 Live from Las Vegas, it's theCUBE. Covering Dell Technologies World 2019. Brought to you by Dell Technologies and its ecosystem partners. Welcome back to Las Vegas, here at the Sands Convention Center at Dell Technologies World 2019. I'm Stu Miniman. My co-host here is Dave Vellante. Two sets, five hosts, three days, wall-to-wall coverage, all of the action for Dell Technologies, all the component pieces. Happy to welcome back to the program, Ashley Horak-Perwala, who's the president of the server and infrastructure services at Dell EMC. Ashley, thanks so much for joining us. Thanks for having me. Good to see you. All right, so we actually had Sam Grokot on, and we were talking about all the product lines, and he said he's the father of power going across the line. He did admit that the power line goes back to PowerEdge, which of course is your baby. That's right. Give us the update, lots of discussion at the keynote, always change in your world, so give us the latest and greatest. Sure, we're about 25 years old now, so PowerEdge has lived on for quite a while. We've got to be over 30 million servers out there by now, so we've had a really good Dell Technology World so far, more to come, but some of the, maybe some of the list real quick of announcements that we've had, and we can talk a little bit more about them in servers. We actually went a little bit early from Dell Technology World and lined up with Intel to launch Cascade Lake, bringing Optane into a server class memory. I think the industry's been waiting for it. We're ready to deliver now, and so that was earlier this month. We've put quite a bit of advancements and enhancements in our OpenManage Enterprise and in securing the platforms. We also, this week, talked about a PowerEdge, it's not called a PowerEdge, so we call it the DSS8440, and really a capstone product to our AI ML portfolio. So today we already support one, two, three, four accelerators per server. Now we can go up to 10. We can support the latest NVIDIA B100, TensorFlow, GPUs, and it's really a unique system within the industry that's going to help customers scale their training loads further and further, faster performance, more MIPS, a very, very intense box, but one that's going to be, I think, well-received within the marketplace. You said MIPS? I said MIPS. I love that term. So, Ashley, we got a lot of pieces that your solutions fit, but you mentioned one item that I wonder if you could just explain to our audience the important of. SCM is something that, how does that impact solutions, the applications, it's something that a lot of times get lost in the whole general storage discussion, so maybe explain the importance of SCM in the marketplace. So, it's a game changer, it really will be, but it'll have to go in our mind through the technology adoption curve that a game changer deserves. So it's been a long time coming, we've been working on it, the industry's been working on Intel, hasn't been working on it for more than a decade. And if you think through it, we see customers using it in two different ways. In memory mode, expanding the capacity within nodes to levels that you can't reach with DRAM today at almost DRAM-like levels of performance is something that a lot of customers already have models for, they can think through TCO, they can think through their performance characteristics, and it really becomes something they can consider to enhance their portfolio today. At mode, a little bit different. As we think through software from the OS level, kernel, hypervisor, application, cache, log, database, all these levels, we're going to have software that has to catch up and allow this to be the game changer it is. But already, I'll tell you the demand for systems that we're providing customers to begin their evaluations, their proof of concepts, their software development, has actually doubled what we thought it would be, and we were pretty ambitious. So I think the demand is there, and we're going to see that adoption curve when the software catches up. And the specific use cases you're seeing early on? Well, like I said, memory mode, I think people can get their heads around already is, are they performance, are they capacity bound? By DRAM, start to do the economics doesn't make sense. At mode, caching for sure, putting log, changing kind of the structure of how you do logs, and database is really going to be the killer app when we get there. Across the different vendors already, we've seen pretty significant increases in performance and we're early still. But I think there's a few things that our customers want to get through and we're trying to help them with. So if you have persistence in the system, you have a new level of something you have to secure. And so we're spending a lot of time with our customers, helping them to develop technology, methodologies to say, wait a minute, information. I turn the machine off and there's still information besides the hard drive or the SSD. Also, can I trust the data, even though it's persistent? Or do I have to have stored services at that level that help me with things like replication or snapshot or archive? So we've got a long way to go, but we're really, we believe this is a game changer and we're developing towards that. And cost wise, you're saying slightly more expensive than DRAM. But probably a little bit more than slightly. Yeah, okay. More expensive than DRAM and relative to Flash, obviously more expensive than Flash, but much higher performance, right? Much higher performance. And so it's just a modeling exercise, but it'll reach levels we haven't had before. And then from a software developer point of view as you go forward, you can really think about scale out systems differently. If your application was bound by capacity of DRAM or memory, this changes it quite a bit. So you're talking about a new programming model, essentially, right? That's why it's going to take some time. But you would expect maybe uptake in financial services early on. Is that fair or not necessarily a health care? Cross all verticals. I think it's going to be where enhancement or performance can, if you pay three, four, or five X the cost, but you get three, four, or five X the capability, or even less, you have to think about it. But there's some applications where latency, where performance of the database are so sensitive and such the bottleneck today that it's well worth it. When you look at the innovation pie that's going on in servers, how much is like architecture, hardware, architecture versus sort of software and management? Can you sort of, I know it's a sort of general question, but give us a sense. Sure, I think as interesting as we are investing as we go forward, I think into a brand new era. So I mentioned earlier, we made it to 25 years old. What's going to happen over the next 25 years? So I think most of the architectures that we developed today are highly, highly optimized for bringing data into a processor, calculating, and we have very balanced, efficient, high-performance systems for that today. What are we doing going forward? Well, we're not necessarily bringing the data, describing the rules called software, and then getting the answers anymore, right? Now what we want to do in a lot of situations, we want to bring the data, which is the most valuable asset. We actually kind of know the answers already. We want it to calculate rules for us, and that's the output. That's a different architecture. That's a different way of computing, and that's why you're seeing these heterogeneous architectures starting to form accelerators, a lot of technology going, and innovation, and venture capital, and talent going towards really building that new model going forward for the next two decades. Okay, Ashley, we heard a lot about cloud this week. When I looked at many of the solutions underneath, I kept hearing the same answer. VxRail, VxRail, VxRail. I've talked to some of the team. There is more than just VxRail and some of these solutions. I had Sam on, looked at some of the other pieces, but VxRail has been a rocket ship for the last couple of years, and of course, the servers underneath, driving a lot of that. Can you talk about how that plays into your portfolio, and some of the architectural discussions we were seeing? How does that bleed into the HCI and hybrid cloud discussion? So, if you think of the journey we're on, 10 years ago, perhaps, maybe even more recently than that, customers really were making two different choices. As a matter of fact, you guys know as well, I was organized into two different organizations, one to deal with hyperscale, and one to deal with enterprise capability. And customers can see that. They want to be able to operate in both domains, but even we were organized differently. If you go maybe five years ago when people started talking about software defined in HCI, we finally had a mechanism to say, you can build scale-out architectures. We can automate this capability for you. You don't have to actually spend all your OPEX, your administration, your talent, your time, just keeping the infrastructure up and running. And so people broke out of IT by project, by Gantt chart, and into flexible architectures. Next thing they said is, but we still aren't really operating. We're operating in silos of very flexible architecture here in my data center, very flexible architecture in the Colo, very flexible architecture in software defined, or SAS or cloud. How do I bring it together? So we believe there's a consistency of platform and infrastructure that allows us to move to a consistency of operations. VxRail offers that today because we uniquely can integrate with VMware and VCloud Foundation to build where now we can take care of the automation, the life cycle management of the hardware, VMware together integrated now can take care of the life cycle of the software stack, all the way up to the IaaS layer or beyond. And now we have the ability to say, you can look upwards, you can develop, you can build on that. And even more so, if you want to then stitch that together and have that be the control plane, you can now build that out to other native public clouds. Now you have a hybrid cloud. We can actually get there. We can actually organize around it, build it. I mean, it's a breakthrough for our customers. And then add on that, some customers have come back to us and said, you have the expertise to do all this for us. Can you, can I just consume it? I don't actually need to control it. And in that case, we can offer it as a service. And we previewed that as project dimension last year and now the teams are really happy to bring it to fruition all the way to beta with customers today. And really give customers kind of that choice. So what's behind that? I mean, you got a team of people sort of monitoring everything. You got obviously a lot of automation. What's the customer conversation like? I mean, it's early days, but what do they want to know about? Do they just want to say, hey, you take care of it? Or do they want to peel the layers and say, okay, I want to peek behind the curtain before I sign up for this? So on the platform side, customers want to know how does the integration work? Really, where do I have to spend time, energy? Can I really live at this IS layer? Can I live at the PAS layer with Pivotal? Can I live above that? How do my workflows get managed? And when you say, we're kind of in the environment and the methodologies you already use today with vCenter and vMotion and things like MPKS, then I think you see a light bulb go off of, okay, I can really leave the administration to the machines and the automation. Then the customer who's interested in moving everything maybe to a consumption model, then they have the next question, which is, can I have consistency, not only of infrastructure and operation, but of consumption? And that's where an as-a-service offering really starts to highlight the fact that we can meet you on your journey wherever you are. Some customers aren't ready for that. Some are just right there saying, that's really the model I want to move to for digital transformation. Okay, you got roughly a $20 billion business growing at now almost 20% a year. So pretty good year last year. Give us the update on your business. Why are you being so successful? And I got a follow-up question on component, sort of supply, but. Okay, sure. So we did have a pretty good year last year. We don't break out servers, but servers and networking, as you said, about $20 billion growing at 28%. Why? Well, I think we have one of the most capable portfolios of infrastructure. We're uniquely trying to make sure that we are operating within the Dell Technologies portfolio. And so most customers, Dave, have not come to us and said, you know what I'd like to do? I'd like to have like 10 more of you guys come meet with me and talk to me about a portion of my business. They said, why can't you come and provide all of my needs? But I don't want to compromise. I don't want to have one best of class and then have to compromise across my other needs. So really building kind of number one, all in one place, is that promise that you don't have to compromise. Really, it's changed the dynamic with a lot of customers being able to say, this is my essential IT infrastructure provider. They have what I need. So that's helped quite a bit. The nature of our business, I think, is that we are operating from the smallest customer. You need one all the way up to customers. You need a million servers. And we're able to operate in a consistent power edge tenant across all of that space. Then the, I think you didn't mention it, but in hyper-converged, we're seeing growth rates that kind of put the server business to shame with, you know, we were 65% in Q4. In an industry that's growing 40% of that's on fire. It's a new business model. It's still emerging, but customers, the demand for hyper-converges continues to go forward because that operating model, simplicity, elastic, scale out, automated, is extremely powerful. A component supply right now, a component pricing is a tailwind for you. For years it's been a headwind. Is that right? It's flipped? Certainly, yeah, certainly the last two years has been sort of an unprecedented rise in some of our commodities in terms of cost. We're seeing that be deflationary or stable at this point. So it's really changed a little bit of the dynamic of how customers were operating within their own budgets. So now I think we're more in what we're used to in the beginning 23 years as we go forward. Actually, last thing you talked about, you used to have kind of the hyperscale business. Just give us the update. I saw a quote out there that, you know, Dell puts more gear out there in hyperscale environments than anyone can do. Just give us a little context as to what that means. Sure, sure. You know, as we go forward, I think we've seen others say that they don't operate in certain businesses. They don't want to be in tier one. And you won't hear that from us. I think where we can add value and we have incredible assets in terms of engineering, modular data center capability, capability at the edge, real assets of software, supply chain, delivery across the board. We want to be able to help customers build their infrastructures. And in the service provider community, I think we've already built up relationships, credibility and technology to help them compete. Our standard is if you do business with us, we want you to win in your segment. We want you to transform faster than your competition. And we think we can do that for people. I think we continue to see quite a bit of success in the service provider space. All right, well, really appreciate the updates and congratulations on all the progress we've made out. Appreciate it. Thanks for having me, guys. All right, for Dave Vellante, I'm Stu Miniman. Getting towards the end of day two, three days wall-to-wall coverage. Thank you, as always, for watching the queue.