 This is Stu Miniman with wikibond.org. Here with SiliconANGLE TVs, live continuous coverage, VMworld 2012, digging in deep to the software-defined data center, transformations in the networking world. And joining me today is a CUBE alum, Dom Delfino from Cisco. Dom, welcome back. Hey Stu, thanks for having me again, great to be here. Great, so last time we caught up at EMC World and now we're here at VMworld to two very close partners of Cisco. So, you know, what's new since last time we met at EMC World? Well, lots of things new, Stu. I think, you know, a constant buzz about the industry, you know, our relationship and our partnership with VMworld and EMC has never been stronger. And a lot of changes going on in the industry today. So, you know, we talked a little bit last time about kind of what was emerging and I think we've got some more to talk about today. Yeah, so Dom, you know, let's try to get the hard-hating questions here. So, you deal with Cisco's field a lot, right? A lot of them report up to you, especially when we talk about storage and virtualization. So, you know, there's been some changes in the marketplace. You know, Cisco supplies networking for, you know, VMware, you've got the 1000V, some updates we want to talk about there. And, you know, EMC is a strong partner, NetApp's a very strong partner, but a change in the conversation is, you know, VMware bought NICERA. So, you know, what do you tell your field and what is your field, you know, getting asked questions from the end users? Sure, and Stu, great question. I think our field understands very well, you know, what's going on with the evolution of SDN. And I say evolution, not revolution, right? Because, you know, SDN, although a new term, building software-defined networks is not necessarily something new. If we go back to the TDM days with ATM and frame relay, these were sort of software-defined networks, right? Service providers need to support multi-tenancy in those environments. Fast forward to IP, MPLS, VPLS, IPsec, VPNs. These are all forms of SDN networks. So, this evolution has been going on for a very, very long time. We've got a new term to coin it with now. And we're going to look at moving forward. How does this play with the advent of virtualization in cloud as SDN involves further? And I think what we're going to find out is that Cisco has a very, very comprehensive three-pronged approach to that. You know, so, I'm not sure if I completely agree that SDN is not revolutionary. What I would say is that the technology, if we focus on, you know, networking is, you know, you just listed off a ton of acronyms that non-networking people probably aren't going to understand. You know, things like VXLAN and NVGRE and LISP are not exactly things that your average VCP, you know, are going to know. The CCIE that is also a VCP. So, for those that don't know that, that's the CCIE is the network-trained guy. The VCP is the virtualization expert. Those kind of skill sets, when those are put together, it is a powerful combination. And there are some transformational aspects about who owns, operates the network and how operationally things happen. And Cisco's been making some moves in, you know, how operational changing. So, at Cisco Live, there was the one-PK APIs that are coming out. Probably can't comment on the spin in NCME, but, you know, a lot of moves, like Cisco, you know, is not sitting still in this space. No, certainly not. And I think you make a good point, Stu, is who owns the network? And I think that what we are seeing is operationally more and more transformation going into these environments, right? Certainly server administrators, virtualization administrators, want to be able to provision elements of the network as they provision the VM, the host, the guest operating system and the applications that sit above that. But what's most important here is that we can't break the operational environment, either, right? So, if the goal is I'm just going to build an overlay network, which is one aspect of SDN, how do I troubleshoot what's happening in the virtual overlay network from what's actually happening on the physical underlying network? Do I troubleshoot those separately? Or does there need to be some form of integration there? We certainly believe that there needs to be some form of integration there, and the two need to be tied together and linked together, right? Yeah, no, no, great point. I mean, that has been the discussion since, you know, VMware came in is that the physical versus the virtual, how do things break? Storage has been maturing that environment and networking, there's a lot of focus on making these changes here. So, let's talk a little bit about some of that virtual networking. So, you know, people in networking are very familiar with, you know, the 1000V from Cisco and environments are starting to become multi-protocol. So, Wikibon recently did a study of virtualization environments and we found 56 of the customers that responded actually have multiple hypervisors in their environment today. So, when we talk about physical networking versus virtual networking, it's not just VMware. Where's Cisco's stance on this and can you give us the update on the latest product stuff? Yeah, so certainly, you know, there's an industry. We all need to play together collectively in the industry and, you know, as much as there's market leaders in every segment of the industry, nobody owns 100% of the market share. I think, secondarily, the piece that we kind of skip over in all this discussion is not everything is virtualized either. So, we still do have to deal with and may for some very, very long time bare metal-based operating systems and applications as well. So, how do you take an approach where, A, we need to have something that's standards-based because proprietary is not going to win in this market. So, we all need to work together collectively and Cisco's going to take an approach spearheaded with the Nexus 1000V which has been in the market some time. Today that's in VMware's offering. It's coming down the pipe in Microsoft's offering as well as well as other hypervisors there because, you know, they do have some play in the market as well. Secondarily, there needs to be a physical attribute as this as well. So, if I encapsulate a packet from the hypervisor and I send it across the network, what does the network say? Pretty much nothing, right? There's only so much we can do with that. So, if we believe that we don't need A6 in Silicon to do things like a class of service and security and all the other aspects and attributes that networks do today, then that's an approach you can take. It's probably just not going to scale very well over the floor. So, you will continue to see the Nexus 1000V which is kind of an SDN product historically, right? It's at least control plane centralization in a virtualized environment. As we add VXLAN to the Nexus 1000V, you will see us get the ability to build an overlay network in conjunction with that centralized control plane environment as well. Not a lot different from what you see with VMware doing with NYSERA, okay? So, customers will have options, right? And what's important to what you said earlier was does the server guy own it or does the network guy own it? Does the server guy wants an interface to it, right? But he doesn't want to necessarily own the troubleshooting, the diagnostics, the ongoing management and monitoring of the entire network. Just wants an interface that he or she is used to using to do some abstraction from the host to the network and do some configuration and make sure that his job or her job is easy and they can get the things done that they need to do from a networking perspective to deploy those VMs and those applications. So, if we look at the software defined whatever, software is important, kind of that abstraction or virtualization is a piece of it pooling. And one of the things you kind of look at as a litmus test is how tight does that software have to be to the underlying hardware? So, some of the knock on Cisco has said, merchant silicon is taking over a large piece of the marketplace. Cisco even uses some merchant silicon in some of the high performance environments. Why does Cisco still have some of its own chipsets and in a software defined world, what Cisco's feeling on software versus hardware? So, I think certainly it's actually three pieces, right? Software hardware and silicon, right? I'll separate hardware and silicon there. And I think that merchant silicon has a play in the market. We do use merchant silicon chipsets in some of our platforms. But what you tend to see from the merchant silicon providers is everything that's out there is been out there already, right? Innovation doesn't necessarily always come in a mass form. It takes somebody to start it and then turn it over to the environment and make it standard space. MPLS was Cisco tag switching, right? You see a lot of features that we've innovated that were proprietary and then we opened them up and make standards. So, what you'll see from us is take an approach that encompasses software, hardware, and silicon and ASICs, right? And those ASICs will be for the most part our own and we will use merchant silicon when and where relevant and necessary, okay? Secondarily, we think that SDN overlay networks is one aspect of it. Network programmability is another aspect of it. And thirdly, the control plane. Whether that comes in the form of a controller or it's embedded in the network device, that control plane is very important as well. And we know part of this value of Cisco, right, is iOS and NXOS and that feature-rich control plane that we have that has enabled customers to do everything they've done thus far. And as we move that into this next generation of SDN, you'll see all three of those things play a very, very relevant piece in this. And we don't think just taking a one-pronged approach is going to give customers a comprehensive solution across the bar. So we think we're going to have to make all three of those very, very relevant and in place systematically with all of the vendors and the hypervisors and the other components in the network as well. Yeah, so some of the critique on the networking industry is that it's been kind of boring for the last 15 years. I mean, Cisco, I know you can't say it, but you dominate the marketplace in many places and networking is plumbing. So it's one of those things that needs to work. I want to keep going, but it's not the sexiest thing. It's a frothy time in networking right now. It's exciting times. What's your experience been when you're talking to customers? Is networking raising up the level of what customers are caring about? What's the priorities? And what's exciting you out in the marketplace today? It's an interesting perception, right? Because I think the pendulum tends to swing very far. And I think maybe last time at EMC World, we used the analogy of customers were going to buy these storage virtualization appliances and just put a bunch of J-Bots behind them and the whole storage world was going to go, the whole storage market was just going to go to this commodity based approach. Yeah, it didn't quite happen. Didn't quite happen, did it? So are we going to replace the 25 years of Cisco innovation with a startup with 50 people who comes out with this nifty whiz bang little software strategy? Those things will have a role and elements of that will evolve, but our customers are concerned that they need to provide an enterprise class or a cloud class capable production ready network, which is extremely comprehensive. We have a lot of implications in the network. Scale, right? The number of IP addresses, particularly with virtualization is exploding, right? Which means table sizes on switches and routers need to be there to handle that. Those are certain things that you can't do in software, they have to be done in silicon. So if that silicon's not talking to the software and I can't do that pervasively throughout the network, I'm going to leave myself some opportunities to have some major issues. And customers who take a one pronged approach may get it to a certain point and realize, oh boy, I'm in trouble now because I can't scale beyond this or I'm starting to have vulnerabilities or security concerns or scalability or redundancy issues in the network. And I may need to kind of regress and go back and rethink this as well. So definitely some of the big themes I'm hearing at the show here is really scale out architectures and distributed systems. I mean, nobody has more deployments than Cisco does on the networking. There's a bit of a schism when you look at deployments. There's the cloud and the service providers and then there's the enterprise. You talked about overlay networks are not the only solution. The enterprise can't just say let's start from scratch or let's just put in this old environment. One of the things customers have, lots of catalysts out there. They've been deploying the Nexus. Where are we in the adoption of some of these new architectures, things like Ethernet fabrics, fabric path and the like? Yeah, I think we're certainly seeing mechanisms to scale L2 much greater, right? And we know one of the implications of virtualization is we need a lot of L2 adjacencies which is not the always most efficient way to build a network. We see the ability to desire transporting VMs across L3 boundaries. Now, L2 over L3 is something, again, we've been doing for a long time, not specific to virtualization, but we now need to adapt that as we move forward. We see other things coming down the pipe. You see, obviously, latency. Not only latency from port to port on a switch, but latency of the entire data center fabric become more of a concern as I get VMs moving around and applications moving around my environment. We talked last time about the adoption of FCOE and how quickly that was growing. We see things like 40 gig FCOE coming to the market at the end of this year via Cisco's, one of Cisco's new products in the pipeline as well. So I think we see a lot of different growth areas in conjunction with SDN that are gonna kind of emerge together as the next generation data center fabric. Programmability, we think, is one of the key things because as you look at the overall orchestration, provisioning, self-service of this cloud type environment, whether it's private cloud or public cloud or a hybrid of the two, is a very important aspect when you see Cisco's 1PK, our open networking environment, our APIs, our developer kit in order to allow the orchestration components, the provisioning components to interface with the network in a much more granular and robust fashion than we have been able to historically. Okay, so we're getting close to a time on here. Last topic I want to touch on is OpenStack. So Cisco participated pretty heavily. What's Lutucker talking about when it comes to OpenStack? What's the positioning on kind of those things? You know, I think certainly one of the drivers of 1PK and the open networking environment is to give customers the ability to choose what mechanism they use to orchestrate and provision their environment. Obviously, quantum is the network component of OpenStack and we need to allow them to interface with our infrastructure, expose our capabilities to them in a programmatic fashion so that they can deliver the most robust cloud service to their customers, whether they be internal or external customers as well. So Lutucker and his team are working very diligently on that. Lutucker actually has a team of developers specifically working to OpenStack as well. And OpenStack will be the only strategy. You know, you'll see us interface with cloud stack, you'll see us interface with vCloud director and dynamic ops as well as the Cisco intelligent automation for cloud of a BMC. You know, customers are going to choose what's right for their environment and we need to make sure that we're able to give them a solution that fits in with that, not forces them to choose a path. Great. Any last highlights you want to talk about that things going on in the show that, you know, interesting solutions or new technologies that people should be learning about from the Cisco portfolio? You know, I think obviously, Lou, you know, we've, I'm sorry, Stu, not Lou, you know, I think the SDN conversation is a great conversation for us to have. We've been having a lot of those conversations throughout the course of this week. And I want to continue them because I think customers are becoming more and more educated about, you know, let's take all the hoopla out of this, right? And let's really get down and get granular and how is this going to work? How is it going to deploy these concerns from a server layer, from a hypervised layer, from an application and operating system layer through the network into the storage layer as well. And we need to make sure that all of these work together consistently as a system. So while some people may think the networking is plumbing, right, we think it's a key component of that system and as critical as the other pieces. We think all the components are equally as critical. So let's continue the conversation. Let's keep digging in here. And let's make sure that we work together to get this going. Okay, excellent. Well, Dom Delfino from Cisco, thank you so much for sharing us Cisco's viewpoint. Cisco is a, you know, a major sponsor here at the conference, reaffirming, you know, the long commitment to partnering with VMware and the whole ecosystem that goes beyond it. So this is Stu Miniman with wikibond.org, continuing live continuous coverage here from VMworld. We'll be right back after this quick break. Stay here.