 Okay, welcome back to VMworld Live in San Francisco. VMworld 2013, I'm John Furrier, the founder of SiliconANGLE. It's theCUBE, our flagship program. We go out to the events, instruct the student from the noise. I'm joined here, my co-host this segment is Stu Miniman, wikibon.org, and we always have great guests, usually start-ups, CEOs, but this is a special start-up, and Sony, G and Donnie with Nsimi Networks, which is a very special start-up because it's funded by Cisco. That's right. Welcome back to theCUBE. You're a CUBE alumni. You were on two years ago at VMworld when Cisco. Welcome back. Thank you. So tell us, obviously Cisco, and they're a big player. They know a little bit about networking. And so when SDN hit the scene, there was always a lot of naysayers out there saying, oh, Cisco's missing the boat, but Cisco, I taught the folks in there are excited. They know they have leverage. They have a lot of big accounts. And so, is this part of Cisco to be nimble? Is it a different company, separate company, wholly owned subsidiary? Can you display in the company? Sure. Nsimi Networks is a majority on Cisco subsidiary. One of the key areas that we are focusing on is to drive the vision of application-centric infrastructure which embraces some of the constructs of SDN and looks beyond those to see the needs of our customers, data center needs, and also how we're going to drive a more holistic approach, trying applications, as well as networking teams and the security teams closely aligned through a common policy framework. So give us some stats. Employees, kind of the size of the company, what's, did it just get like a pivotal, it's the big, a lot of people involved, or a small team, what's the general size? So Cisco has infused over $100 million within NCMA over the last 17 months of the company being in existence, and we have over 260 employees, primarily development teams, that we have accumulated here in Silicon Valley from startups, as well as some of the fast-moving internet-class companies. So it's a super-charged startup? Yes, it is. Kind of like what VMware did with Pivotal. Something similar, but here it is, yeah, we are structured like a majority on subsidiary. John, I think one of the differences, of course, Pivotal is moved outside to be able to partner with a lot of people out there. I think one of the things that Cisco's done really well with their spin-ins before is it allows them to stay separate, create that innovation without having to deal with the traditional business product lines out there, and typically it's pulled back into the power company. So you think it's, from an innovation standpoint, it's all about speed? It's about going to be about speed while leveraging the ability to bring together the ecosystem, as well as the ability to access such a wide range of customers. That's just co-cators to whether it's the internet-class data center customers, commercial customers, enterprises, and service providers. So what is the big opportunity that you guys see? Obviously, to put this team together, obviously you want to hit a certain market, and how much creativity do you have? You have freedom? I mean, you have to go back in and get signed off? We have the latitude to innovate, where applicable leverage existing technologies, so we're not recreating the wheel, and the innovations will be done across all segments, which are starting at A6 hardware and software. So we do have tremendous latitude to innovate, and bring the talents that have system-level views, and productize and make the solutions available to the customers globally. So, Sony, I think keynote, I know you didn't get a chance to see the keynote this morning, but Pat Gelsinger came out and talked about how virtual switching ports are now greater in number than physical switching ports, and he said, his logic is, congratulations, you virtualization admins, you're actually the largest gathering of network administrators that we've ever seen. So we understand that the boundary between physical and virtual switching is something that's been a little bit of a challenge for many years. Cisco, of course, had the 1000V, but with SDM, things are changing anymore. So what do you see as kind of the state of physical and virtual switching, and how does the kind of the network virtualization and SDM fit into that? So it's, let me try to take your question and first split it out, right? I mean, for starters, Cisco and VM would have a very good strategic relationship in server virtualization and desktop virtualization. That's where the customers are demanding that we continue to work very closely on elements like the unified computing system, which has really taken an industry-leading approach and has become a market leader in such a short timeframe. If you look at the networking world, we have a different view vis-a-vis the NSX world, and this is coming straight from vendors like Gartner that have stated that only 15% of all computers, of all servers, out there are indeed virtualized, and 42% of all customers are multi-vendor from a hypervisor point of view, and over 60% of the workloads will, in one way or another, be cloud-based by 2016. We, as a networking company, have to not just think about physical networking, but also have to factor in virtual networking, and while keeping in mind that a lot of the growth with big data applications and the levels of traffic that are coming into the network are bare metal in nature. So the application-centric architecture, which we are putting forth as where we want to take the industry and take a leadership position in, has to factor in the fact that physical and virtual need to coexist, and that the boundaries of a software-defined network are not just around providing the flexibility of software, but you also have to keep in mind the scale, security, and multi-tenancy attributes that come in from a hardware view as well. So it's about innovating at all levels, ASIC hardware and software, where we feel that the SDN models of today fall short are in areas of mapping physical and virtual, lack of visibility in the physical in real time, so when I'm trying to diagnose or troubleshoot a problem, the onus is on the shoulders of the customer, because the way NSX is trying to bring these architectures together is through loosely-coupled interfaces. How do you deal with Microsoft as a hypervisor? How do you deal with security in this loosely-coupled world? And how do you keep check of all the revisions of all the plugins and the shims that you need to all keep together in this software-only-defined world? Let alone the ability to deal with bare metal, which the scale needs that come about. How do you measure end-to-end latency? How do you look at it from a system point of view and not a box-by-box or a hypervisor by hypervisor point of view? And last but not least, how do I tie it together from an orchestration perspective that is agnostic whether you're using OpenStack, which is something that we will continue to put a lot of energy on through our open APIs, but also how do you tie in customers' existing tools from a lifecycle management perspective and expertise in bringing the worlds of applications, networks, and security all through a common policy framework? So while we are taking an approach of embracing some of the innovations, we are also looking beyond it to see where the problems lie for customers. Ultimately, it's about the applications and the agility with which they want to move and they want the infrastructure and the software to respond to that. That's where the problems need to be solved. You know, we hear VMware talks a lot about their software-defined data center and networking is the next piece of it. Very much a virtualization focused and my understanding NSX can work across multiple hypervisors that can tie to cloud. Not exactly sure how it ties to physical environments as opposed to, NCME has really the application-centric focus. You talked about how UCS fits in. What does the NSX announcement mean for things like the 1000V and the networking relationship between Cisco and VMware? So as I said, in terms of the relationship between the two companies and server virtualization and desktop virtualization, we will continue to be very strong partners and respond to our customers, our mutual customers' needs and tie very closely from a solution point of view. In the area of network virtualization, we have a different strategy. Our strategy is around the application-centric infrastructure which thinks about not just the innovations at the virtual network layer, but also the innovations in the physical networking layer and providing a common pane of glass. Adopting the models that we already put forth in fabric computing with UCS and bringing those same notions and empowering the networking teams, but also tying the security teams and the app teams into that common policy management framework. We also believe that recreating the problem which is what the attempt being made by network virtualization using NSX is not taking away the problem. You're just moving the problem from one layer and adding another layer analogous to ATM LAN emulation. I've been in the networking industry now for 20 plus years. And one of the fundamental reasons why ATM LAN emulation did not take off was because of the complexity as well as lack of scale. So simplicity is a core part of the application-centric infrastructure. And last but not least, having the ability to have real-time policy-based frameworks that can tie all three different groups together is ultimately what will drive the agility in the infrastructure to cater to the varying needs of the application administrator. So I get asked a lot as I spend a lot of time looking at converged infrastructure. Are solutions like VCEV Block and Flexpod, are those really UCS-driven and therefore unaffected by some of the visionary differences between Cisco and VMware? Or will this have some impact on those solutions? I find the VCEV as well as the Flexpod types of converged infrastructure hugely benefiting with the application-centric infrastructure because when you envision the networking elements of these embracing the ACI constructs, you truly experience the model of a converged management model beyond just the physical infrastructure but also tying in the virtual as well as tying in the application areas, which is ultimately what customers are embracing the Flexpod and the VCEs for. So I guess from a visionary standpoint, one of the big question I have is, who really manages this environment? VMware puts forth the vision that the virtualization administrator, they'll pull it all into vCenter, they'll manage that, some of it will go via hybrid cloud. Is it a network administrator in your view of the point or what does the operations team look like in the future? So in the model of the application-centric infrastructure, we clearly are empowering the network buying center through automation. At the same time, we are also ensuring that the security operations team members and the application administrators as well as the cloud administrators are empowered to provide and to have the ability to use the same framework so that you're not manually going across these traditional siloed organizations, which is the case today and bringing the network services in from a separate angle. But because it's a common policy framework that ties in the network, the network services like firewall, server load balancing and all the other network services, as well as the security teams and application teams, it's a common policy framework that now automates while empowering the network buying center is also providing the baseline for these different security and app teams to use the same framework, driving consistency and a common framework across these different models. This also empowers the cloud administrator that has the ultimate job of rolling out applications using a common infrastructure as a service model so that you can drive velocity and you can drive agility at the speed at which your apps can be deployed across this infrastructure through automation. Sony, I got to ask you about the market. Obviously when I get to the VMware ecosystem question because I think that's a big conversation underlay, overlay, are they tool for any job or is it one monolithic software? But also OpenStack, obviously the momentum with Hyper-V and Microsoft showing some traction, you see obviously VMware, they got OpenStack out there, it's kind of like this halfway house for people who want to get some private cloud and public cloud and hybrid cloud and then you got Amazon. So tie all that together. Why is OpenStack so successful? SDN had a nice little role in that and a lot of innovations happening in these cloud models with networking. What's your take on OpenStack? And then compare it with VMware's ecosystem play. So let's look at OpenStack. One of the biggest reasons that customers are telling us that they are latching on to the OpenStack model is that it is giving them the flexibility to leverage an underlying infrastructure including hypervisors that are multi-vendor in nature. And it also gives them the ability to bring in their bare metal applications and it provides them with the flexibility to tie in different elements coming at every layer of the IT stack. That is the beauty about the OpenStack model that leverages again the open source community. No one vendor owns it. No one vendor owns it. And it's giving the buying center the latitude to move at the speed at which their respective IT organizations are moving. It gives them tremendous flexibility. At the same time, it's extremely important to note that when you compare and contrast it to the overlay models of the network virtualization schemes that are being put out there, you would need to make compromises around what level of functionality the overlay network can provide and where you would need to go for the ecosystem or the other pieces. The flaw that I see in this ecosystem play is that you have to, on one hand, be very open. But at the same time, you cannot say I am open to allow for these loosely coupled items to interoperate with each other because who at the end of the day is going to manage the version control, okay? I've got a shim or an adapter or a plugin on vendor A's implementation. I've got another controller that is managing a piece of that. When I change the revision on my controller, who's making sure that the appropriate interoperability testing and all these other loosely coupled elements are held together? And how do I deal with Microsoft Hyper-V in that environment? How do I also deal with security in that type of an environment? Or if the customer buys another vendor's solution, how does that play into this overlay model? These are important questions. So is the book written here, is it still an open book relative to what the ecosystem strategy is for VMware? I believe that it has to reconcile how would you take something like this into production. It's one element to try and put these pieces together through technology building blocks, but the verdict is still out on if you were to take the solution into the market and make it production worthy, how will this fit into a real world production environment? And how does this fit into the lifecycle of a cloud management tool set in terms of version control, security and all the other fundamental items that are very important for the enterprise customers as well as the cloud providers. So Cisco has been very successful in building networks. Obviously Web 1.0, Cisco was the company. In Web 2.0 they've struggled a bit, John Chambers admitted that, but now with the cloud is an opportunity and now the applications are in charge. That's right. And so that is different from to the network. So what is Cisco and what are you guys doing particularly to make that policy hardened or that functionality enabled without a lot of rework or missing capabilities? How do you look at that? And how would you explain to an IT guy, hey, you know, App Tsunami's here, you got some cloud, here's the network. And you're considering OpenStack, you got Red Hat, all this stuff out there, the network, the Cisco or something else, but older networks, what do they got to do to modernize? So if in order for you to think about how do you move into the cloud based environments, keeping in mind your legacy applications, the network has to offer some very baseline functions in order to accommodate the cloud models and the evolving cloud models. It has to have a certain level of SLA capabilities to deliver quality of service, security, reliability as the core elements of the building blocks. The same time we are driving towards a more open southbound API and a northbound API construct that allow you to plug in different elements of heterogeneous components that exist in a data center in terms of infrastructure pieces. And through the open APIs that are northbound, have the ability to accommodate whether it's OpenStack that we need to support, including extensions that you would make to OpenStack, as well as have the opportunity to tie in existing lifecycle management tools that they have today for their networks and empower the cloud administrator to drive a certain level of automation in the networking stack. Third key element is tying in network services. Your security firewall devices, your server load balancing devices, 95% of which are physical, but an architecture that also ties in the virtual environments. How far, how baked out is that today? All that functionality, on a scale of one to 10, 10 being fully baked out. That level of functionality today is a manual intervention. So my load balancing group has to manually assign the policies, my firewall group has to do the same and so does my networking teams and so does my compute teams, right? These are all connecting the dots manually across these different siloed organizations. Expansion, labor issues. Higher OPEX, right? So you have five or six different groups that are trying to manage this rollout of applications in a manual manner. The key is to provide an infrastructure that is independent of the underlying app whether that app is running on Hypervisor A or Hypervisor B, whether it's running on bare metal and providing the basic services in a very agile way to the underlying applications that are expecting those services to come out of this agile infrastructure. So our philosophy in the application-centric environment is really to bring, to bridge the gap or to normalize things, whether it's a virtual environment coming based off of Microsoft, whether it's based off of VMware, whether it's based off of KVM or whether it's a bare metal application. Should have to take away the argument of am I routing, am I switching? It shouldn't matter to your application. Security should be an embedded element of that underlying infrastructure. So who's your competition as a startup? You have to have some benchmarks, but you're kind of a startup but you've got Cisco behind you but you're really out there in the trenches. Who do you see as your competitors? So we do see the traditional overlay models as one competitive alternative but we believe that they fall short of really being in the true competitive landscape because they've yet to prove themselves in the real world. They're competitors on paper. Competitors on paper, right now, you're absolutely right. And then we also see the notion of providing to our customers the same levels of SLAs that they have gotten for their traditional environments in this new cloud era. In this new cloud model, we want to be in a position to offer those same services to this heterogeneous pool of applications, those same level of services at scale. So goals for the next year, I'll see what's your objective for the next year? I'll see you've got to grow the team. Execute. Build the technology. What are some of those execution milestones that your team has? Well, our goal really is, as we have stated when we launched application-centric infrastructure at Cisco Live, is in the second half of this calendar year is to roll out the solution base and to provide our customers with a set of solutions that they can put into their production environments and cater to a diverse set of customer needs. And that's where all our energy is right now is in focusing our energies on executing on what we said we would do in the middle of this year and to execute in the second half of this calendar year. Well, very exciting. I love to see Cisco doing the spin-ins. That's a move that they've done in the past. We'll see how fast this could move with that. And if spin-ins are going to be viable, I'll see Cisco has a lot of muscle and install base. Congratulations and great to see you in a startup mode. Congratulations. This is theCUBE. I'm John Furrier with Stu Miniman. We'll be right back with our next guest after this short break. Thank you.