 World 2015, brought to you by VMware and its ecosystem sponsors. And now your host, Dave Vellante. Welcome back to Moscone, everybody. This is Dave Vellante. And this is theCUBE. We're here at Silicon Angles continuous production of VMworld 2015. This is our sixth year here. George Desheres here is the president and CEO of DataCore software and Dustin Fennell who's the vice president and CIO of Epic Management. Formerly CIO of Mission Community Hospital. Gentlemen, welcome to theCUBE. George, great to see you back again. Hi, senior day. So, VMworld, we have just seen an amazing transformation of VMware as a company. You've been in this industry, you've seen just such amazing evolution. What do you make of this show? What do you make of where VMware is today? To me, it's pretty exciting. This whole virtualization thing. I mean, I was there at the forefront with storage virtualization back in 1998 when we kicked off our company, DataCore, to really do storage virtualization. And when I look at it at that time, we were focused on how the industry was really stuck on IO performance and storage performance. But storage virtualization, even though I thought at that time was gonna explode, it took a lot longer to take off. And when you really look at it, the first big explosion in the market was VMware with the hypervisors. And they were able to do a great job of basically using more of the capacity of the CPUs out there, get more work done on the same platforms. And that led to the whole consolidation, more VMs. And voila, VMware has become an empire out here. And it's great. I can't believe when I walk around the show. And now I'm, you know, the only bad part is instead of all the technology and architecture, it feels like architecture is everywhere. It's hyper this and hyper that. And, you know, the fundamentals I start worrying about because things like virtualization in the early days was real pure. It was basically abstract away from the hardware and, you know, just let this thing run on top. Now what I'm seeing is people are talking, quote, VM aware, which sounds a lot like making sure some flash devices are tied to a VM, which goes right back to sort of the antithesis of virtualization. But overall, it's pretty cool. The VMware has really moved forward. So Dustin, I have to ask you the CIO perspective. A lot of CIOs I talked to, spent a lot of time on the application portfolio and trying to drive business value into business process, but then they get sucked into infrastructure. As a CIO, do you just want to forget about infrastructure or is infrastructure management essentially a fundamental part of what IT has to deliver? I think it's a fundamental part of what we have to deliver, but I think we need to trash the way we are used to delivering it. What do you mean by that? So I think we need to look at how we deliver services and get away from trying to manage and hold on and own everything and deliver our services in a way where we're not managing the end device anymore. We're not worried about hardware, whether that be storage hardware, whether it be server hardware, and deliver our services in a much more distributed manner because our end users are demanding that they want to access those applications via whatever device they happen to bring and wherever they happen to be, whether we own and manage the device or whether it's their own personal device at home, work, wherever. And we have to be able to do that in a way that we can manage it and keep our data secure. So George, we were talking off camera about the heritage, your background came out of the technical computing world, on-core days, and that world was always data-driven. It's a lot of scale out. And you started data core in an era where that hadn't seeped into the enterprise, but now this is a big tailwind for you. So talk about the evolution of architectures, if you will, and you were ahead of the curve, and now it's wheelhouse for you. We would describe that a little bit. Sure. I mean, it's kind of funny how things kind of go into cycles in our industry, but if I look back in the 80s and 90s, which kind of dates me, and... Seems like yesterday, George, to me. The craft at that time, you looked at microprocessors and it was really obvious when they came out, it was going to create the PC revolution. And the other one was supposed to be take a bunch of those microprocessors, make a big box out of them, and in effect do parallel computing. So there were two trends. Parallel computing for the really big jobs and personal computers. Well, at that time, parallel computing was also focused on sort of three embarrassingly parallel things. First one was sort of time sharing, it was the easiest one. You know, basically multitasking switches out and it becomes have a big machine with terminals. The terminals become PCs. The PCs become virtual machines and they get wider threads and you become containers. Okay, I just went through history in a very quick way. By the way, what's the cloud? Time share. Bunch of microprocessors doing the same thing. The other big one back in that period was databases. Databases, in fact, if you remember Oracle and Ellison, Oracle 7, parallel server was going to be the way they buy you in cube, they do all this kind of stuff. What actually happens? Distributed computing clusters, they're good enough. He's a commercial guy, he makes money and it kind of gets sideline. What was the third one? Parallel IO. That's what we used to do at Encore. That was our big claim to fame. When we got into data core, we actually were a parallel IO company. And the first thing you got to do is you don't want to deal with all this hardware so you create an abstraction layer, i.e. virtualization. So that you can make it parallel. Problem was, multi-processing was awfully expensive, awfully complex, and even though the promise was it was going to get there, it didn't quite work at that time. So it took a lot longer. So the byproduct was storage virtualization. 1998, we declared it. We went out, we made a company, a very successful company that's been out there called Data Core. Now fast forward, what's been happening in this industry? You got Moore's Law going along. It talks about every two years doubling the transistors. Except what people heard was the gigahertz was going to get faster, the clock speeds, frequency scaling. So we all remember it went one gigahertz, two gigahertz, three gigahertz, four gigahertz. Where are we today? Two gigahertz. What happened? Multi-cores. Last five, six years they've exploded. There is more multi-cores out there than anything else. They're sitting on every server you got. They're like free. Who's the first guys to take advantage of it? VMware, the hypervisors. They start putting VMs on these. But you start getting diminishing returns. Why? Because as you add more VM density, all the IOs concentrated and you got an IOS drop. Diminishing returns from a performance standpoint. Performance. Not at your price per core though. Good returns. Good returns from a commercial. But basically the biggest IT storage problem is the IO performance. It is bottlenecked this whole industry. But those multi-cores are sitting out there and they're completely available. So what's data core done? We went back and re-looked at our own code. What we have in San Sifany. We did some benchmarking and what we're seeing is pretty fundamental. We can unlock those multi-cores to basically sit there and suck all that IO through them and go back to parallel IO. Fundamentally we're seeing an order of magnitude better performance. What does that mean? That means you get more VMs density on those same platforms tremendously faster IO. We think this is the breakthrough for the next generation. So much like storage virtualization kicked off in those early days, I feel like I'm right back and I've got sort of all that motivation to drive this forward. But right now we're sitting on servers out there that are not doing what they could and data core is going to bring that to the fore. The virtualization, software led, distributed computing, all things that have sort of come around. Yeah, absolutely. Tailwind. I want to go back to Dustin and get your perspective. So you've been a data core customer early on. Correct. So your previous rollout to hospital. What led you to data core originally is sort of, you were taking a chance back then, right? It was different. And actually data core customer even prior to that when I was in the higher education domain at Scottsdale Community College, we implemented data core. Okay, so some cost sensitive environments, right? Hospital, higher ed. Right. And so our whole strategy was to really become hardware agnostic, not only on the server side but on the data, on the storage side. And in addition, those environments typically do not have a dedicated storage guy. You have some infrastructure folks who, that is just one of the many hats that they wear. And we have varying different types of storage and really don't have the expertise or time or budget to make sure that we have the proper training in the management of all of those various different storage types. So being able to put our storage behind data core, have one pane of window to manage all those varying storage infrastructures from and to learn one management console is a huge advantage to us. Okay, so that was the simplicity that was the allure at the time. And now you've been a customer for quite some time. How have you seen data core evolve? What, what kinds of things are you asking for? You know, would you push them to deliver and? I think it's just the, especially on the performance side for us to be able to, especially in healthcare, we have all these varying systems. If you look in a typical hospital data center, you can actually point out, well, you know, there's a project, there's a project, there's a project, there's a system. Everything's just kind of built in these little silos. But with healthcare today, they're wanting all these siloed systems to actually work together as one cohesive system. And most of them have varying different types of storage behind them. And to be able to wrap all of those storage types behind a single package, single pane of glass and be able to manage them independently, but through a single management console is a huge advantage to us. And the performance as more and more data is being collected and leveraged, all that throughput is extremely important. So it wasn't just the cost savings, there was a business process angle here as well. Is that right? Describe that a little bit. I mean, just in terms of what did that do? It just got you to market faster, it allowed you to re-architect business processes? It allowed us to re-architect our infrastructure faster. It allowed us to leverage our current investment for longer. We could take what used to be tier one storage and just move it right down the line and add additional storage and just move things around much better with lower staff and really lower expertise. Didn't take the expertise that it would have in the past without data core. All right, so George, I got to ask you about a couple of other trends that we're seeing. So you got VMware now getting finally getting into the storage business. UFC finally let them. So they're going hard after VSAN or hard in quotes. You see the hyperconverged guys now taking off. Where do you guys fit? What's your take on each of those trends? Well, first of all, the hyperconverged, there's a couple of things I see with it. One is they're putting all this stuff together and if that's the case, you better be able to do the storage component if you're going to be doing also the compute component and the network component. And what I find is they're way underpowered for this stuff, just ask the database guys trying to do database workloads on these things. So one of the things that's really key here is how do we take hyperconverged to the next level in performance. So obviously the parallel IO is a key component for me. And what we want to prove for instance is that you can run with data core and get the performance of sort of the enterprise class storage on a hyperconverged platform. Now, we're working with the likes of Lenovo, Dell, Huawei, Fujitsu to build systems so that people can have hyperconverged. The problem with these hyperconverged though is they also live as separate islands. So people go, let's go get a hyperconverged system. But what happens a year from now when you want to now tie it back to the rest of your organization? What do you do? You're stuck again. So part of the thing with data core is I can work across that storage network everywhere. But also I can help you build a hyperconverged system that is part of a network because the reality is this thing about all storage being on a hyperconverged platform is a little bit idiotic if you ask me. The world was always sort of some stuff is centralized, some stuff is distributed. And it's not just because you can centralize it all. It's because some workloads you want to have across different locations. Some stuff you want to do in a central location to get management efficiencies. So there's always going to be an issue of, you know, a network between some of these. And what you want is a system that can basically work as Dustin said, with a single pane of glass that will let you manage your hyperconverged as well as your sand. And that's what data core can do. So you're the umbrella? We're the umbrella across that. Okay, how about vSand? VSand is interesting. The one thing I don't get is every storage system, and you know this, that's out there has been done with two nodes. And they've done it for a long time to do high availability. Have you noticed that all of these hyperconverged need three or four? And then when you go to multiple sites, you're gonna have three or four? Or you have witness nodes and all this kind of stuff? I don't get it. How come the whole storage industry did it in two nodes? So fundamentally, part of the thing we're doing at the show is talking about, get, you know, do more with less, right? I know it's not original from a marketing side, but it's true. But it resonates to IT guys. And you know, complexity is when you add a third node, you're adding more complexity. So we can do what VMware Virtual Sand does with two nodes versus four nodes. We are five, 10 times the performance. So the big difference for data core is we're focused on, I would call the more enterprise class workloads. But in terms of what they're doing, in terms of these sort of hyperconverged, the two trends that bugged me the most is one, they're not really looking at how to tie it with the rest of the architectures. How does it also tie with existing storage and existing equipment so that it's not just by one piece? And then the other disturbing trend is they're going against virtualization by trying to hardwire to specific flash devices or specific hardware, and then they use some architecture terms like it's VMware to say that it's somehow special, but that special is what locks you in. And from data core and from our view of virtualization, it's about being independent and decoupling from that hardware. So that flash affinity to VMware affinity, you're saying is a reversion back to the old ways. It's a complete contradiction to virtualization if you think about it. Don't get me wrong, I love flash, it's a good thing. Why do I care which one? Why do I care how it's tied to my system? Data core, we do 15 levels of auto-tearing. You want more cash, you want more flash, put it in the system. Software should be smart enough to manage it and handle it. This idea of quote, having to get back to specifics, what are we going back to the hardware days? So how has that, Dustin, how did that change the way in which you managed infrastructure? You said earlier on, you wanted to be sort of software-driven. How did that change your environment? I guess you didn't have the resources to throw bodies at the problem, so that enabled you to move forward. Maybe not do more with less, do more with the same staff, maybe less hardware or whatever hardware you know, but talk about the impact that that had on your businesses over the years. Well, one of the byproducts of the Affordable Care Act in healthcare is reduced revenue, reduced reimbursement rates. So we're actually being forced to actually provide more integration, more access with reduced revenue, and so we have to figure out strategies of how do I take my same staff or even less staff in some cases and still provide even better services and integration? So from my viewpoint, it's got to be simple. It's got to keep me away from being tied in to any one particular hardware manufacturer so that when I have a project or a new system, I can buy just what I need, so whatever I buy needs to be scalable, and it needs to be whatever's the best at that particular time rather than having to be locked in until I'm able to refresh that entire hardware system. So keeping with the true reasons you want to move to virtualization, I think this tying back to hardware is just kind of locks you in and takes you back the other direction. It needs to continue to move forward where I can just, whatever hardware I want to use is what I want to use. So, Joe, we're running out of time here, but I wanted to ask you about the company. I mean, you're very successful. You've got a stack that's pretty hardened. Storage stacks are very difficult to build. You're seeing a lot of innovation come out, particularly these flash guys are coming out and the stack is not hardened yet. Some of them are, some of them are. They're still pretty elementary. Or some of them are like the three-par guys. They did a decent job and they've got these services, but a lot of them are sort of fragile. You've built up over a decade. 18 years almost. Plus, this really robust platform. What do you want to do with the company? You want to stay private. You want to do IPO. You want to get bought. I mean, it's rare, but a company so successful is stayed private. What's great, but talk about that a little bit. Sure, I mean, from our side, the funny part is I really thought when we started the company, seven years, everybody would figure out storage virtualization and we would be sitting out there. Now, it's taken longer, but what's interesting is right now, from our standpoint, we're cash profitable. I look at all these VCs out there and all the startups and I'm blown away because most of them are spending $4 to make a dollar in sales. So a lot of marketing and a lot of hype going on. We're doing this cash profitable and frankly, we're running a very successful business out there, so where do we go next? If we go logically, it's an IPO and that's probably a very good way to go for us. From a bigger standpoint though, I think the revolutionary breakthrough that we're talking about with Parallel IO could change this game so big for us that I'm just amazed again because where we're sitting right now is I'm watching an industry that has gone to virtualization that's done all this consolidation, that's sitting on multicores and yet is sucking IO through a straw. And we wanna smash that straw and multicores are sort of like these nuclear powered turbines that can do it and we think that we're there and with that, you know what, I may be getting old but I'm ready to go take it on again and we're gonna drive this thing all the way out. It's great, well George, congratulations on the success. Dustin, thanks for coming on and sharing your stories with us. Really appreciate it. Thank you, appreciate it. All right, keep it right there everybody. Check out CrowdChat.net slash VMworld. We've got a CrowdChat going on. Check out Wikibon.com for all the research. Siliconangle.com for all the news and Siliconangle.tv is where you'll find all the replays of these videos. Keep it right there, we'll be back right after this short break.