing a great job. Pauline, it's great to have you on the queue, but my final question is for the folks out there, you've been on a CUBE alumni, you know how we work. We like to strike the signal from the noise. I've been critical in the past about HP with this turmoil at the top in a constructive way. Hopefully, I don't make too many enemies over there. But it was deserving. But the company is very healthy. And so I want, at least we're seeing that and we're reporting on that. From your perspective, you've been around the block and the industry. Share the folks out there, what's happening here at HP Discover? Everyone wants to know, I mean, HP's in this cloud right now of a lot of negative press. They have a black eye from that whole autonomy thing and the EDS and it's really horrible. A lot of people are rooting for the women to win as Meg Whitman said. And there are, and I've known people here, I worked here nine years back in the day and I know people who are still here 30 years at HP, great people and great company, great customer base. So they're not running out of cash. They've got a great product portfolio. What's your take on it? Not an Intel spokesperson, but from your personal perspective, share the folks what's happening. I agree with you. I mean, I think this autonomy thing's been a big distraction particularly since it's a balance sheet issue. It's not a cash issue. The company still has as much cash as it's ever seen. It's just gotten its balance sheet straightened out because the auditors made it do that. But they have a huge depth in breadth of technology and I think a couple of the points that Meg has made that are worth making again, not just for eye wash, they're everywhere. I mean, if you're a global company and you want a solution, it's not like just deploying something in one location or two locations. You can use them around the world. You can use them for branch offices. You can buy servers. You can buy services. I don't think there are a lot of companies that have got that depth and breadth. I mean, there are only a couple of them that you and I could name. And I think people want choices and all of the ones that are out there, HP is committed to open. They're committed to pushing the envelope. Innovation. Innovation and I think it's a mistake for, it's very easy when the negative PR gets going to kind of jump on that bandwagon, but I think it's a mistake to write them off because they have a loyal customer base. They're deploying this stuff today. The customers aren't unhappy with the technology. And the exciting thing for me is they seem to be moving a little bit faster. Hopefully that's what Meg is doing in terms of rolling out products, being creative with some of the server technology, the storage technology. I mean, I think it was a real surprise to a lot of people to see the autonomy stuff in the storage box. I mean, that's a perfect example of what they need to do more of. Yeah, I agree. HP is not down and out. They are just a little bit bloody, but you know what? That's a balance sheet issue. And the stock market too. I mean, yeah, people lost, that's going to come back. So I think, you know, we're documenting the comeback. We're excited too. We also documented the good stuff, like Gen 8, we're at the Moonshot launch. So you guys are doing a great job. I just love the direction. I mean, I want that Moonshot box because for our Hadoop stuff, it's unbelievable. So, okay, Pauline, this inside the cube. Great to have you. We'll be right back with our next guest after this short break.