 Live from San Francisco, California, it's the Cube at VMworld 2014 brought to you by VMware, Cisco, EMC, HP, and Nutanix. Now here are your hosts, John Furrier and Dave Vellante. Okay, welcome back everyone, day two of VMworld 2014, this is theCUBE where we extract a signal from the noise, I'm John Furrier with my co-host Dave Vellante, 50 year doing VMworld, we're excited theCUBE is extracting that signal, sharing that with you, and of course we've got a great lineup, wall-to-wall coverage, three days of back-to-back-to-back interviews with senior executives, entrepreneurs, leaders, venture capitalists, we're here getting all the stories for you and sharing that with you, and I'm pleased to have Sam Greenblatt who's CTO and VP of, VP and CTO of Engineer Services at Dell, senior executive in the industry. Sam, great to see you again. It's great to see you guys. Thank you for coming back. You know, you and I were talking, we've been around the block, you've certainly been in the trenches for many, many decades, you've seen cycles of growth, you've seen up and down cycles, you're now at Dell, we are kind of back to infrastructure again in kind of a new way, it's the same game, same architecture with a twist, what's your take of the current situation, share with the folks out there, what's going on in today's landscape, hyper-converged, hybrid cloud, and user computing, it sounds a lot like the 80s, but with a twist. Well, one of the things that's very interesting, if you look at, and I hate to use Ray Kurzweil curve, and you look at it with the factor, yes, computing is going up by a factor of 10 to 6 over a period of time, but the problem is there's not a lot of new concepts. If you look at the mainframe when we, and you look at everything else we built, that was hyper-converged, you know, you had DASD, you had the CPU, you had networking, may not have been in the same box, but it was very closely coupled. When you look at where we are today, the word converged has now been moved by our friends at Gartner to integrated systems, because you can't really hyper-converge everything, but you can get in a box such as we did here with Evo Rails, a 2U form factor with four nodes in it, and actually get 100 VMs on it. So is that much different than the 80s when we had those huge beast? No, it's scaled hard work. So if we go down history, we'll look at down history, memory lane in history, and look at what's going on today, some are saying with cloud servers, does the server go away? I mean, servers are not going to go away. They shift, you guys are in the server business at Dell, and also want to get your perspective on where the servers end up being, and two, what ends up happening to the architectural wars, and as an example, Token Ring versus Ethernet were topology wars, both networking. The stack is kind of an architectural, you know, wrestle match right now with vendors, certainly with OpenStack and Linux, we've got virtual machines, so talk about those two factors, servers, are they going away where they live, they shifting, and then certainly this architectural wrestling match. Servers will never go away. There will always be a CPU. I don't care if it's a smaller form factor or anything else. There is going to be a chip somewhere that is going to serve all the work, whether it's a phone, whether it's a PC, whether it's a large server. What's happening now is we're recognizing as an industry that disaggregation of the server. With the Internet of Things and everything else going, we've got to move the server closer to the point of information, and as you do that, you get into a whole new style, and I'm not saying that Dave got, we were talking about another company that uses, you get a new functional view of the architecture where everything is now interconnected because the ubiquitous network, high speed, but let's go back to the 80s where you started. You and I both had a modem, 9600, IBM made a fortune selling type two cabling at that time because nobody thought you could get higher than 9600 without shielding and all that. Today in copper, on Ethernet, we're up at 200 gigabits. By the way, one of the things that's very popular from your networking companies is they're running networks through electrical grids in your house that are interconnected. The world changes, but it doesn't change the functionality. OpenStack sitting on top of VMware, or VMware sitting on top of OpenStack is a big discussion point that's going on last night at the parties, besides the beer pong and the lip cup action that we were involved in. Of course, I won the beer pong match, but get that in, Dave. He was deadly. My son taught me everything. He's a lot of practice. We didn't have beer pong when I was in college. But what's the architectural match? Am I putting OpenStack on top of VMware, or am I putting VMware on top of OpenStack, or is it the choice? What's your take on that? Well, if you really look at the VMware architecture, you can run Linux in a virtual machine. If you really look at OpenStack, it's a Linux. It's basically an infrastructure service sitting on Linux. So one of the things that the OpenStack community has done is been able to run ESX on top of OpenStack because it's sitting on a Linux. The way that they're doing it here is they're moving it into a virtual machine running Linux. So you basically get OpenStack on top of Linux, which is the right way to go. It depends on your point of view. OpenStack on top of Linux in a virtual machine. Isn't OpenStack Linux already? So why the virtual machine? The virtual machine is so you can run other operating systems in tandem with OpenStack. That's VMware's take. If you were to take a company like Red Hat, they believe ESX should run right on top of there. And it's an argument that you really have to think through when you go through what is your style of compute? Is it open source Linux or is it a VMware style of compute? Okay. So those sort of two counterpoised styles. And then you talked before about the processor getting closer to the data. Are you talking about function shipping there? Like a Hadoop style of processor here? Here's the thing. We have done a lot of work at Dell in putting a processor right on top of a disk. Right. And the reason for that is if you're going to do a Hadoop big data thing, you want the processor not to have the latency of going back to a processor of your network. You want it right where the data is. You've seen our server saying stuff, right? Yes. I mean, bringing this, so I'd say it differently, bringing the storage back to the CPU maybe closer. The best I always know I owe. Maybe that's how I'd say it. You just quoted Gina on the course. I'm an old guy. Gina, I'm doing used to say data rush should always stay at rest. Don't move it unless you have to. So let me get this right. The big trend. I just tweeted this in the crowd chat, crowd chat.net slash VM world. But the, if I get this right, moving, moving closer to the data to the processor in the cloud, that's a mainframe, right? I mean, that's what mainframes that's back to the mainframe architecture. That's how we started this conversation. So, okay, we Dave. Okay, so I want to jump in there. So VVols. Yes. Right. Look at how long it took IBM to get the equivalent of VVols. And it was a long slog. So I, I would suspect we're going to have a similar long slog here. Do you agree? It's everything is a lot faster than it was when IBM was trying to do VVols. Uh, I think you're going to see it in the next 18 months horizon. Okay. So that's pretty good. You know, the thing is that sped up, we're not using spinning media that much anymore. We're using SSDs. We're going, matter of fact, Dell is the leader just to mention it in flash technology to reason, we have flash at the price of disk. So you're saying leader in terms of cost? Yes. Not, we don't make the SSDs. So if, yeah, right, but if flash is the same price as disk, why would you buy the spinning disk? Exactly. Ever. When I look at my PC, I don't have a solid state. No solids, no, no spinning disk here. Right. And so what's the optimization for flash? I mean, the flash vendors are being pigeonholed right now is biting off more than they can chew because they're positioning themselves as transactional. So for, you know, high end for financial service throughput and then cost per gigabyte, can you do all three with flash or is there an optimized use case for one or the other? And or they by boiling over the ocean. Now, see, you know, we've talked with the solid fire you guys last night and they claim that they they're going to bite off all all three of those categories and you have pure and you have everybody else here. The interesting with flash is we're just learning what the different tiers are and what the optimization is. We have a tier zero flash. We call fluid cash for sand. It has certain use cases. But what's happening is we're finding use cases that we never expected. Can it do all three? I'm not sure. I know we have VC segment coming on. We have Jerry Chen coming on. He's on scene of any other technologist. But I got to ask you the valuation question. So Dell certainly participated in the storage virtualization craze. You saw data domain get bought for 2.5 billion for a $3 billion market. Essentially, yes, three par. We were sort of talking about what HP really sort of had to fill a hole. So they sort of overpaid in the bidding war with Dell. Do you think the flash valuations are justified because they're going to replace spinning disk? Will it be greater than that last wave, which was what, 2011, 2012? Well, I'm not a banker, so that's why I can ask you that question. I really believe that what you're going to see is the price of flash has been coming down dramatically. Faster than spinning disk, obviously. Spinning disk. And Dell will optimize the flash in our compelling line. And we're actually with the ecological also and also in our power board, which is our low end. But what we're finding is we still need spinning media. So we go cheap and deep. We're now using MLC for a petabyte. You know, it's funny, I was looking in our lab, we have a 16 new rack with a petabyte of data. If you would ever told me I would have a petabyte of data in the size of a filing cabinet, I would have thrown you out right on to Powell Street. You know, the economics are changing. The form factors are changing. We're moving faster than we've ever moved before. So you've seen a spate of acquisitions of all flash companies, obviously, you know, guys like, you know, EMC HP chose not to IBM bought TMS. What's Dell's position on that? Do you I don't comment on that position. You have a but what's your position on all flash array? Do you guys have a solution there? Yes, we do. Do you have a gap to fill in your portfolio? We have an all flash array. We are shipping it today in our compelling line. It's doing extremely well in the marketplace and it's growing as fast as we would like it right now. I won't comment on other companies. I watch what they do. There is still some trepidation because guys like me, who I've been around a long time, worry about high availability. We're now into a ratio encoding for recovery. Raid doesn't always cut it. So we're really looking at a not an evolution, but a revolution and people have to deal with it. Sam, how should we look at EVO rails? You guys have got a good relationship with Nutanix, right? A lot of people saying, oh, this is a shot across the bow from from VMware at Nutanix and SimpliVity. We were talking yesterday, so not really so much. Help us squint through that. Okay, here's the thing. Each one has their strengths. We're very, we're very good friends with Nutanix. Actually, we have an OEM agreement with them that was signed recently. We announced an appliance with them. Their great partner D Raj has been fantastic for us, whose Nutanix, Nutanix really optimizes something that nobody really tried to do, which is optimization of IO degradation and impact analysis. And they do it as good, if not better than anybody. Well, they also they also prove the winning consumption model still is software, but the appliance is not dead. Obviously, Nutanix is winning their D Raj. You know, we had him on the cube really first year we started and he was telling me that he almost didn't get funded. They're really convinced the venture capitalist that his model looked weird. They thought his model was weird, but it turns out it's a winning formula. So I got to ask you the question of pivot on that. What is the thing that you see right now that no one's seeing relative to things that are opportunities for the next Nutanix, that next startup, that next approach dealing with hyperconverged, also EVO rail validates certain things, but also you got the hybrid cloud. Is there a winning formula that's unique that's not yet on the hype curve? They'll basically unlike listening to Gartner and Forrester and all that, we embarked on a strategy of going out with appliances. And why we want appliances is because every customer came to us and said, we don't want to have to stand this up. If you look at EVO rails to stand up what is in the EVO rail, it would take 14 hours today on stage. They claim 15 minutes. We have faster time in our labs. They want to plug it in and they want to go with it. They're not happy of, and my analogy is a mechanic always wants to go under the hood in the new cars. And when it gets under there, he finds out he can't touch the spark plug. Every technician wants to go under the hood of an appliance. And for all intents and purposes, it's a 2U. And if you don't know what a 2U is, that's one and a half inches and it's downstairs. And basically, it has four nodes in it and it can stand up 100 BMS. Every customer we talk to says, that's perfect for us either in a rack and we're definitely involved with the EVO rack and EVO rail is very, very good for people who want to use campus type computing. So what's the difference between Nutonix and VMware? They're complementary. You could put an EVO rail in and you could put Nutonix in and Nutonix has plugins for VMware. And I'll tell you, whoever optimizes the IO Blender issue where... What is the IO Blender issue? The IO Blender issue is everybody believes that when you do vSAN, you've got unlimited capacity under there and if you don't plan the way you have those disks, you think you're the only one having the disks, there are 20 other VMs trying to hit that disk and they're queuing. And as they're queuing, it's usually one storage controller handling it. So what Nutonix did is they moved the storage controller up and so did VMware up into the hypervisor. What Nutonix has that we liked was some very interesting technology for having all the pool storage being able to fix the IO Blender. VMware is working on the same thing. There's nothing, by the way, if you look at it when we did virtualization, going back in history again, John always likes to remind me how old I am. If you look at it, we had, if you remember when we did VM on a mainframe, we had 16% degradation on a mainframe. Today, that's less than 3%. Why? Because we kept finding microcode assists and everything to fix it. That's what we're looking at. And at Dell, we recognize that you got a partner to win. You can't possibly believe that you have all the knowledge in one company. And so Nutonix is a partner, VMware is a partner, we have other partners to be announced shortly, like on a Hadoop. We partnered with Cloudera and then we partnered with Databricks to do Spark. We can't do it all ourselves. Sam, we really appreciate you coming on the cube, kicking off day two here in the cube. I'll give you the final word. All this is great stuff. You guys are doing some good work, but the battle seems to be in the software stack. So end the segment by sharing your thoughts on the software stack, because if you do connect the dots, the battleground will be the software stack. What's your vision there and how do you see that playing out? We absolutely believe that. We, our best partner is Dell Software and Dell Software is working with us to make sure that we stay ahead, but at the same time, we're working with companies. Nutonix for all intents and purposes, even though they run on an appliance, is a software. It's an SDS play. VMware is really a software data center play. And what we're going to do is add our value in the hardware and the software to make it the best experience. Look, it's all in service and support. Michael said that many times and we're dead serious. What we want to do is make it manageable, easier and reliable. And if we do those, we're going to win in this marketplace. St. Greenblatt, CTO, Dell's Engineering Services. An old man. No, you're experienced. Software mainframe. Never too old experience matters, especially in this transformation. We're here in the Cube Day 2 VMworld. The tectonic plates are shifting here in the VMworld. No limits, Dave. Earthquakes and all. We're still here. We're rocking. This is the Cube Day 2. We'll be right back with our next guest after this short break. OK, thank you.