 Okay, we're back live here at HP Discover 2012. This is theCUBE, silkenangle.tv. It's exclusive coverage of HP 2012. This is our flagship telecast. We go out to the events and talk to the smartest people we can find and extract the signal from the noise. We don't care if it's a CEO, entrepreneur, analyst, CIO. We have them all on and I'm excited to be here with you with my co-host. I'm Dave Vellante of wikibon.org. And we're here with Craig Nunes who's the VP of marketing at HP Storage Group, a longtime CUBE alum. Craig, welcome back to theCUBE. Yes, thank you. Yeah, I guess you guys couldn't find anybody smart or a CEO, CIO, so you got me, right? Is that how it went? Yeah, you are a CUBE alumni, of course you're smart. You've been on theCUBE multiple times and great perspective. You know, the new line came up this week is nobody ever got fired for coming on theCUBE. Right! I love that. That was in response to a comment that said, I really can't talk about it, I don't want to get fired. Oh. And I don't think anyone that we know of yet has been fired. Not yet. Quite the opposite, guys. We can push that envelope. So you're just coming off a breakout session with the price line, right? Yes. So what was the deal? You had a customer case example? Yeah, so what we were talking about with the price line guys was the, so with all the unpredictability going on as people are moving to virtualization and IT as a service, how do you cope, right? And we're talking a bit about solid state technologies and it's a way out for unpredictable workloads, but money is involved with solid state. And so we are kind of focusing folks on the optimization approaches, right? Solid state's no good without great software and we talked about optimization within the box, right? Subland tiering, three par first on the high end to deliver such a capability. You weren't the first to announce it though. We were not the first to announce it. You were the first to deliver. Yeah. We kind of like to announce when we deliver actually. It's sort of a novel marketing approach. And then the other thing we talked about too, just because the world is shifting is not just optimization within the box, but optimization across the storage systems, maybe even across data centers. And we talked a bit about storage federation and how that can help you with that kind of workload balance. Pure motion. Pure motion was the topic, yeah. Craig, so I got to ask you, because you've been on the queue multiple times. This is again the second year we've done the discover with you. A lot's changing in storage. You mentioned flash and the whole price thing is interesting because even though it's a bit expensive relative to flash prices, the benefits are really strong if you could look at what's happening around the solutions, right? So, and storage has changed. You're seeing storage at the center of the cloud, big data, all the things that you've been talking about. What's changing, I want you to tell us from your view, because you were at three par now that you're back inside HP, there's all kinds of questions of what does tier one mean? Are you a tier one storage provider? And there's a lot of different competing approaches they say, we're tier one, so are you tier one and what is tier one mean? Yeah, so I am absolutely the wrong guy to answer that question. The right person to answer that question is our customer, right? Because it's not tier one storage, it's tier one data. Tier one, mission critical stuff, right? And they need to put it somewhere. And the whole idea that you've got tier one data now in, say, a cloud environment, what does that mean in terms of requirements for the storage platform? Well, still, you've got to have all the performance, the availability, the scale, right? For sure, but what other requirements does cloud kind of weigh on storage and what is maybe an older storage architecture not great at and what do folks need to think about? And it's a lot about, we just talked about handling the unpredictability, being very efficient from a cost perspective, and how you handle that, that's got to be baked into your storage architecture, right? So it's not just tier one for what I call the physical, tied down application that you can model, profile, test, and it will never change. You need, tier one's kind of different when that application changes every day. Right? Talk about the unpredictability piece, because that's, we were talking about this when we were in a different context of HBase being a tailored suit. Fits great, it's been great, but you try to change that suit and put it in someone else or use case, doesn't work very well. Talk about that unpredictability specifically around the data. Yeah, so, I mean, probably the most extreme example is when your IT is serving folks outside of your corporation. So you have a plan, got a budget, right? But maybe you have a big quarter, and suddenly there's a lot of business you've got to serve that is gonna stress what you've got. How do you handle new business without taking performance and service levels away from your existing customers? You know, that's a very real challenge for storage. You don't have a lot of lead time, you don't get paid until that application's up and running, you've got to solve it now. You know, in the old world, if I'm running SAP on your favorite monolithic architecture, never mind how long it kind of takes to kind of get the config right and buy it, but then you've got nine months of pilot, test, install, and then nobody touches it, right? Nobody touches it, that doesn't work, right? In this kind of model. A real simple example, right? VMs are born and torn down every day, every week. And VMs are their own workload generator, right? And you have busy ones and not so busy ones changes every day and your visibility of that is tough. So, you know, simple virtualization environment's going to drive that level of unpredictability. So that's the changing definition of tier one. Is that that tier one is not about what it used to be attached to physically, but handling unpredictability in a multi-tenant environment. That is the core difference. So, I mean, the traditional definition of tier one, of course, is the reliability, availability, the serviceability, mainframe attached, right? You're basically saying, okay, we've got those things, maybe you don't have mainframe attached, but we've got the RAS capabilities, but we have to support a different requirement for our customers. So you're sort of talking about a new definition of tier one. You mentioned multi, go ahead. And I'd maintain the mainframe attach, you know, where we're aiming. That is that predictable, you know, software on a fixed physical asset. You know, I think that's the old tier one environment and we're looking very much at the tier one environment for cloud, right? Well, you know, Meritz calls it the software mainframe, right? The x86 mainframe is really what, you know, the business that he's aiming to build, which is interesting, right? Because, and you think about the mainframe, it's highly virtualized and that it's very specialized as well in terms of the application portfolio. But, and so the goal, presumably that he has and others is to be able to support any app, any workload, you know, on x86 virtualized platforms. And obviously you want to support VMware, Hyper-V or whatever it is. But so multi-tenancy is an interesting thing. I was saying, I think it was David Scott. Yeah, it was David. And I've shown you, I've got the original red herring all marked up from three-par. I collect those things because I like to go back and John, you know, you're from California, I'm a horror, a storage hoarder. You're California, you know the term pivot, but a lot of times you read the red herring and it's nothing like what the company does. That red herring was, the term utility storage must have been used a thousand times in there. Now, we don't use that anymore, but it's really cloud. That's really, you design for those, you know, carrier class, you'd carrier class was used in there, utility storage. So I think three-par was very true it's mission, but the reason for my dissertation is multi-tenancy. Everybody says they have it. Everybody says they have cloud. Why are you guys different in that regard? Yeah, so let me talk about what we mean by multi-tenancy too, because I think folks are like, well, I've got more than one app, right? More than one user, it might not multi-tenant. Look, here's multi-tenancy. Being able to, as you consolidate, right, in storage, mixing workloads, they can kind of fight with each other, right? Best practices, you know, still kind of keep transactional, you know, database applications away from the sequential, you know, kind of applications because you're going to take service levels away. And as you add, you're going to, you know, adversely affect your existing customers. So what part of what we're talking about with multi-tenancy is handling that mixed workload and adding, you know, an unpredictable mix without affecting your service levels to your existing customers and future customers. And so, you know, being able to deal with transactional app, sequential app, simply, easily, without having to, you know, have separate storage arrays or different partitions, different directors to serve different people. That is what, you know, a big part of multi-tenancy is handling the mixed workload. Being able then to physically, sorry, provide some isolation around that data. We call it virtual domains, but it's like a virtual private array. Think of a virtual private network. Think of that for your data. Having some level of security, so as you're consolidating, your business users within the enterprise go, oh, okay, I get, you know, an administrator for that stuff over there can't get to my stuff, they can't even see it. Okay, I'm willing. Playball here on a shared storage resource. So multi-tenancy has an aspect of security, it has an aspect of service levels under mixed workloads as you're growing. You've got to have that, and it's the unpredictability that makes a difference. If it were predictable, maybe you carve out a little spot for your transactional stuff, your sequential stuff, but it's not, right? VMs every day, right? And we heard that from a partner at SHI this morning, they were like doing demand by the cloud, and they're seeing real high performance computing, and they use an example like retail, just the workload of demand, servicing just powering up and the demand there, so that's a key feature, so here's one. So there's an elephant in the room here. I just want to, you know, the cube, right? So we want to have it out here. So you guys are growing, David Scott said you're growing at 100% three-part, he's growing at 100, three-part on store once, I think he said combine it, growing at 100%, but I think he said that three-part itself is now HP's biggest storage product is growing 100% a year. You have contended that your primary competition is VMAX, not Clarion or VNX, but VMAX. Now VMAX, as we know, that marketplace is not growing at 100%, no, we're close. Let's say it's flat, maybe it's up a little bit, maybe it's down a little bit, whatever it is. You guys are growing at 100%. So now EMC would say, well, three-part is not tier one. They would focus on their fast, you know, automated tiering as a real advantage. All the knobs that you can turn in their environment, their proven history, a lot of good stuff going on. There's a great franchise, you know, the Symmetrix franchise, I'm sure agree with that. How do you compete with them? Yeah, so, you know, I get what EMC is saying, but the results simply don't wash, right? Cause if you look at, I think the most recent published quarter from the IDC guys, a quote, in the high end segment, we're up 49%, HP storage, it's up 49% in the high end price band. EMC down 10, down 10%, right? So I'm going to look. In the high end space, why is it we have hyper growth and they're shrinking, right? So their statement doesn't wash with me. The data doesn't support it. The, you know, the notion of fast, right? Here's the thing. We talked about unpredictable multi-tenancy. You've got to be built for this. This is an architectural decision, right? That's the market's demanding that, it's not it. And the issue is, you know, bolting on features, and we talk to our customers about this all the time, cause marketing guys can write great data sheets, but we can write data sheets faster than engineers can write code, right? It's the bottom line. And the end of the day, fast on a VMAX, on paper can look pretty damn good, but if it doesn't really provide that level of capability, functionality to handle that unpredictability, it's not going to work. And it's architectural, right? You can't add a feature and turn, you know, turn a Volkswagen bug into a Ferrari. It's the architecture. You got to start with something, right? And, you know, and there's smart guys at EMC. I love those guys, but they have an architecture they're living with today that doesn't do that, right? And a feature doesn't change it, right? That's the difference. Now, my next question is you mentioned the market data. That's good, you know, congratulations on that. Let's talk about backup appliances. So, store once, you guys have made some claims about performance. Clearly you're aiming at data domain. It's unambiguous. Yeah, I hope I was clear. And we've had Sean Kinney in the queue before and he's, you know, taken off the gloves. He used to work at EMC. So, you know, we love that, right? Because customers are trying to make that decision. They want to know, okay, let's do point counterpoint and then we'll make our own decisions. The data shows that, you know, EMC's dominant there. They've got, you can see like two thirds of the market. So, store once is growing 100%, but it's growing from a small base. So, how do you move that needle? Yeah, so the, I think, so step back for a second. Because deduplication changed our industry, right? It was a great thing, right? Folks were struggling with the tape backup window. And, you know, some really hot startups, you know, rolled in and a great rifle shot approach solves some big problems, right? Dispaced backup needs great deduplication, right? But you roll the clock forward, you know, five or 10 years and those companies have been acquired. They're part of a broader portfolio and our customers struggle with fundamentally having to deploy one kind of deduplication for, let's say, your branch, a different one for your data center, maybe a different one yet again for virtualization. And, you know, they're hamstrung in where they can actually do the deduplication, you know, doing it where they want versus where the architecture dictates. And, you know, there's other problems like, you know, single-node approaches, availability's an issue, people don't talk about restore getting your data back. Why do you back it up in the first place? To get it back, right? At some point, if you don't talk about that, there's a terrible deduplication tax in this first wave of deduplication. So it was great, but it created some problems that need fixing. And so our approach is, look, we're going to come into the market and handle those problems and really a new wave of deduplication. Call it deduplication 2.0, next wave, whatever. But an architecture two years new that had line of sight on those problems as it was being architected to handle those. One deduplication technology everywhere in your enterprise. So dedupe where you need it, replicate any place you like without rehydrating your data and marry that with what HV Storage has actually been pretty good at for a few years now, which is scale out architectural approach which solves the performance headroom, availability issues that folks are suffering through. So it's a magic combination. So did you talk about deduplication, how it changed the industry? John, did you interview John Toygo? No, I was watching them. You watching? Yeah. You know him, right? Sure. Not really love that guy. No, I don't know. But the things he says are really controversial. He says, oh, dedupe don't do it, thin provisioning, it's no good, it's garbage. His approach is just delete your files, just delete your data, that's, be efficient. Well, I wonder how big data is going to change that scenario because people, maybe this gold in them are a bit... I think we've heard, David, in talking over the past year, Cube, when this comes up about data backup, is that people, if there's any risk of their data, any risk at all, a hint of risk to the business, they will not delete their data. Because they have to keep it. At any risk, not even like a flinch or hint of a risk. Which is a whole other conversation that we should have at some point about tape and the resurgence of tape in more of an archive space. We're talking about dedupe and disk space backup, and I think, you know... Well, we dominated the direct conversation at NAB with Tom Joyce. Yeah, yeah. Absolutely. He was like, yeah. That was an eye-opener, that whole event. That was the first NAB I'd been to. I've never really been a big tape collector. Great front-end guys, but horrible back-end backup. Old school, meeting guys. Huge, huge, well, I mean... Still cheap. Yeah, and for all that data. Archive data, that's great. All right, we're getting the hook, Craig. All right. Love to carry on here. This is a great conversation. Thank you for your candor, and congratulations on great progress with 3PAR. We're watching with StoreOnce. Love the competition and the options. I didn't mention it. 100 terabytes an hour from the StoreOnce system. StoreOnce, 100 terabytes an hour. Congratulations, yeah. That's going to be debated. Take that to the bank. It's going to be good, and I think the idea of disrupting and controversy is a testament. I love this conversation of tier one. This architectural conversation is relevant, and the fact is that the customer experience is in preferred ways of doing things is changing, and I think big data is a testament that the fact it's unpredictable marketplace around the data, where to store it, when to get it back is interesting, so. Yeah, John, I'd really like to explore this whole tier one discussion. Maybe we can do that on Wikibon, maybe get the community involved, so. Well, it's very relevant data, so I think it's something we want to watch, and I'll see, you know, we'll watch the speeds and feeds too as well. There you go. Craig Nunez, VP marketing here at HP's Converge Infrastructure Group, Storage in particular, X3PAR as well, so congratulations on a great event. See you next time. Thank you. We'll be right back after this break for the next guest.