 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 Austin, everybody. We're here at Dell World 2014. This is day one. Stu Miniman and I will be here as well tomorrow. Ashley Gorak-Puwala is here. He's the vice president and chairman of Dell Server Solutions. Ashley, welcome to theCUBE. Good to have you. Thank you. Thank you guys. So you guys, are you exiting the X86 server business by any chance? We are not. No, you haven't announced that. We are not exiting the business. We saw that, you know, last time we were here, there was a lot of rumors about HP spinning off its PC business. Michael said, we're not doing that. You obviously, it was a tongue in cheek, but you're seeing all these trends. People sort of splitting up. You guys have referenced it a lot, really driven from the financial community. People wanting to maximize the sort of short-term profit. Not really thinking about customers, not really thinking about the long-term viability of the company, but thinking about short-term hits. Maybe not such a bad thing for some investors, but from your standpoint, what's life like as a private company? Well, it's been great. So we just celebrated our first-year anniversary of it this past week, and one of the benefits of it is that Michael likes to call it the shot clock. We're off the 13-week shot clock, and so when we make our decisions, it's no longer about week 13 and then into the forecast a week one. We don't have to worry about what the guidance was. We're really in a mode where we can do what's right for our customers, do what's right for our business, do what's right for our partners, and it's pretty freeing. And so, for example, we are able to take a much longer-term view of what an investment needs to pay off on. If we have a fiscal year, we still may look at ourselves on a fiscal year, but we can look at two or three years, and that's really how you get ahead of some of these trends is you need to be investing forward with partners and our own IP, and so we've really kind of left the tyranny of the 13 weeks behind. And some of the engineers in one of our labs were very, very concerned about some of the engineers in one of our labs put up a fun little poster about others are busy splitting, some are busy merging, but we're busy shipping, so we're really, you know, in the mode of excited about our 13G servers, just launched them last month, and we're shipping and bailability is going great. The ramp is awesome right now. It's right on track for what we predicted, which is the same kind of trend for our 12G. Uptake is good, so it's been very good. Yeah, so I want to come back to the products, but it's such an interesting topic because it's not like you guys don't have internal targets anymore. It's not like if you miss them, everything's okay, right? But I would imagine that maybe the highs aren't as effusive because you're not getting all this crazy, you know, stock market action and the lows aren't as, you know, damaging because you don't have all these negative headlines, but what happens when you miss or you exceed expectations now? I don't know what it's like to miss. So what happens when you exceed? You know what I'm saying? If you're part of a public company and you blow away the quarter, it's like, oh, everything's fine, we're indestructible. You know, sometimes you make bad decisions as a result, but so, okay, so you haven't been missing. You know what I'm saying? The highs, too. I want to tell you a struggle of this is that we would like to be able to say a little bit more about how well we're doing. And so we're actually on both ends. You know, if we do have a miss, we're not going to be talking about that, and we're going to be investing right past that. But if we have great momentum as we do now, right, we're not going to be, Michael has taken the tactic, we're not going to release financials except to the folks who own our debt. And so we won't be announcing those. And so we don't have as much of a direct indicator for the market at large versus one customer at a time or through the field. So even with the analyst community, it's a little, it's got to be a bit of a struggle because we're private. Other people going through quiet periods or changing their dynamics, and it's harder to get a read on the whole market right now. And we would actually prefer there was a read on the market right now because we know that the competition's suffering right now and we're actually growing at double digits. It's kind of a two-sided, but we'll always take this side of it. So Stu, we had to squint through a lot of IDC data this morning. Ashley, I'm wondering if you can help give us your viewpoint on the server market because those that don't look at the market might say, ah, it's all commoditized, it's all x86. There's no differentiation there. We're going to have Matt East foot on later and we'll talk about density optimized servers. The ODMs are taking over some of the really hyper-scale environments. You've got Cisco who's been attacking the performance side of the market, took a lot of market share from IBM who then got out of the business. You've now got Lenovo owning that piece of the business who's going to come strong after your business. You've got Supermicro underneath trying to undercut you on price. What do you guys see as the marketplace? It's a broad spectrum of solutions. So how do you look at the marketplace today and where does Dell's portfolio fit? So good question. First of all, it's big. Depending on how you measure it, we're shipping eight or nine million servers a year. So there's a big opportunity and there's also room for, as you said, very well different segments. So in hyper-scale, let's take that as a segment for a second. We've been very successful with our DCS business. We continue to be very successful, but maybe two, three years ago we started to blend those businesses together because there's so much synergy in where some of that is going in terms of scale-out, different file systems, learning about how to do a workload or an application that is a little bit more vertical, east-west traffic, things trends like this have really been interesting for us to be a part of, and then we're taking that down to, from hyper-scale to your scales, the tagline of moving to different levels of IT. But there's still things you can take advantage of beyond the procurement leverage of scale, which is a lot of the applications are going to go to smaller chunk size. When you need very dense, when you need more, you don't scrap that and start a whole new application stack that's vertical and think, I've got to go get servers, I got to get storage, I got to get networking. You can start to say, I've got to know that has all of that integrated and I can start to just stamp out more and build more and that's a great one. We talked a little bit from hyper-scale to your scale, we've got a converge system that we actually expect most people to buy one per place they want to put it and so that's our vertex platform. So it's converged but it's actually more of a remote office, back office, small business, applications like that. A lot of interesting usage there where people are taking it out to danger zones and a data center right in their arm. I wondered if we could talk a little bit more about the hyper-scale market. What's your take on things like the open compute initiative? Are there any success stories that you can talk about in large cloud or service provider environments where Dell is the standard, my understanding that if I look at the biggest guys, Google doesn't buy from you, Microsoft has been a client in the past but they buy from a few suppliers. What can you tell us about that market and open compute and where you see the biggest 10 to 20 data centers in the world? If you took our data center business which we don't anymore as a standalone we'd still be probably number four biggest server provider and we don't name and we're asked not to name customers obviously and so we'll maintain that confidentiality but we're having a lot of success there. Open compute. Had a big logo slide in there this morning. I was taking pictures. There was a big logo slide. So there's open compute. There was open compute conferences last week. We've been there. We were actually donating some systems management and working with others to donate. We supported Microsoft's window cloud server donation and so we have a pretty rich history of believing choice is pretty important. Openness is pretty important and we like to compete in the spaces around choice and openness for our customers and so we'll continue to do that. It's really, really important up in Hyperscale because you have a set of customers who absolutely know exactly what they want and so we want to be in a position to tailor exactly to what they want and that's that learning and that institutional tribal knowledge we have now is really paying off for customers who are not quite the scale of the customers you added or you referenced. Yeah, Michael talked about this morning the DCS, the new Exascale. It was the XA90. More of a storage solution but when I look at that architecture kind of going beyond, you know, it's not a rack. It's not a blade. It's more of kind of a cartridge or module type solution. Can you give us the update? I believe it was Copper and Zinc were projects that you guys had. Things like the HP Moonshot. Sometimes they might leverage X86 but not only X86 architectures. Can you give us that update? Yep, so whole range of things there. So what Michael referenced, the XA90 is a massive storage capability for you 90 drives. Really, it's probably a better term for it but I'll use the cheap and deep. And where folks are really looking at probably colder data, not cold, but colder data or data that's not right there at tier zero and high performance. And everyone knows that data has just been created at astronomical rates. This is a great product for folks who need to stand up large, large amounts very quickly of storage. And we have a lot of customers you can imagine tied to apps and things like that that are doing that very quickly. In the middle, what we think, we announced FX2 as well and that is a very modular, scalable product that we think is going to be great for scale out and for optimizing per workload. So that's really important to us is why have an average infrastructure that may be good at one workload, not good at another, or is just the average of all. And so some competitors, one you referenced, might take an approach of let me give you an architecture per workload and I'll try my best to hit it. But if you want to go to the next workload, which may be in the same data center, same company or you may even want to reprovision, you need a different architecture. We're taking the approach that we've got the architecture solved flexible enough that you can add or take away all the elements you need to build. Is it OLTP, transaction processing on one side? Is it static web page, front end on another? We can do that in the same infrastructure. And so that's more of the trend we see and I think it plays to the things we've been talking about in hyperscale and what we've been looking at is applications that scale and scale out east-west, direct-attached storage is very important to these applications and we need to be able to build just the right amount of networking and IO but what we can bring to it is put a wrapper around it, a single management layer that's designed from the beginning from a single company. So I wonder if we could talk about this concept of we were talking before you came on actually about server sand. I like to quote Gene Amdahl, the best IO is no IO. So you put the storage as close to the CPU as possible and if it's persistent and it's flash that sort of changes things. So how do you guys look at that notion of flash and how it's changing system and storage design? Yep, so we have a great set of products in the storage group and I don't know if you'll have someone on from there who's coming on. Great, Alan asks that question to Alan and he has a great set of flash products. If you look at where Alan and I are working together there's no question that compute and storage are coming together and that's the way of these future applications these workloads at scale. And so we've taken that tack for quite a while. So in our last generation to our non-Gen 2 we introduced PCI Express Flash that's plug-able. Semantics and feel of a hard drive that people are used to but it's got the new future of PCI Express and flash-based. Then Alan and I worked on technology that if you have a component on the back end you can bring the first tier or tier zero of the flash and put it on those drives. And then you can federate that over a whole cluster and scale it out. So think about now instead of a round trip to write back or get data through switches and through devices you're within basically the server compute complex. Second, if you look at our 73013G set of products we introduced the only 1.8 inch flash drives super dense. Now the capacity isn't there and we're working on that to make sure the capacity gets there in time but it'll always be a little bit behind 2.5 and 2.5 will be a little bit behind 3.5 in terms of capacity. But here these SAT SSDs create another tier so even if you aren't using our and you're using an application just on a standalone you can have 3.5 inch for your data this 1.8 flash for your tier and acceleration tier and then it's all still right next to the computer where you're not making the hops over the internet and out and going north and south and then coming back. We're all in on flash and we think we've got the best set of products by really engineering how we get as much flash into the box as possible. So talk about what that means for customers so they're reducing hops the latencies, they're not spinning devices necessarily unless you want it cheap and deep is it changing the way they're writing applications, are they not there yet are they just sort of using it to deploy less resources how are they getting value out of that? Well performance is easy or maybe straight forward is the better way of saying easy way of measuring it so and early on it was the adoption rate was really based on performance and cost and where that ratio met for a customer is when they started to adopt it. Now it's become prevalent enough that it's being used for tiers and so what you need that's your hottest data or the data that you need immediately or is always being in cash or is the most that you're writing for instance is what we keep in the flash and then we use technology such as deduplication or other methodologies to make sure we have the right data in the right spots and move it as it gets colder out to cheap and deep or rotating or even beyond out of the box. What people can do with this is now instead of outgrowing your mailboxes I'm always having to delete my email so I can send more a lot of the mail applications today can scale if you have enough storage with them and so now you can instead of setting up a whole new rack, procuring more storage configuring everything separately and adding data center room you just add another node and you've got either deeper mailboxes or more mail users that's kind of the virtual storage you've got vSAN or other methodologies storage spaces for Microsoft others that are starting to make sure that where we got to on virtualization for the compute side and the guest OS would start to get to for the storage side as well and that's really important so that we don't separate those two domains and end up managing this very separate from this because that'll be virtualization sprawl. Now another big mega trend obviously is the whole converged infrastructure so you're not only working with your storage but you're working with the networking guys as well talk about how that's changing the way in which customers deploy server technology. I'll talk about Vertex for a second so we have in Vertex standalone really meant for either your dentist's office all the way up to your mega stores meant for data centers of one compute comes in so you have your four blades to run everything you need firewall on one exchange on the other and some backups for that you have server side storage inside of it and then networking and what's really important there is that the usually the consumer is not interested in the networking at that point they're interested in plugging that into the wall and someone else is managing it from afar and so what you have to have is a standalone, let me fire up something different let me get a network technician involved it's usually a remote employee from a data center managing thousands of these that he's never seen and so it has to be seen with, it has to be part of the experience of managing the box not I'm managing the separate networking now but then you go all the way up to some of the scale out things we're doing today where why have eight times more cables than you need why not just take that away imagine a world where you didn't have to do that and someone didn't walk by and take one out and you had to find in that nest the one that was removed why not go down to something where you can aggregate your I.O. and get the benefits without the penalty Ashley, I wonder if you can tell us what you're hearing from customers about refresh cycles when it comes to the servers look at the biggest you know, data sets in the world they buy the cheapest hardware completely something that they use it, they get rid of it because you guys are turning the crank fast 13th generation server, every time you come out with a new one it's faster, it uses less power it has more feature functionality has the rest of the market shorten the refresh cycle or you still find in that they're going to buy it and try to squeeze every ounce of it as they can every year it's a continuum too because as you said the refresh cycle on someone who uses every ounce whatever they measure in there whether it's I.O. or storage or data or CPU the TCO is so low they can get by with refreshing very quickly and a lot of ways they see the IT is their factory and so it's worth put a dollar in, you get a dollar out or more for general IT, I'll give you a stat there's somewhere between 8 to 12 million servers out there, they're still running Windows 2003 servers come next year we're going to have a migration that's going to have to happen but we don't migrate like a consumer product we're going to migrate in a more controlled way where people can migrate their applications so we think this is for us the best time to have some of our customers moved over to this future IT stage instead of just buying the next version and putting Windows Server 2012 on it if they can put that same OS on the kind of architecture of the future that's a great place to refresh I'm wondering if you have any statistics as to how server utilization is for your customers when I first became an analyst I had a lot of arguments because VMware would come in and say we're going to get you 40, 50, 70% utilization and David Floyer the co-founder of CTO Wikibon it's like Stu, 15% is like really good and you know it's heavily, maybe you're getting 30% and if I'm one of the three best in the world maybe I'll get to 50% so where have you seen that discussion move? We still have a lot of folks who are running it single digit but no doubt consolidation for virtualization is the majority now well into the majority and so that is helping so we're seeing that come up all the time but we're still running mostly underutilized in a lot of cases there are some customers though that have figured out how to use utilization and virtualization hand in hand and so they're really running up to the speeds and feeds and when they scale they scale by adding they add more they get more that's where we're trying to get a lot of our customers but there's no doubt that we're still underutilized depending on there's still more choice though in the market do you buy less cores more cores and this can really have a big impact on what your utilization is right Ashley we have it leave it there thanks very much for coming on and sharing with us the visions of Dell server business and good luck going forward have a good rest of the show I appreciate it thanks guys alright keep it right there everybody we'll be back with our next guest right after this this is theCUBE we're live from Dell World