 Live from New York, it's theCUBE. Covering Big Data New York City 2016. Brought to you by headline sponsors, Cisco, IBM, NVIDIA, and our ecosystem sponsors. Now, here are your hosts, Dave Vellante and Peter Burris. Welcome back to New York City, everybody. This is theCUBE with a worldwide leader in live tech coverage. We're here as part of Strata plus a dupe world. This is what we call Big Data NYC, kind of a show within a show. Todd Brannon is here. He's the director of UCS Marketing at Cisco. Todd, welcome to theCUBE. Good to have you on. Thanks guys, good to be here. So I have to admit, back when Cisco announced that it was getting into the server business, that was real skeptical. A lot of people were. I said, wow, this is like oil and water, fish out of water, whatever you want to say. And I have to say I was wrong. You guys changed the industry. I misunderstood the strategy. Obviously hindsight is 20-20, but I got to give you props. Thank you. No, well, I mean, I think it's because we didn't enter the server business per se. Exactly. That was the last thing we wanted to do. What we wanted to do is really re-architect the way it was all connected together. Because if you looked at the compute, the networking, the storage elements, the complexity in each of those wasn't too great, but it was all the connections together, especially when you dial back to something like, oh nine, when we entered the market, everybody was trying to virtualize. I don't think anybody had in mind that continuous delivery models are out there today. And we actually, what we built then, really kind of shot ahead of the duck in that sense in terms of programmability, but at that point in time, we were really focused on creating a singular platform with the network fabric at the center, and it's really paid off. Well, even then, I have to say, I was skeptical. And the reason was organizational. I said, well, it's going to be hard to heard all these storage and networking and server cats. But what happened is, correct me if I'm wrong, is the organization said, we're doing it. And then the organization sort of followed along. Is that what happened? Yeah, that's fair. I mean, there's still some of that, but I think you're right. I mean, the technology's converging, and the organizations are converging. They have to, right? You can't have people that are just kind of very narrowly focused on one of the technology domains. The real premium, I think we're all seeing now, is on the IT generalist, especially in infrastructure. Well, yeah, and we just wrote a piece on this. We called the Iron Triangle, we're overcoming the Iron Triangle of IT, where the Iron Triangle is the technology and the invested interest from specific technology vendors. The degree of automation which has been built around specific types of technology and the skill sets that have been specialized to specific types of technology. If you're gonna have hyper convergence of technology, you also have to have hyper convergence of automation and hyper convergence of talent. And that's not a trivial undertaking. It's not. So one of the things that's interesting, if you think about what is happening with infrastructure for the longest time, the server was the centerpiece. It was, yep. Now we're talking here at Hadoop World and elsewhere about the ascension of data as the dominant resource within how you think about your IT organization. We're increasingly thinking about data and the role that data plays and the physical constraints associated with data from a speed of light standpoint or an error propagation standpoint, whatever else. Data gravity. Data gravity, how that impacts your architectural decisions. So we're seeing a shift from the server out to the storage. But what everybody forgets is all of that is made possible by what the network's able to do. Well, it's crucial. I mean, when you think about distributed applications of this sort, the data center fabric becomes the backplane of the application or the backplane of the system, if you will. So that fabric element is critical. And that's why, you know, when we looked at the hyper convergence as a particular solution, we didn't know it at the time when we designed UCS but you take the clock forward to when we introduced that this year, we realized we'd already done the heavy lifting of making a fabric-centric approach and then we could put that software-defined storage stack on top and all that east-west connectivity and the programmability there, really essential and a really great foundation. The network's the backplane. And Flash changed the game. The disk used to be, the spinning disk used to be the bottleneck and now with Flash, it's like, oh wow, network has become so much more crucial. Or puts, and David Floyer, who is one of the Wikibon analysts, does or not has done a leadership job of articulating this. It's the fact that spinning disk created enough latency, which you didn't have to worry about how everything else coming. Now with Flash, that goes away, which has an enormous impact on the characteristics of designs and the architecture elsewhere. Only shifts the presser point back out to the fact. Right, right. So, and that has affected your partnerships, right? Because people are now realizing how difficult that is. Cisco's a leader. So a lot of companies are like, wow, we do business with Cisco, that's great. So you started with the whole VCE thing. You've got now business with IBM. I mean, you work with, I mean, NetApp, you work with all the major players. So that's been a brilliant market entry strategy. Well, I mean, we, I think with VCE, it was really ahead of its time in terms of converged infrastructure. And it just made sense to follow our customers, whatever storage infrastructure they were most comfortable with, because people build a lot of their data life cycle management around that. Historically, we're building a lot of that around the sand. So it was incumbent upon us to have partnerships across the spectrum there. But I think the pendulum's swinging. You know, if you think about when we, when X86 computing first came about, we called them PC servers, because they were just big PCs, and the storage was there in the system. You know, as an industry, we took it out for a lot of reasons, put it in the sand, but now the pendulum's really swinging back to servers in many ways, especially what we're seeing here at this show. So that idea of storage, you know, and compute being very closely married in the infrastructure is critical to a lot of these apps. You know, we think about something like video surveillance as a problem five or six years ago is really a problem of data ingest and storage, but now you throw on facial analytics, right? In a security context, try and find the bad guys, or maybe it's license plate recognition with a transit type system. So now you're doing a lot of analytic component on the data as it's streaming. Everything's moving from a stack to a flow. And so all of that requires the storage and the compute to be very closely coupled. So it's coming back into the server just full server, full server. But it also requires, and I know you've a strong architectural background and a lot of your brethren, a lot of user companies are now saying, well, do I do on-premise? Do I do cloud? What's the combination? One of the things that's clear is, again, we used to always think about the server and everything else was a peripheral. One of the things that's clear is wherever you put your storage and compute. And again, speed of light's going to have enormous impact on this. You're always going to have a network on-premise. Correct. So talk a little bit about how that's going to impact decisions at an architectural level for everybody. Absolutely. Because to my thinking, and this is why, to my thinking, I always thought, well, that Cisco was going to end up a major player across the hyperconverged because you were always going to need a network. Sure. And I think we see a lot of hybrid cloud today, but we don't see a lot of hybrid applications because to your point, moving that data in and out of the cloud, there's a latency element, but there's also just a cost element. In many ways, the cloud service providers, it's kind of a loss leader, the compute side of it. They make all their money on the storage. So for a lot of customers, the cloud's going to be right obviously for immediacy and the incubated applications, that type of thing. But when you really do the math, the on-prem infrastructure outperforms in terms of TCO, and then there's all the elements of cloud around data sovereignty, these types of things, there's still a lot of server huggers out there. But to your point, the architectures are getting much more horizontal, and we're definitely in an advantage position with the fabric play we've got. What are some of the more interesting use cases at scale, maybe you could give some examples. So Strata, we talk about big things at Strata, right? We just did the Olympics down in Rio. So we did London, Rio, we're going to go to Tokyo. And the scale of that thing was just enormous. So we had, if you think about the event itself, there were something like 15,000 athletes from 200 countries. On the ground, they had 85,000 security personnel, 70,000 volunteers, 25,000 members of the media, and that's before the spectators even got there. And then you've got all the venues, right? All the cameras, the timers, the sensors, and we had one network there that connected all that together. And what was incredible is that that was the most attack network on the planet during the period of the Olympics. We took on about 23 million attacks and blocked those. So kind of a new record, I think, for kind of a security play. But the cool thing about it, what that pervasive connectivity enabled was, NBC, we partnered with them. They broadcast the games to five billion people, but that's kind of the traditional monolithic broadcast approach. There were 170,000 hours of online video content posted out of the Olympic Games. That's almost 20 years of video, right? Viewed 24-7, and my daughter, I think she is online 24-7, watching video, right? And she's of a generation that can watch two or three video streams at the same time. So it wouldn't take her 20 years to get through it. But the thing about it that really changes everything is all these people that can't make it to the games to follow a specific athlete or a specific sport can now connect to the games in a completely different way because we were able to completely connect everything at the games and just all that data and then securely offer it up. So, and there's a lot of weird stuff at the Olympics, right? Like dressage, horse dancing, right? And so there's a lot of things that, you know, these little niche audiences, just the story of the long tail. Well, and it was interesting, the story of how NBC dealt with that. Instead of sending a couple thousand people to the Olympics, they put 2,000 people in the warehouse in Connecticut for the remote video for all that online stuff. There's a lot of tricasters there, boys, at outstanding. Okay, good. Can we go back to hyperconverge for a minute? Can you talk a little bit more about your strategy there? Maybe people aren't as familiar with what Cisco's doing. Yeah, so back in March, we announced a new product series called Hyperflex. And so what we did, we looked at this market over the past couple of years. And what we observed was that the first movers, the first generation players, had really done a pretty good job with maybe the server and the software-defined stack, but they hadn't really looked at the fabric. And these are clustered fabric-connected systems where you're distributing the storage. So the fabric's critical, and we didn't see a lot of that in the first kind of generation design. So we thought UCS would be a great platform for it. We found a partner called Springpath, and they did something really interesting where, you know, again, a lot of the first generation players use the EXT-4 file system out of Linux as kind of the basis. And they kind of glommed their platform on top. The guys at Springpath, they wrote a file system, rather, from Scratch. So an object-based file system. And really kind of changed up the way the data is distributed over the clusters. So we invested in their technology. And so we productized that, took it to market. I think in the first four months that we were shipping, we've taken on 500 customers onto the platform all over the world. So the thing that surprised me, though, is the type of workloads that people are putting on a hyper-converged platform. So I think we've all kind of thought of it as kind of a lightweight play in many ways for virtual machines. But we're seeing customers put CRM systems on there, ERP, heavy database, because the performance is starting to get there with a system like Hyperflex. So we're really excited about it. I mean, it's all about simplicity, right? And again, the automation tools, the control tools are all starting to scale with the ability to add this stuff to the fabric. Absolutely. I mean, our customers want to be out of the business of administering hardware. And we've been talking about this for a long time in the industry, but now we've got to a point where we can come in, after they do the rack and stack, it only takes about 34 minutes, we've timed it to deploy a cluster. And then it's all managed out of vCenter, right? So there's no new tools to pick up, right? That's really what customers are coming to us and saying is like, don't give us new islands of things to manage. Give us something that integrates into the existing paradigm, but make it just dog simple to run. And that's where we're going with Hyperconverge. So strategy-wise, you guys were, I guess you weren't the first in Converge. I guess Teradata was the first, right? But anyway, it's probably right. And maybe Exadata, okay. But in terms of horizontal infrastructure, you really changed the game. And a lot of people, and I know you can't talk specifics about M&A, but a lot of people predicted, oh, Cisco's gonna buy NetApp, or Cisco has to do something in storage. And you sort of listen to Chuck Robbins and you say, he's thinking the internet of everything. And so my question is, when you think about how you change the game and compute, and you look at storage, I've never been one to say, oh, Cisco must expand its TAM by going out and buying a storage company. But rather, things like you're doing with Springpath allow you to sort of change the game. And pure storage and a number of others. Yes, that's right. And using a partnership approach and a software-defined approach, maybe you could talk strategy for a moment. Well, I mean, I think, again, that the pendulum is swinging back, more and more workloads are natively distributing across x86, software-defined storage obviously does that, but like a lot of things we're talking about here at the show like Hadoop, a lot of these applications, they're natively scaling across x86. So a strategic focus for us are the products in our server portfolio that can better power those types of workloads. So you're gonna see us, actually in November at our worldwide partner summit, we're gonna have an announcement in this space because we're investing pretty heavily in technology in that scale-out server domain. But strategically, obviously, the SAN isn't going anywhere. I mean, it's a fantastic platform, especially when you're talking about shared storage for the virtualized environments. A lot of customers, they've just made that leap to virtualizing all their applications or workloads. And a lot of that is based on that paradigm of centralized storage. So our strategy is obviously, we're gonna continue to drive hard with all of our partners on the strategic storage side. But as the pendulum swings to more of the x86, you're gonna see us announce and come forward with some really innovative technologies that kind of pivot back into our fabric approach. And then how do we help people, because the scaling's the key here. These big data environments that don't get smaller, they just grow. So being able to rapidly scale out and do that through automation and give the developer a storage target that they can control through APIs, right? And infrastructure is code. Those are the areas we're really focused on in terms of our investment, is that automation, the simplicity, the control plane, if you will, on the x86 side. All right, Todd, we'll leave it there. Thanks very much for coming by theCUBE. Appreciate it. All right, keep it right there, everybody. We'll be back with our next guest at theCUBE, we're live from New York City, right back.