 Okay, welcome back everyone. We are live in Silicon Valley. This is theCUBE, our flagship program. We go out to the events, extract the signal from the noise. I'm John Furrier, the founder of SiliconANGLE, drove my coast. Stu Miniman, chief analyst at Wikibon.org. Our next guest is Dave Ward, CTO of Engineering and chief architect at Cisco. Welcome to theCUBE. Hey, thanks guys, appreciate it. Yeah, we love getting tech athletes like yourself. We call them tech athletes. Guys that in the trenches making big things happen. So I got to ask you, your take of this, Cisco is the big player in networking. Networking is the future of networking here at ONS Summit. What's your take on all that? I mean, what's the take on the whole scene right here, the vibe and the future of networking? I think the vibe is great. You know, I've been working in this community for a long time, really trying to catalyze programmatic interfaces to networking. It's good to see that as well, the ONS conference and group is now also bringing in network function virtualization, discussion of controllers, but my favorite part is that products are emerging, deployments are emerging. That was a long time coming and so I think it's a great change for ONS to start seeing the products in deployment. So what's your take on software defined networking? That was an interesting thing when this year was bought by VMware, it was kind of the shot hurt around the world. Whoa, something's happening. And then the marketing hype kicked in, software defined data center, which still is kind of like, what the hell does that mean? But software center, the value proposition still is what people are talking about. So what is the software paradigm that's changing the game at the networking layer? Well, let's first talk about this. So there was a lot of hype. I actually think 2014, we're now in the post type SDN cycle and what I mean by that is again, back to products. And so what's key is that not only are there some early mergers like you talked about with NSX and Martin Casada who spoke earlier today, but there's now a number of SDN protocols that a number of standards bodies and a lot of them being represented, not standards bodies, but the technologies being represented here. SDN is just a pillar of a networking architecture. Network function virtualization is a pillar of that architecture. So it was cloud, you know, let's just throw all the buzzwords as long as we're going through the whole litany of them. The thing is, is that to build out a solution or a deployment, all of those pillars have to come together. And being one who's, you know, help bring OpenFlow out of Gates 104 out of Stanford, I'm, you know, I'm extremely happy that programmatic interfaces are here for now for networking. So let me, let me tell you this, John. So programmatic interfaces, we actually, there's been a long history of this. OpenFlow has gotten a lot of the attention over the last couple of years, deservedly so as it did help catalyze what we're doing here. But look, there's been Corbus, there's been GSMPs, there's been a zillion of these different protocols out there. Why weren't they successful? It's kind of an interesting, interesting case and a lot of it has to do with the protocol architecture. A lot of it has to do with history. And SDN is here now. We're beyond the hype. We are seeing products. We're seeing people deploy this. We're deploying with customers. In fact, we're deploying with dozens and dozens of customers, our SDN solution. And really even a big player like Cisco has fundamentally turned how they think about programmatic interfaces from maybe someone was thinking a threat. He said, oh my God, something's going on here. Let's just call it what it is. You know, people were worried about the threat and worried about disrupting even Cisco's traditional hardware. But innovation is there, right? I mean, talk about that. What are innovations happening? Share with the folks out there your perspective because something happened, but what's the innovation coming from that? What's the plan? So the innovation not only is around the technology, but the innovation really is that there are now services and new services that can be deployed. There's new markets that are emerging associated with this by enabling faster service delivery, easier to operate networks, easier to operate, both whether it's enterprises, service providers, data centers. SDN is not just limited to the data center. It is a piece that allows programmatic interfaces, not only physical kit, virtual kit. That's really, really key, being able to tie together what's going on in the data center to what's going on in the network and link customers to what they want out of the network really, really quickly. So Dave, you know, one of the tenets of this show is it's about open networking. And of course, when you say open networking, there are plenty that say, well, the opposite of open networking is what we've had before. Cisco is the incumbent. You came from Stanford, but you're at Cisco now. You know, how does Cisco fit with this whole open networking trend and open source? Wasn't at Stanford, just to work with some friends there. I just want to make sure we don't mislead anybody out there. But so what's open about this not only is, of course, ONF, ONS, and working on OpenFlow, but there are a large number of protocols out there. The open piece is that number of SDOs, standard development organizations are working on it. There's now open source projects out there, not only multiple open source implementations of OpenFlow itself, but of these other SDN protocols. Open Daylight was launched, you know, just about a month ago. That's a great effort for the entire industry to come together. The fact that open means open standards, the fact that open means open source, the fact that open means open interoperability between a large number of vendors, to me, that's what we're talking about. And the fact that we can get to the point of open systems integration, and what I mean by that is virtual appliance from one, hardware from another, these are great things to create the ecosystem around the network. Dave, you've got to deal with a lot of different challenges. You've got this organic bottoms up community model technology within Cisco and within Cisco, and the customer requirements. Cisco has huge customer base. So you've got forces kind of coming from both sides. How do you balance that out from an architecture standpoint? How do you get these products to market? Just share some insight into what it takes and what are some of the architectural decisions you make, trade-offs, pros, cons, take us through the mindset of that and how that renders itself and ultimately into the product market. Sure, so first and foremost, when building out this equipment, it's always in the customer's mind. What problem are they trying to solve? First and foremost, there's no doubt. Sometimes it's go faster, sometimes it's make it easier to operate, sometimes it's create these service mashups that we talked about. So first, what did it take? Open interfaces into the operating system across the portfolio. The ability to build out these SDN protocols across the portfolio, virtualizing security, mobility, video, collaboration, voiceover IP as much of the portfolio as possible and be able to orchestrate across it. This is a big change for a big player like Cisco, which had been categorized as a hardware supplier now to really take a look at how to combine servers, switches, storage, routing and the solutions on top. So it's been a lot of work over the last couple years. Can you talk a little bit about how the decoupling of hardware and software is impacting what you see inside Cisco? So let's face it, there's a couple of things that we do differently and that we focus on and the decoupling of software means that, let's say we're talking about, we just came from Mobile World Congress a week ago, we virtualize evolved packet core from mobility. That's a service. Needs to run on top of an orchestration layer. That needs to speak SDN protocols to the hardware. That is part of the decoupling. Have the service actually be able to work on, let's say servers, maybe UCS servers or whatever the case might be, but have the service decoupled potentially directly from the exact hardware that's there. This allows us to take things like Mobile Gateway and be able to have features associated with it running out of the cloud. Given the use of those pillars of SDN, NFV and orchestration and that's the key. That's how they come together. You know, I always look at Cisco, looking at the history of the company. It's been a fabulous run. It's my generation. I remember when it was founded, watch it grow from a small company and now it is today. It's like that, there's so much going on. It's complex. You can't just change that airplane out and just bring in the new jets. So that's one challenge. The other thing is interesting. You go back just a couple of decades ago, it's like a pretty static environment. The provision, automation, configuration and the edge of the network was an endpoint. Now you have Internet of Things or Internet of Everything that you guys are doing. That changes the game of what the edge of the network is now with virtualization. How do you deal with that change of the edge relative to virtualization? How does that impact the products and things that you work on? So working across the whole portfolio, bringing in the Internet of Things, what's interesting is that Internet of Everything is just mind-blowingly larger of trying to discuss that. So we have to realize that even in Internet of Things and giving access to all those sensors, those happen, let's say healthcare. Let's say it's within oil and gas. Let's say it's within mining. There is still a lot of segmentation specialization that needs to happen. So therefore, yeah, a lot of basic stuff that happens around the perimeter. I need to provision this stuff. I'd like to automate it for a few minutes. Security, of course. But what's interesting is that what emerges next, I believe, from SDNNFV orchestration is the platform for the network. It is network infrastructure as a service. It is connecting all of these sensors and endpoints where let's just not talk about devices as endpoints, but let's talk about human beings and notion of identity. We don't want to be known as individual people being associated with different devices. We want to be able to have that common across. So the perimeter now also becomes the identity of the user. Things start getting really interesting at that point. It's complication. Dave, so talking about the Internet of Things, the two things that come to mind are analytics and security. Can you talk about how those are impacted by Internet of Things and what's going on with SDNN? So there's so many aspects to securing Internet of Things, of course. Not only the sensor itself, the device, how it attaches to the network, the data it's producing, this does, so how does this affect things? Security and analytics means that you're collecting, not only do you have multiple billions of devices attached to the network, you need an information hierarchy to be able to collect that data, process locally, and then aggregate it up, but those analytics on top are really end up being part of the service that's being delivered. Whether it's smart grid, smart cities, or in any of these segments that I mentioned before, where this is driving towards, is a platform to be able to understand the data that's being produced, the state of the device itself, or what it's monitoring, and then producing it into information so somebody can make a response. You mentioned earlier, customers at the forefront of your decision criteria. Obviously that's a huge customer base. Talk about some of the specific things that Cisco's doing, that's the reality. You said there's some real stuff hitting the table right now. Meat on the bone, as we always say. What's the meat on the bone right now from a product standpoint with Cisco that you can share? Sure, so a couple of things, and I'll give just a couple of examples that are perhaps known out there. At CES, we discussed and described cloud-based DVR system. This takes your DVR out of the house, puts it in the cloud, access on any device. How did we do it? Programmatic interface is out to both, not only inside the home, but to the networking devices that they attach to, et cetera. Programming out the cloud, orchestrating those virtualized network functions. We went from, because of this technology, we went from zero to greater than two million customers in less than 90 days. Now you start talking about how many folks are having DVRs in the house. We're talking about tens and tens of millions. That type of ramp was not possible without the technology that we're talking about here at O&S. Second, when we just come back from Mobile World Congress, it's all about virtual evolve packet core, virtualized GI LAN services interface, self-optimizing networks. These are different commercial segment-based words for programmatic interfaces to the network, virtualized network functions, and orchestrating it. What do we get out of this? We get the ability to combine video over mobility, bidirectional conferencing, and collaboration. Things that didn't exist before, but are now possible by creating these mashups. These are driven directly from our customer base. Let me give you one last example before I turn it over to you guys. The tradition, now focusing on service provider, looking at the traditional VPN business case, yeah, we provided some separation of address spacing and et cetera, but it was primarily a circuit-based business. How can we get enterprises to attach to the internet? Well, let's say, for the sake of this conversation, let's say that we're doing pretty well there. But what does every business wanna purchase or acquire or use? Firewalls, anti-DDoS, anti-malware, intrusion detection, et cetera. What's a great way to do that? Virtualized network functions. What does it require? Programming the access devices, the provider edge devices, programming the WAN, programming the cloud. These are all driven directly from our customers into our product base. Without this transition having occurred, we as an industry would have had a much, much harder time being able to estimate this. So orchestration and automation are huge there. What can you talk about that in terms of making that fully automated? Sure, so what we've demonstrated to a number of customers is that, and we'll go back to this VPN case, and we've also done this for residential users as well as mobile, here's my portal. My portal to a service provider, let's just say, says it's got a huge service catalog, Firewalls, anti-DDoS malware, LTE backup, whole voiceover IP, web scanning, email scanning, parental control, the whole deal. It's a big service catalog. I start clicking on these things, putting them into my shopping cart, hit buy, or hit go. We're able to orchestrate those into the cloud, connect that user directly into the cloud in that service chain being discussed through the principles we've talked about. The only human being who was required to activate that service is the person who wanted to buy that service. Fully automated antenna. Fully automated antenna. So Dave, you talked about some of the carriers and the service fighters and what they're doing. How about the enterprise? What tangible solutions are there today for them with the SDN NFE bag? Sure. In addition to some special built products like application-centric infrastructure, what this allows us to do in the enterprise is we take the notion of a workload and the policy associated within that policy, of course, can be security, load balancing, and a variety of other pieces in the data center. So in the enterprise, if we link that out to how am I attached in my, we'll call it my office, how am I attached over the Wi-Fi, I'm now able to take that user and their identity and link them to the workload and the policy end to end. And linking people to the activities they wanna do was not possible without the technologies that we're here discussing. So for the enterprise? Get, not only just fire up my workload in the way I want it and where I want it, but also have it meet the quality of experience I need and link me to it based upon my identity and the attributes associated with me. One of the things we're seeing in our data that we're monitoring across these different verticals is, this show is hitting a couple different markets, DevOps, SDN obviously, and software-defined data, so it's essentially cloud, hybrid cloud, those markets. So the question is for you to share with the folks out there is, everyone wants developers, give me influencers, I want to target developers. Well, what the hell does that mean right now? I mean, the old days Cisco had a very specific developer profile, now you got software as a big part of it, is it DevOps, is it a certain programming language? What does the developer landscape look like for you guys? Because now it's the disciplines are intersecting, it's some overlap, it's some blurring of the lines. Are there certain types of developers you guys look at that you target that are working on your systems? Are you guys doing any outreach? Could you share your perspective? Yeah, you bet. So what we had to do, and this started a couple years ago, was we realized we needed to restart, reset, de-goss, however you like to think about it, our developer program. Our developer program was centered around we're not talking about developers for Cisco, but instead I'm talking about catalyzing a developer community around networking. Outside of Cisco. Outside of Cisco. Now remember that we've got a huge channel partner, we've got a lot of IT professionals out there, we've got hundreds of thousands of CCIEs, all that have lots of finger memory associated with typing in those commands. Rule number one, never leave your community behind. Also, if we want to create network infrastructure as a service, if we want to, and then get this networking information up to the SaaS layer, we need to catalyze that developer community. So we needed to absolutely revamp the developer network to enable a DevOps model, to enable our partners to build custom solutions on top of the platform. Remember that developers use platforms. There's been far too much talk in my opinion of, you know, in Darth Vader terminology, Luke, use the API. You know, APIs are cool, but it's about a platform, and the platform isn't just give me a load. I got a REST API, great, good for you. Yeah, I mean, God bless you. SiliconANGLE's got a REST API. I got some of the back, you can take them on the overhead, you know, but nonetheless, my point is, catalyzing a developer community required us to do a couple of things, change our developer program, fundamentally change our strategy to utilizing open source as an industry de facto standard, when standards aren't yet to emerge, this is what's going to attract all sorts of new partners. Is that new, when did that happen, and what's the update on the developer traction? So the, right now, so we're going to launch it soon, but we've been working in beta with a lot of folks, it's out there, we're training up our partner community, we're developing the courses for CCIEs, we've discussed that at our, you know, our big conferences, which are called Cisco Live, as the case might be, and we have started catalyzing hackathons towards these platforms, merely because, not, sorry, not really, but because, developers are going to give you the best feedback. They're your users. Exactly. They're your users. They're the new IT professional. Yeah, if you're service cataloging, well, that's definitely happening. Service cataloging is happening. They're the users. So I got to ask you the next question is, as in Cisco internally is culturally changing, you might have a unique perspective on this. The always the term was, we have to move up the stack. We have to move up the stack. What does that mean, moving up the stack? Is it simply the developer as a user model? Is there a certain part of the stack that you say, hey, above that point we're done, or so from a Cisco internal to external bridge, what is moving up the stack mean? Sure, so what's the target? What's the lunar landing? Getting networking information and being able to modify the desired quality of experience, security, and other attributes from a SAS layer, from the application that's running a business layer. That's landing on the moon right now in the history of the internet. So what does that mean we need to build? We need to get those APIs out there. We need to get those controllers and orchestration mechanisms. We need to get to the point where network infrastructure as a service, which, look, I'm not pointing here, but most folks talk about SAS, PAS, and infrastructure as a service as a traditional layers. Our role is to get network infrastructure automated, fast, agile, easy to use. Working up the stack means get it into those other stacks where we currently don't build a complete virtual machine management stack. Talk about the customer now. The customer is obviously the pressing force to them as economics in the cloud is great. I got on-prem, that's not going anywhere soon from our data we're seeing. Across many shows we go to, of course, hybrid cloud will be a vehicle for some cloud stuff. But what is the mindset of the customer? How far are they along the lines of having a DevOps mindset? I mean, DevOps mindset is a whole, completely different ballgame. Pushing code, having code, write code, as we've heard here at this show. That's automation. You have all kinds of coolness going on around this world. So where is the customer mindset at this point? Are they just tipping their toe in the waters? Is it evolution? Top of the third inning? I don't know. Huge, huge continuum of where customers are at one end, which a lot of the industry and the market hyperlix to focus on are those specific folks who have large developer community as part of their business and let's rotate towards those. There is a huge continuum following from there all the way to, I want to buy a turnkey solution either from a supplier or from a partner to, you know, I know something is coming. Please don't leave me behind to help me change and transition my IT staff to help me begin to get them to be coders and build customized solutions. That continuum is broad. The vast majority are in that tail, right? There are only few, whether it's large enterprises or certain MSDCs or certain providers who have the capacity, capability, skills, et cetera, to be those big, I want a self-sufficient DevOps model. There's a big long tail of everyone who wants things from turnkey to help me get there. Yeah, so Dave, talk about these large cloud service provider customers. One of the big debates we're all having is five years from now, will most applications live in a data center that I kind of control or will it all go off to the cloud? And of course you look at the buying habits of, you know, Googles and Amazon to the world is quite different from the traditional enterprise. What's your guys' viewpoint as to where the market's heading and how that market split goes? So five years, tough to say, but needless to say, due to security, regulation, compliance, there is going to be and continue to be a ton of private data center. Those are, and let's also talk about one other thing which is value versus volume workload. The value workload is one which has to have very tight cycle completion, very tight features, policies surrounding it. That's what we're talking about with those private data centers. Flipping everything all the way over to the big public cloud, that focus there potentially could be on a volume workload. And I'll make this up for sake of conversation, a 365, you know, mail, web, et cetera, et cetera. All those are perfectly great things where you're going for lowest cost, not requiring cycle completion. In between, I think what's been left behind and is absolutely 100% going to emerge is the provider cloud. They've got the stickiness to the customer. Customers built out in enterprise cloud, they need to burst out with guaranteed of experience and capacity. And we see a ton of these, or a number of these very, very large cloud companies that are focusing on the value workload that are guaranteeing cycle completion time. Wasn't a lot of the players that you mentioned, by the way, but what I'm trying to call out to this conversation is there is a continuum. A number of new markets are going to emerge and that's what I'm predicting for five years is that it's not going to be the two polls that you mentioned, it's going to be a whole lot of points along that. Lots of private cloud. And don't forget that local, regional, national, international, not everything is going to the mega data center. Lots of stuff is going to be closer to the edge on-prem, on-site, simply because of transport cost, that might be one, or simply because I must have a guaranteed quality. So if I'm running a financial trading algorithm, I'm not running that with an indeterminate cycle time. I'm not running that where speed of light is going to be an issue. I need that right next to me. Yeah, but the New York Stock Exchange has a cloud and they've got their data in one place and they've got kind of industry specific environments that might be useful. Absolutely, moving data is hard to do. I think there's plenty of room for it. Moving data is easy to do. What are you talking about? Come on. Well, let's talk about it. You saw the speed of light issue, right? Yeah, exactly. Let's talk about data, big data, fast data, little data. The role of data with analytics and automation is critical, so internet of everything also includes data center devices and networking information. It's been around for a while. What's your take on that? Any evolution or changing in the industry with big data? How it's impacting some of the innovations that you're working on? Yeah, absolutely. The type of big data analytics that we need to be able to build out are frankly, let's just talk about products for a second. We need to build a guaranteed capacity. I need to crunch a petabyte of data, right? We know that's going to be a certain amount of compute, a certain amount of storage, a certain amount of network, a certain amount of interaction. Building out both turnkey solutions in a pod-like piece, because generally we talk about pods just being data center capacity. The portfolio that I'm trying to change and I'm trying to push through the system is a notion of solution capacity, which requires all of these pieces. You asked about big data, trying to build, here's what it requires. And we formed a number of partnerships, whether it's with SAP and their HANA solution, as with some others, that here's a set unit of capacity that can be guaranteed that it's going to work. Now, you mentioned internet of things. Realize there, that's a totally different information model. Talking about big data is like tossing bugs words out there. It's data, it's ultimate data. And that data does come with a specific information model, a specific processing model, that is going to be an extremely interesting emergence in the industry of what do I do locally? How do I aggregate it? How many levels of aggregation? Where's the final consumer? Where is it near the compute power? Is it co-located? Do we build compute power into some of these, let's say IOT gateways? Do we push them even in lower? Is that part of the information hierarchy? Still early, I mean that's open book at this point. That's probably going to be more headroom, right? Exactly, and I think it's great headroom for the industry. Dave, as you look around the industry, I'm curious, what gets you excited? Is there, maybe you can't speak of a specific startup, but what area is there a lot of room for more innovation and a lot of activity that you'll see, maybe over the next year or two? Sure, I think that one major piece is that a lot of the open source pieces are early. And I'm not just talking about open daylight, I am talking about open stack, which those conferences are now getting enormous and there's a ton of attention around it. These pieces aren't complete yet. We know that when we get the ability to have a full stack, let's say built in open source, built out of a variety of consortiums that can be pulled together to tie together all these pieces that I keep reiterating, SDN, NFV, and orchestration, we can get to the point where there is potentially an open source cloud platform. So what gets me- From a composition standpoint and an operational? Correct. What makes this interesting is that our customers have choice, not only that, there's ability for immediate influence on what's being deployed. What gets me excited is the pace of change. Things right now are moving as fast as they were in the mid 90s, all the way up to the early 2000s through the internet boom. It's just a new internet boom with new services and solutions that are out there. And I'm totally pumped up. I think the opportunities are outstanding and also frankly the market's changed quite a bit. Our customers are in transition, the suppliers are in transition, the technology's in transition. Man, it doesn't get more exciting than that. So go back on history, let's do a history lesson, go on memory lane. What in your lifetime do you remember? Can you point to that this kind of resembles? Is it the compilation of client server and PC revolution? Is it the networking era, TCPIP, all the above? Can you kind of compare and contrast what environment that we're in now and what it looked like? What could have looked like? I know it's different, I mean, I think it's different, but what is it like? How do you look at it? I mean, compared to other inflection points? Well, I think all the comparisons you made were correct. This will be a successful movement. This is, and we are looking at something as revolutionary as we can compare it to Ethernet. We're talking about the new protocols that are around the network. Could be TCPIP. We're talking about the new protocols that change the network. But in addition, what's fascinating is that this is the first revolution around ops that we've really seen SNMP. This is not comparable to SNMP, which is a simple network management protocol. This is the first revolution targeted at the IT professional, targeted at the service delivery piece. And that's absolutely fascinating. But I am cashing the check that this is a successful movement because we're going all in, I'm all in. I think it's a very good bet. I mean, obviously it's amazing. The Internet of Things, although is a pretty obvious bet too, Internet of Things versus, say, Industrial Internet, which is a GE kind of positioning, where they're talking about airplanes being run, using data from turbines and equipment, billion dollars of ROI on a 1% improvement. Some people say, some people, this is a quote I heard, that when people die in a hospital because of bad management of data, that would be the equivalent of a 747 dropping out of the sky every two weeks, but no one's taking notice. So these are the kinds of, like, shifts on the consumer end. To what you take on the Internet of Things versus, say, the Industrial Internet or Industrial Cloud or... Well, I'm going to take it a little bit differently to begin the conversation. So what's interesting is the last time, well, sorry, the most recent time, and we're still living through this, that an additional 6, 7, 10 billion devices were added to the Internet was mobile telephony. And you know what? The Internet's still standing. So now we're talking about adding another 10 billion over the next several years over the same timeframe. This is a fascinating time that the Internet's going to stand up and man, man, 10 billion new devices at a time. So let me get to your exact point. What's interesting about what you're mentioning with respect to healthcare and some of these segmented industries is that a lot of those devices are not attached. A lot of those devices, if they are attached, are not associated, for example, is the EKG associated with me in that room. We've got very interesting security issues in the manufacturing plant. You talked about the Internet of Things with the turbines being run on the jet. We are talking about automated and sensor-based automation of factories, manufacturing, mining, et cetera. Where in the mines, they actually want to have driverless trucks inside all run by sensors. It's intoxicating. If you want to go geek on it, you say, oh my God, we're instrumenting our entire global planet. At the end of the day, it's completely instrumented. A revolution in ops, I love that. I think your identity thing is huge. I mean, devices connect. Do you think about the identity role of the person? It's just mind blowing. So much more work to do. Thanks for coming on theCUBE. We really appreciate it. We are here live at the Open Networking Summit. I mean, we just had an amazing conversation about developers, the future of networking, the future of the world. This is Silicon Angles theCUBE. We'll be right back with our next guest after this short break. Thanks.