 Live from the Austin Convention Center in Austin, Texas, it's theCUBE at Dell World 2014. Here are your hosts, Dave Vellante and Stu Miniman. Welcome back to Dell World, everybody. This is theCUBE. Stu and I have been here all day. We'll be going all day tomorrow. Alan Atkinson is here. He's the Vice President and General Manager of Dell Storage. Alan, it's great to see you again. Thanks for coming on. Dave, well, it's great to see you. Still always great to see you. So, you guys, a bunch of announcements today. We're going to get into it. But I wanted to first, I'm asking everybody, life is a private company. We've got Michael coming on tomorrow. He's been very vocal about that. You've worked at public companies. You've had your own startups. What's it like to be private again? That's that argument. You know, it's a degree of freedom, right? So, it's the beginning of November. I don't have to sit here and give you earnings reports and data about how the quarter did. It gives us the freedom to make certain investments and long-term decisions to go do some strategic things that, frankly, just are really difficult when you're running a 90-day shot clock with Wall Street. And so, you know, from my perspective, it's been all positive, all offside. And I'll tell you what, the company is really energized. I don't know if you're getting that vibe when you're talking with folks at Dell, but the energy's there. All right. So, we talk about a number of things that you guys got going on. One, in particular, is the announcements that you guys made today around all-flash arrays. And what I wanted to ask you about is, you like some others, or unlike some others, chose not to go out and buy an all-flash array company. There are some other examples of that, but certainly, IBM bought TMS, EMC had to buy Extreme IO. You guys chose not to buy a company, and you're an acquisitive company. We are. Why is that, and what does that say about the zillion all-flash array companies out there? Well, you know, I really can't speak for what other companies felt they needed to go do in the market. But, you know, when we looked at our portfolio, we really felt like, you know, first of all, we have a fully virtualized infrastructure. So that's a huge advantage. We don't move bits and bytes around on physical disks. So, it wasn't that big of a stretch for us. Now, to be totally honest, we did have to rewrite all of our tiering software, right? So, it wasn't like we just said, hey, let's shove an SSD in there and see what happens. You know, we rewrote it all so that we could really take advantage of multiple tiers of flash. And, you know, Dave, if you look at what we actually do inside of the box, we're actually using that tiering algorithms that made us so famous back in, from the compelling days on forward. And obviously, we're leveraging a single-stack architecture now, which is exciting as we've brought the ecological and compelling families together. But that tiering algorithm is giving us an opportunity to use different types of flash that others can't use because we can control the rights. And so, effectively, what's happening is we're writing into a high-right endurance type of flash, and then we aggregate the rights and push them down into a much lower-cost, read-intensive type of flash. And this is totally transparent to the end user, but it gives us a huge cost advantage. And then beyond that, once we do that again, because we have the IP for all the tiering, we can then push down the spinning disk very, very easily, and we do. So you're tiering at the flash layer as well as the cheapest and deepest spinning disk layer, but you're writing to the flash first, correct? That's correct. And if you look at sort of under the hood, you can obviously buy it in an all-flash configuration or you can buy it in a hybrid configuration. It's essentially just a multi-tier architecture. So, if you were to buy a hybrid version, what you would end up with is two flavors of flash, plus a spinning flavor, probably a high-capacity disk. So, the reason I'm asking about the right to first, we've done some work sort of modeling out the impact at scale when you start to scale out on writing to flash first. And at smaller scales, it's not as big a deal, but the resources that are required to support an application when you scale out, if you have a right first flash first architecture, are significantly less, so the performance goes through the roof. What have you seen there? Yeah, I mean, our performance numbers are really staggering. That box that we announced today at a $25,000 entry price, I mean, that box will scale up to close to 200,000 IOPS, which is pretty remarkable at that price point. Right. You know, and obviously you can add more boxes and you can aggregate and get more IOPS. But, you know, if you look at it on a cost per IOPS basis, it's pretty attractive. Well, the interesting thing is, too, I've been around and talked to a lot of customers. I must have talked to five or 600 customers between, you know, April and June this year, just sort of roundtables, talking a lot about flash, and most of them were still early days for flash. Now, they put flash inside their existing array, sort of bolted on, if you will, but their big concern was still cost. Yep. It sounds like you're attacking that head-on. Yeah, absolutely right. We want to bring flash to the masses, if you will. I mean, if you look at it, I think we've actually gotten to a point where certainly a year ago, six months ago, maybe even today, flash is largely a performance-based conversation, but what I've discovered from talking to lots of customers is that's actually morphing. And, you know, I look at it as analogous to where we were, you know, maybe 10 or 12 years ago with flat-screen TVs, right? There was a vendor that was pretty dominant in the TV space that made, you know, CRT-based, very nice TVs, and then a couple Korean manufacturers came in with much lower-cost flat screens, and in the space of about 24 to 36 months, they had won the TV market. And, you know, people didn't really need a flat-screen TV, but they wanted it. And we're seeing the same thing with flash. People want flash. The only reason they don't buy it is cost. That's it. And if you can solve that problem, they'll buy your product. Even getting close, you would think, given the performance benefits, it's going to flip. Absolutely. But I think we can do better than close. Yeah. Now, why? Because you can do data reduction, you know, right in the platform? So, I think, you know, data reduction is a good thing, and it helps. But I think it's the ability to use multiple tiers and multiple types of flash where we can use very inexpensive flash that doesn't have great write endurance cycles because we don't write hardly ever there. We write it a different layer of flash. We aggregate the writes and push them down. And the performance difference between those layers is essentially no. So, the architecture is fully virtualized. So, you're gaining benefits there from an efficiency standpoint. You're claimed to fame, you guys created the whole tiering, you know, notion. So, you save a lot there as well. And you are doing data reduction inside the platform as well? So, today we're doing data reduction on all of our secondary data, right? Yeah, okay. And that's about, you know, your mileage may vary as is the case with data reduction technologies. And you're going to see us continue to iterate on that pretty rapidly as we go out. Okay, so that's another knob that you could turn, which is futures. What do you make of all the flash? I mean, it's great. I don't mean to disparage the flash startups and the VC money that's going on. It's wonderful because it's innovation. But I wonder if the flash vendors of today are going to be able to achieve escape velocity the same way that compelling and say three-par and even isolan were able to achieve escape velocity and get to be public companies. Some of these companies will go public, but the virtualization guys did it during the downturn. You know, and then some got in there like Phil Soren and David Scott before the music stopped. They raised up a ton of dough. And when the economy tanked, back to after the dot-com blow-up, the big guys sort of sat in their hands and really didn't innovate and were waiting around. And then they said, okay, we got to do virtualization. It seems like the large companies, I'll include yourself, have responded to flash a lot more quickly than they have to the virtualization trend. I wonder if you could comment on that. Well, I think there's been some lessons learned, right? So, and to be fair, there have been acquisitions in the flash space and some companies have reached escape velocity, if you will. Yeah, but not 2.5 billion or 2 billion or 1.8 billion. No, 1.2 billion. You know, I'm not smart enough to know what happens when, as you said, the music ends, but it seems to me that flash is just becoming ubiquitous. And if I look at kind of what's going on with price curves and density curves and where I think we'll be in a year with just flash technology, it just seems to me that it's going to be ubiquitous in all of the products. Now, some vendors are going to do it better than others. I think we do it particularly well. And I think, you know, our market gains this year would indicate that's true. But, you know, it's anybody's guess, but I will say it's a timed game and the time is drawing nine. I know Stu's dying to talk about your Nutanix deal. Yeah, so thanks, Alan. Glad Dave only took the first 20 minutes of the interview. But, you know, when I look at Dell, Dell is really in a unique position at this blurring of the line between the compute and the storage. When even at Dell World here, I walk and look around the show floor, you've got people that use Dell hardware, like SolidFire and Scale Computing. You've got, you know, various investments in OEMs, like the Nutanix deal we want to talk about. You've got investments in Accenta. Can you talk a little bit about, really, this kind of, you know, software-defined storage, Hyperconverge, whatever you want to call it, and how your team looks at kind of the, you know, build-by-partner, you know, options? Yeah, I could probably go for the next 20 minutes, at least on that too. So, you know, software-defined storage is one of those things that's on everybody's tip of the tongue, right? You know, you can't talk to a customer without, hey, what do you think about this company or that company or what's your software-defined storage strategy? And by the way, it's not very different in networking and it's not very different in infrastructure. Software-defined is kind of, you know, one of the hottest topics you got to have a position on. You know, we announced our project Blue Thunder effort over the summer, which was really our umbrella label for what we were doing with software-defined. And if you think about it, you know, we have the perfect platform for software-defined, as we like to say, your path, our platform, which is PowerEdge. And we talked about a couple variants of that today here at Dell World with the FX2 coming out and some of the other platforms. In fact, you just mentioned it. A lot of the biggest SDS providers out there actually are on Dell, which is kind of an interesting statement. We're very proud of that. We have a very, very healthy OEM business that, you know, we're very proud of and we design with that type of thing in mind. But, you know, when you really think about, okay, what is the value of software-defined? Some people say cost. I mean, I've got to tell you, I'm not entirely convinced of that. But one of the things that it is is it's very workload-specific. It's almost impossible to have a good software-defined conversation without also having a workload conversation. And as such, you're going to run the right stack for the right workload. So a good example of that would be Hadoop, right? You're probably not going to run Hadoop as a general-purpose file system. You're going to run HDFS because you've got a big data analytics problem and you're going to run MapReduce, right? So that's one example. There's others. You mentioned Accento and sort of what they can do with their ZFS-based technology and some of their other newer technologies. Nutanix being a great example of scale-out. So our view is that we've got to have a healthy partner ecosystem. Of course, that includes some of our premier partners such as VMware and Microsoft as well, right? And I think that doesn't mean everything to everybody, but it does mean partnering with the best-of-breed folks so that we can offer an NN PowerEdge-based solution that includes support, integration, and all those things. In the case of the Nutanix partnership, this one is a very deep OEM partnership. It's one of the ones that's fully Dell-branded and we don't have a ton of those. The Dell XC series is in every way a Dell product powered by the Nutanix web-scale software. And it's a very tight partnership, and we're super excited about it. You know, I think the Nutanix web-scale software is one of the biggest plays I've seen in enterprise infrastructure in many, many years. I mean, they are very hot and for good reason. They really do solve the problem. And of course, we're also going to be developing our own IP in the software-defined space. So you can expect us to have a hybrid, if you will, of, you know, tight partnerships with premier vendors as well as Dell IP-based solutions. And, you know, it's all of the above because, as I said, it is workload specific. Yeah, Alan, just a comment. I think you're right. It's applications and where it's at. And the problem we've had with infrastructure for as long as I've been in IT is we started with an application, then we had to make some bespoke hardware and I had to have my IT staff spend, you know, tweaking dials and adjusting configurations to fit that application as opposed to we want infrastructure that's built to support that application without having to do all of that extra work. Alan, from our understanding, you were, you know, instrumental in that OEM relationship with Nutanix and from what I've heard, it was really fast from making the decision all the work that's been happening over the last couple of months. I hear the engineering teams between Dell and Nutanix had a big party last night. Can you give us a little bit insight because out of all the solutions that were out there, Nutanix was one of the only ones that didn't use Dell hardware. So how did that one come to fruition and how did that partnership happen? So you said a bunch of things there. Let me see if I can knock down a couple of those, right? So first of all, you know, I'd like to think that we do things pretty rapidly at Dell these days. If you look at how many product introductions we've gotten out in the last 12 months and the pace at which we've moved, I think it's safe to say, you know, we're pretty motivated and, you know, as a private company we're moving pretty fast. So that's the first one. The second one is, you know, we went out and we talked to lots of customers and we talked to lots of the Silicon Valley companies and honestly, Nutanix just had a unique level of buzz around them. I mean, their pedigree from the early Silicon Valley to scale out data centers, their buzz in the industry, their customer penetration, frankly made them the type of company we wanted to partner with. And then I'll also tell you that the teams just had a very good synergy and, you know, I'll let the Nutanix folks speak for their side of it, but the Dell team and the Nutanix team really developed a really good symbiotic relationship and, you know, we're excited about, you know, taking this forward. So you did that deal, you and others, but you initiated it, right? It was sort of something that you helped catalyze. Let's put it that way. I think that's a fair way. Okay, so now where does it sit inside the Dell organization? So it is in the storage group, right? So the relationships own an enterprise and specifically in the storage group, but also tell you increasingly, you know, the lines get so blurry between compute, networking and storage, you know, even internally with our products and the way we think about things and the way we develop things that, I wouldn't read too much into that, but it does live in the storage group from a product management point of view where we bring out platforms and those sorts of things. And it's sold through the Dell channels, correct? It's sold through the Dell channels and direct both. And direct, but Nutanix does not sell the solution or does it? Nutanix does sell their solution, right? Yeah, right. So they're going to keep a Nutanix branded product out there, but the Dell product that includes Dell support, includes PowerEdge, includes the things that we think, you know, most companies find attractive. You know, is a Dell product that can be bought, anyway you buy Dell products, which means channel or direct. Now, I want to also ask you about sort of the, you mentioned ecologic before and compelling, those two worlds have come together. Can you talk about that? Where are you, I know you're at the single stack now, but how did that come about? What does that look like now? Yeah, so, you know, one of the things that we really wanted to do in order to get more leverage, more wood buying in the arrow, if you would, was to get everyone in Dell Storage pulling in the same direction. So we wanted to finish integrating the acquisitions. I mean, Dell Storage is really a collection of five acquisitions. So, compelling ecologic, perhaps the best known ones, but also RNA on the PCIE attached flash, which is a really innovative, cool product. That's the Fluid Cash for Sand product now. Exanet, which is our Fluid FS. And Ocarino, which is our data reduction technology. And so the mission was really, let's bring them all together in a way where everybody's protected. If you're an ecologic customer, you can use the combined stack. If you're a compelling customer, you can use the combined stack. So we're not all the way there yet, but we are beginning to bring out the 4020 being the most recent example. New products that are based on a single architecture, the SC series in this case. And we have announced very clearly that we're going to be providing common management between both existing ecologic component and the new products, as well as replication between all the products and common management. So everybody's investment is fully protected and it should be nothing but upside for customers. So essentially, that's where your R&D really is, is the net integration, right? That's where you put a lot of your emphasis. Well, integration and acceleration. We've added a lot of new features, things like data reduction, things like improvements in our failover technology, things like huge performance increases, things like rewriting for flash, right? So that we can support multiple tiers of flash. And I could go on and on. I mean, our feature list is very rich, but you can do more features when you have more engineers working on the same thing, right? Right, right. Okay, but it's fair to say, Alan, that a big chunk of that R&D was on integration. As you say, you added some other pieces. And the integration was also a lot of leverage in the model there. So the reason I'm asking that is, as you've, I mean, early on, you had to bring in big assets. You had to swallow ecologic and that was not trivial. Will the R&D then naturally shift toward other innovations? What can we expect there? Well, never say never, right? But I think for as far as I can see, we're going to keep the same or more, in fact, we just opened up a new design center last year on the existing technology. Now my caveat there is, as things like software defined become, you know, even more ubiquitous, we definitely want to have our own IP there. And there's multiple ways to get there, right? So I'm certainly not going to say that we may not redirect some efforts to the software defined side, but kind of the way we're seeing the Dell IP version of software defined is we want it to be fully integrated with our mainstream stack. So we think it makes sense to have the same folks working on that. You know, I think the value for customers is really if you can buy a product, a Dell storage product that comes in small, medium, large, or software only. That's got an awful lot of value. Especially if you happen to have a really good market sharing service. Well, right. So, okay. So that software only piece. Can you talk a little bit more about that? Can we expect you to sort of extract the function out of the hardware and allow that to reside? Well, I'm not announcing anything today, Dave, but you know, we're certainly thinking about, you know, hey, what would a software encapsulation of our technology enable? Now, again, we're not announcing anything. That's not something I have today. First, you know, priority is let's get this fully integrated family now. But, you know, as we think about the way things go forward, software to find is clearly a force to be reckoned with. And it's clearly an area where I think the Nutanix and XC series is a pretty good existence proof. This is an area where we're going to apply. Well, speaking more generically about the industry, I mean, we've seen for years that you take a disk drive that you could get from whomever for a buck and we in the enterprise to sell it for ten. Now, the reason we're able to sell it for ten is because there's all that function in the controller, which is software. The industry seems to be wanting to to extract that and then, of course, charge for that. That's fine. It doesn't really... Does it change the economics you said before? Not really about the economics, necessarily. I kind of agree with you. I think that's mostly hyped, Dave. And I agree with that because you got R&D and you got to pay for that somehow. You got innovation has to get paid for. I agree it's mostly hyped. Now, I mean, at a macro level, I think there's cost pressure on the entire industry to, you know, get more competitive. And as densities improve and technology prices come down, that's happening anyway. For as long as the storage business has been out there. Yeah, so in the course of cloud, maybe there's this other little wild card. So you're going to continue to see an economic curve. It's going to provide more IOPS and more capacity for the dollar. No question about that. But is SDS going to be a driver to really move that? I think it's going to be a driver to get much closer to the application and much more workload driven. That's what I think it's going to bring. Yeah, really. That workload driven piece and automating that the knowledge to be able to apply resources to given workloads and orchestrating that is, that's new. Now, what that does, I guess it is economic from this standpoint is it attacks the biggest problem in IT, which is the labor cost. Yeah. Right. Now, that doesn't necessarily mean, you know, the industry revenues are going to drop. It won't. What it does mean is that IT organizations will be able to spend money elsewhere. Do you buy that? Yeah. You know, I think so, but again, the data center is morphing so rapidly and flash technologies are making things so much more dense that I think people are going to use this in ways that are pretty imaginative. I mean, I certainly, you know, given our server share, I'm very excited to make sure that we're a leader in this space and I think things will get redeployed and maybe labor costs come down because you're going to have less physical footprint, data center maintenance is less and some of those variables change around but it's actually pretty complicated when I start managing, I don't know, five, six, seven workload specific stacks, you know, one for my big data problems and I've got to put another one out there for my cheap and deep and I put another one out there for object storage and I put, I mean that is what we're talking about, right? Yeah. So, okay, so let me, let me, because you're a former practitioner, so this is, you got street cred here, so that's interesting what you said. We might be just sort of shuffling the labor deck but maybe not so much on what some people call non-differentiated heavy lifting of hardware and configuration. Well, Dave, let me tell you the way we're thinking about it and, you know, again, this is, you know, I keep repeating that this is a very emerging area so some of this stuff is maturing right before our eyes, right? If you look at what we talked about when we announced Blue Thunder, it was you know, if you start with a common platform, right, and that platform is PowerEdge you're already building in a common management infrastructure at the hardware level. Okay, so that's interesting to begin with because now I can buy the same hardware for all those storage stacks. Okay, that's new, right? Second thing is well, if I'm starting to build in sort of state-of-the-art orchestration technologies I'm putting RESTful APIs and all these I'm working with vendors and most of the open source platforms already have this some of the proprietary ones do and some of them don't but we start making them in a way that they can be instrumented in a common fashion that we can use tools like ASM or other things to leverage orchestration across them. You can make the problem better. In a perfect world you would have common infrastructure and you could mix and match those storage stacks as your workloads dictated, right? That would be perfect. We're certainly not there today. Right, and you could do that based on policy or based on whatever analytics you can pull from the system. So you could take the can you take the humans in the future? Will you be able to take the humans out of that provisioning equation? I think provisioning will get far, far more automated. I actually have little doubt about that. If you look at even our XC series that we just announced, the Nutanix relationship it's so easy to provision there I mean you can do it in minutes, right? It's extremely easy. And so the ratio of person to box is very, very attractive. And I think you'll see those type of things happen in a wider area, but I don't think there's storage specific to it. I think you've got to instrument that together with the workload and with the hardware. Yeah, so I want to tap on your practitioner experience from your days in the financial services industry. When you think about cloud and forget about the sole security and governance and on-premises or off-premises just forget about that for a second. And I know that stuff exists and it's important. From an economic standpoint, do you feel like your customers will be able to replicate the cost structure of the big cloud guys? Well, okay, so somewhat of a loaded question. Who's cost structure? I don't think that the private guys are going to be able to get stuff as cheap as the public guys just because of scale, right? But I do think that ultimately private cloud has economic advantages to off-prem. Okay. I think you'll probably see some combination of hybrid. I believe that. But when you look at at least all the cost studies that I've seen and I don't claim to be an exhaustive expert on this. Unless you're at a very small scale, when you factor in management costs, networking costs of moving the data back and forth and all the rest, it still ultimately ends up cheaper to do it as a private cloud or at least a hybrid, not entirely private. Renting is more expensive than owning at some scale. At some scale, right? Yeah, and of course it's hard to figure out because there's all these artificial, I mean even with Amazon, it's hard to figure out what pricing looks like. You saw when Google, you know, took a shot at Amazon and said, hey, they announced price cuts every, you know, three times or four breakfast, but in reality hardware prices have come down a lot faster and you guys take advantage of that hardware price. You guys cut hardware prices all the time. Yeah, I mean, look, the industry's price curves only go down for price performance capacity, right? It's amazing industry we live in. It's been that case forever, right? So, and that's a great thing for consumers. I mean, look at the amount of power you have on your mobile device these days. Right, okay, so probably not being able to replicate the cost structure of the public cloud, but when you look at a certain scale, the overall costs, and you factor back in, I remember my proviso security and governance, all this stuff, you get close enough that the business case for on-premises cloud is going to be really compelling. You know, I think so, right? You know, barring some other major movement, of course, you've got to factor people into that equation. Somebody's got to administer that cloud, and so I do think there's an issue of scale. If you're not at a certain scale, I think public cloud probably wins. Right, and well, and again, workload dependent. So, all right, Alan, we've got to leave it there, but I'll give you the last word. Thinking about your storage business, I wonder if you could, you can't give us gross margin by product, but if you could just talk generally as to how the business is doing, and then what are the takeaways for Dell World 2014? Great questions, Faith, thank you. So, I mean, look, our business is, I am really excited about how well we've done. I think you heard in Michael's keynote, you know, number one worldwide in capacity now. I mean, so we've passed everybody, right? And a lot of that is the movement towards software to find, and people moving data internally to servers, and big dense servers, and watching that shift. But, you know, we're gaining share across every measure you want to manage. You want to mention, right? You want to look at external, you want to look at price bands, we're gaining share. And that's exciting, right? I think the strategy we have in place is resonating, disrupting the usage that people get their heads around things like the flash announcement of 25K that's not hard to understand. So, very excited about that. And I think that excitement if you're walking around here at Dell World to answer your second question, you know, this is my third Dell World. And I can tell you the excitement, the momentum, the morale at each one has just built and built. I mean, last year it was right after they go private, people were pretty happy about that. That was a tough battle, right? Got that done. And this year, a year in, I don't know how many folks you've talked to here, but people are pumped. People are pumped, no doubt. I was here two years ago and the mood here is, it was good then, but it's even greater here. So, congratulations down. I love talking to you because like I said, you got practitioner chops, you're seasoned exec, and I think they put the right guy in the job. So congratulations on the success and I know you're not done. We'll be watching. Thanks, Dave. Thanks very much. All right, keep it right there, everybody. Stu and I will be back to wrap. We're here live at Dell World. This is the Cube. Right back.