 Okay, we're back here live at HP Discover in Frankfurt, Germany. This is theCUBE, our coverage, exclusive coverage of HP Discover for two days live and to get all the action and strike the signal from the noise. I'm John Furrier, the founder of SiliconANGLE.com. I'm joined with my co-host. Hi everybody, I'm Dave Vellante of Wikibon and we're here with Chuck Smith, who's the vice president of Blade System and Business Planning for ISS and software at HP. Chuck, welcome to theCUBE. Thanks guys, appreciate it. Big event for you guys, about the same size as the one in Las Vegas, right? Yeah, actually a little bit more. Is it a little bit bigger? Yeah, yeah, yeah. Around 9,000, so you guys reach close to 20,000 in these two events. That's right. Over a span of six months, you just announced next year it's going to be in Barcelona. So one of the things that we've been covering, obviously that's come out of the first day and then first half of this day is obviously big data is native across all of HP. You're seeing the big data seeds being planted, whether it's autonomy, a little bit of HP lab stuff here, you got Vertica and a variety of other software is creeping into all the little nooks and crannies of HP and system software services. So big data, obviously the big theme. Yep. On the compute and infrastructure side, you have the converge infrastructure modernizing with Flash, the new storage stuff. And really the key thing that we're seeing is this software-led infrastructure vision. Wikibon just posted a groundbreaking research report, first of its kind, modernizing the converge infrastructure definition called software-led infrastructure. So Dave Donatelli was up on stage and he might have known her, not known, but he has lightning in a bottle because he said HP is delivering and shipping software-defined networking, software-defined servers, and software-defined storage. That's right. In all three categories, it's a triple threat. No one can claim that right now in the industry. That's true. So I want to ask you, what is software-defined servers? Because you guys have announced Moonshot, which we covered. You've announced Gen 8 in Vegas. We covered that with video. So we know that's going on. The compute side is getting faster, faster, faster. But software is being abstracted out as the key differentiator, making it all work. So yeah, more cores are coming, power efficient, all that's great. So give us the update. So hardware's not going away, but software still is the focus. We call it software-led infrastructure. So tell us, what is software-led servers? Well, I think that one of the key things that Dave spoke to is, when you think about Project Moonshot, one of the key elements, even in the hardware game, is ensuring that you've got the right compute infrastructure to support those applications. And we see with cloud-enabled data centers, public cloud, and so on, a dramatic shift in the requirements. And so what Dave's talking about is, and what we're delivering on, and you'll see us roll out over the course of the next 12, 24 months, it's continuing to support that from a software-led server design. So Project Moonshot, we actually have the ability to tune cartridges, if you will, to support specific application footprints. In another, if you will, approach, we've been shipping cloud system for three years, and cloud system actually allows us to deliver on a cloud-based architecture for infrastructure service all the way up to the application. So you can then take, if you will, resource pools, bring together three years, yeah, 2009. We actually announced it at Discover in Berlin in 2009, and it's actually a little more, April of 2009. So we actually have the ability to take those abstracted compute storage and networking resource pools, bring that together with our cloud maps, and essentially deliver that application. So essentially defining, if you will, the logic or the software to provision the server or the service. So that's kind of a couple of ways we're coming at it, depending on the requirement. So quick follow-up on the cloud maps. Could you take that intelligence and embed it over time into the systems and actually proactively act on that data? Yeah, I mean, so cloud maps, for us, is really a template for what that infrastructure and application needs to be to optimize for an app or a customer in the enterprise, or for that matter, a service provider, what have you. What we do have the ability to do is store that, if you will, that logic of how that infrastructure and application comes together, and then manage it over the lifecycle. So we certainly do have that intelligence to not just deploy it and provision it and set and forget. Static, right? But yeah, but no, absolutely manage it over the lifecycle and adjust to resource requirements, changes in the elements associated with the application, all sorts of things. So yeah, absolutely. And that adjustment is obviously fully automated, right? That's right. Or largely automated. That's right. You've got to do the implementation correctly. It's based on a combination of what we're doing within the enterprise group and with our software group. So a lot of what Meg talked about was bringing together the infrastructure elements and the software together to deliver Converge Cloud. How is HP addressing the notion of the solution sales versus the traditional box sales? Can you talk about that a little bit? Yeah, I think that, well, I mean, obviously, as you said, the hardware's not going away. So we have to have folks that really understand where our differentiation is and understand how we bring these things together. What we're trying to do is focus them on key solutions that deliver value from a Converged Infrastructure perspective. So whether it be blades or the recent storage announcements, the work that we're doing in software to find networking have really clearly specified goals and target hardware elements in the Converged Infrastructure that makes it easy for our customers to understand, makes it easy for our partners to understand what to sell and from our sales teams to bring it together. It's a bit of a journey. So with an infrastructure sales team and bringing them up, that's what we continuously work on. But if you look at our success, I mean, we are the blade leader and have been for well over five years and we've shipped over three millions of blade servers. We've got a significant amount of differentiation there and really is the platform for Converged Infrastructure. So we've been quite successful. You've touched on it, but I wonder if we could summarize that differentiation. How do you guys? Sure, I mean, if you look at it, what we try and talk about is that we have, and it's linked to our value proposition associated with Gen 8, the intelligence associated with provisioning and managing the firmware and the change over the life cycle. The sheer capex costs associated with buying blades up front, your break even is somewhere between, zero and six servers. The connectivity, the work that we've done with our storage team, the flat sand, where we actually directly connect into a storage area network. That takes out costs, improves latency, takes out management overhead. So those are just some quick examples. Virtual Connect is our technology to connect servers to networks and that's been very successful. Over seven and a half million ports out in the industry and really changing the way in which people connect the server to the network. So Gen 8's been out for a while and one of the value propositions of Gen 8 is you're going to attack these mundane tasks. We all know that we have a very overly intensive IT labor component, especially when it comes to managing servers. Do you have data, improve points that suggest that you've actually been able to attack that problem? Yeah, no, we have significant data. I mean, it's in a lot of the marketing and literature we put out in the public domain, but just some examples. I mean, we've been able to improve the update timeframe by 3x from both our competition and our previous generations. So compressing that timeframe. Compressing that timeframe, absolutely. Another key element is also over the course of the lifecycle of supporting the infrastructure and we've actually created the ability to diagnose and manage and capture information on an ongoing real-time basis. And so then if you run into a situation where you've got a server failure or a component failure, we can actually rapidly diagnose that. We can actually phone that home to our services folks or a channel partner and we can actually dispatch the appropriate part very quickly. And we've shown that we can actually improve the response time and essentially coming back up to production by 66% faster. So these are just quick elements in terms of what we can do with Gen 8 that nobody else can do. Chuck, I want to talk about a concept that's being kicked around. We were just at Oracle Open World recently and all the events we've been bringing the Cube to. So you get to hear all the rhetoric from some of the software guys trying to be hardware, not Oracle. We like to give Oracle a little kick once in a while. Well, all the time. But because they're in the hardware game. I mean, it's trying to be like HP. It's face it. Larry wants to be like HP. But obviously they have Oracle software. The big conversation that Oracle SAP and even your partners is workloads. So one of the things that's really impressive about what we heard at Gen 8 and Moonshot is tuneability. Talk about that going on. Because the power and cooling stuff, that's energy stuff. That's a different conversation. Very, very cool. Different conversation. But people want to manage their workloads. They're deploying virtualization. You got all kinds of different requirements now for mobile and cloud. Workloads are changing. You still have that SLA, that enterprise grade workload specific hardware. The old days you stack and rack. I got rack for this. I got a rack for that. Now you got virtual machines. And this tuneability becomes a big deal. So talk about what you're seeing relative to the notion of I got some infrastructure. I want to build it up. Stand up this infrastructure. But I want it to be flexible and agile from a workload perspective. Right, right. Well, so I think a perfect example of what we're doing there. We just recently announced the product called the SL4500. And back to your question on big data as a perfect example. Depending on whether it's a Duke cluster or an exchange server or something else, you really have a wide, if you will, range of workload requirements. And one of the things that we've done is create, within our smart storage strategy, caching algorithms that actually allow you to adjust the cache that's used on board relative to the solid state drive and give you the ability to tune to the requirements of that application either more or less than actually accelerated. And so we demonstrated that on stage yesterday with Mark Potter's introduction to that product. And we were actually able to show a dramatic improvement in terms of the, if you will, transactions per second or the performance profile based on our ability to capture the workload parameters and then adjust cache to meet that. So that's a great example of how we're doing that. And you'll see us extend that out to the storage host and really create a broader view of how that's handled, so. Yeah, and we're seeing also with that tuneability is that purpose-built hardware or purpose-built solutions are, we call it the tailored suit model. Tailor suit, you wear it all the time. It only fits that one, your one size. Like we were talking off camera about VCE, that's on the high end to really bring that kind of capability. And what do you think about those conversations when people say purpose-built solutions around hardware? Because you've got hardware and software. Well, I mean, we could argue whether the VCE elements are purpose-built or generalized infrastructure. But what we see, for example, cloud system, we see as purpose-built for delivering a public or hybrid cloud into an enterprise. It's very flexible and the ability to handle a wide range of applications within that given the cloud maps and what we talked about. We also see kind of another trend with, and this is what Project Moonshot really looks to address. You look at some of these public cloud providers that have relatively homogeneous application profile, but deployed at scale. That's where purpose-built hardware starts to get very interesting because you didn't have a very, if you will, deterministic tiered stack that the hardware can address. And so there's sort of two things going on in the enterprise. Heterogeneous applications, a need to be much more flexible, and then these public cloud providers which have really a different orientation. So say big data, for example, like if I want to have density, I need a lot of servers with a lot of RAM. I might not need a lot of storage, so I'm going to need maybe having different configurations. Right, so I mean if you look at our SL 4500 that I talked about, we actually have a one by 60, so you got one compute with 60 drives, really perfect for an open stack implementation, or if you will, a base level cloud, all the way over to a three by 15, which is much more compute intensive, less drive intensive for things like a doop, and so on. So it just depends on, and we actually have that configuration. So do you see, do you see, if I follow up on that, do you see things like Moonshot eventually fitting into that type of architecture? I mean, I know it's early days and not going to ship it here. I mean, so if you think about Moonshot even so, you actually have, we have this concept of a cartridge. So now we can actually have a cartridge that has memory intensive, or multiple cores, single core with more storage, depending on what that application is. And that's really what Dave was getting at in terms of software defined servers. The software is driving that purpose built design, but we built an architecture to be flexible around what the software requirements are. I wonder if we can stay on competition a little bit. So you see Oracle get into this market as a partner and another competitor. Well, I mean, the Exadata is a converged infrastructure. They're selling them. So let's go down the list. I mean, what's your bumper sticker and how you differentiate, how you compete? Well, certainly we look at Oracle, for example. I mean, if you look at some of the performance benchmarks, you look at some of the capabilities that we have with our converged infrastructure, we actually, if you will outperform Oracle and we can show those benchmarks. And so, you know, I think one of the tenets of the purpose built infrastructure is you actually got to be better somehow. And when we're better with our converged infrastructure than they are in their own stack, that makes it pretty an interesting discussion with customers, so. You're great. So that's cool. And then IBM made a big entrance in the market. I mean, it's blade business was hurting. Yeah. They've essentially replaced the blade business with a pure line. Yeah. Impressive marketing, for sure. Yeah. You're smirking, but okay. Well, and then this whole thing about patterns and getting knowledge in there. So that sounds good. It's IBM, you know, they're big and global. So, what do you see there? Well, when I see that, I look at the pure system and what we've looked at in terms of just the base level infrastructure, where they are with the, if you will, the fabric and how far we believe they can go, what they're launching in terms of the infrastructure. We still believe we've got a better mousetrap. The differentiating items we talked about. We see patterns as being very similar to what we're doing with Cloud Maps. So we believe we've got an extensive lead there in terms of, if you will, patterning or having, if you will, a topology for deploying applications and having it very well scripted, best practices and so on. So we're doing good there. So kind of more of the same, the devil you know, and you know how to compete against IBM. And then we talked a little bit about VCE and you're saying, you can argue whether or not it's truly integrated. I look at the public financials and I kind of scratch my head a bit. I do see them very aggressive. We see it primarily EMC centric. I think with the advent of FlexPod and some of these other pieces with Cisco, there seems to be a breaking ranks a little bit. And so we're wondering whether the consortium itself has legs. The flip side is that we believe with our cloud system and the platforms that we have are much more flexible. We give customers choice. You can start low and build up. You can leverage your own infrastructure they already have. So we can actually deploy a cloud system in an existing blade environment. We can connect to other people's infrastructure which customers require. So it's heterogeneous, not a lock in. So there's a variety of things that we think we differentiate. The converged infrastructure's place used to be lonely enough for you guys. Now everybody's in Nundell, of course. That's right. Well, it's the finest form of flattery. So Chuck, my final question, we're getting the break sign here. But you know, this is a great conversation. I'm going to give you a use case. So you guys have a great footprint on the enterprise but I was talking to a few folks and the startups that grow really fast like the Pinterest of the world. They start out buying commodity, not commodity, we call commodity, not industry standard. You know, they make their own boxes, they do a prototype and all of a sudden they need to scale out really fast. So the use case is a big data startup builds a couple of node cluster they build their own boxes and all of a sudden they got to just grow like crazy. So here's the use case. They need maximum boxes with an unbelievable amount of RAM and as many cores as possible with this being kind of in there but maybe put a different disk system in there but they need to ramp up as fast as possible and build the biggest infrastructure for that spec. Ton of RAM and a ton of cores. What does HP offer that? Is it a moonshot? Is it different servers? What is the ideal solution for that kind of use case? That seems to be like the Hadoop use case. Yeah, now we actually see, I mean there's a variety of answers depending on actually how they're using Hadoop. We actually see that the SL4500 platform being that really purpose built for that Hadoop cluster you can scale them and add them together so you actually have a quite a capable platform so very excited about that. We think ultimately that- It's RAM intensive? Yeah, it can be. That's why I said we've got a variety of configurations but yes it is and you can actually start to incorporate the disk caching to create a much more accelerated workload as well which is interesting from a Hadoop perspective so that would be our pick to deliver Hadoop and big data but it's not the only one. We actually think that for specific deployments for example Memcache and so on that moonshot actually has a very interesting architecture so it really just depends on again the software and what you're trying to get at. We actually think the moonshot was actually the purpose built God box for big data because the way the amount of service you can put into it was just amazing. Power, power, obviously. Yeah, I think that that's true but what we see is that in some cases a requirement for fairly large data set so I mean we can go up to 180 terabytes in the box with the large form factor drive. It's disk but it can be a mix of SSD and spinning media so you could actually have that combination. Of course the moon box of the SL. That's actually the SL 45. Yeah, okay. Yeah, you know. That's more versatile, more flexible. It is. And it's actually. I think that's not shipping yet, is it? We actually have it at selected customers. Shipping and production. Yeah, generally available. Generally not generally available, yeah. But lots of conversation. Hey guys, thanks a lot. Hey Chuck guys with the servers and obviously you know, if you like hardware, man, hardware's not going away but as we said in our blog post it's the software and the hardware. Hardware's a great delivery mechanism as Dave Donnelly says and you guys are converging it all together. Software led infrastructure, software led infrastructure's the hottest area. You guys got the software defined server storage and networking. Congratulations, thanks for being on theCUBE. We'll be right back with our next guest after this short break. Thanks guys.