 Live from San Francisco, California, it's the Cube at VMworld 2014 brought to you by VMWare, Cisco, EMC, HP, and Nutanix. Now here are your hosts, John Furrier and Stu Miniman. Hi, this is Stu Miniman with wikibond.org, SiliconANGLE TVs, live continuous coverage from VMworld 2014 here in San Francisco. Our fifth year at the show, so much going on, cloud, software-defined, data center, converged infrastructure. I'm going to dig into a second segment now with Cube alumni, Howie Jou, who's a senior director of networking, I've said senior director of engineering with cloud networking in services with Cisco. Howie, thanks so much for coming back on the Cube. Thank you for inviting me here. So, Howie, you've been at many of these shows in the past. You're a VMWare alumni, work in some of the startup world, and now with Cisco, who is a large partner of VMWare and some places where there's disagreement as to whether future things go. What's your take on the show so far? Interesting you asked me this question. This is my 11th VMworld. I was at VMWare for almost a decade. I'll tell you a story first. Okay. Back in 2007, and Diane Green, of course, at the time CEO of VMWare, did a keynote speech. And I guess who still remember who was the other person who did a keynote together with her? It was John Chambers. It was John Chambers. Since I was running the networking team at VMWare, everyone was asking me one question. Why John Chambers as the CEO of a networking company here at a server virtualization conference? Seven years later, last year, VMWare the 2013, VMWare announced NSX. And I just joined Cisco. Everyone asked me, from VMWare asked me this question, or outside of VMWare asked me this question. Why Cisco's logo is not on it? Now, one year later, people asked me a more interesting question. You guess what it is? But I think, you know, that's sort of the story aside. I think if you look at networking overall in the last five, seven years, it's a huge shift, right? I think the awareness of the problem definitely surfaced dramatically. Now, in terms of how people are really adopting the technology, is that doing things significant differently? I think it's still lagging behind. If you look at a server memory virtualization, it's definitely out there, right? Storage virtualization, it's almost there. Networking virtualization or link between virtual and the physical, it's still, there is quite a bit difference. It stands away. So, Halle, help us unpack that a little bit. So, VMWare has always had some networking. I mean, you ran the networking group inside of VMWare, the virtual networking switch, the distributed switch, put out an API so that Cisco could create the Nexus 1000. Then, of course, the shot heard around the world two years ago, put this whole SDN discussion on the map with the NYSERA acquisition. And really, it spawned a whole revolution in the industry. We'll have Martin in on Wednesday to give the update from the VMWare side. How far have we come and how far do we need to go for kind of this discussion of the physical versus the virtual networking? I think there is a long distance away. I mean, if you look at it from the VMWare perspective, 70% of the workloads are virtualized. So, if I have a networking virtualization solution for that 70% of workload, almost there, right? 70% there. But if you look at it from a networking admins perspective, I have 10 ports and one of the ports connecting to a vSphere, which runs 20 VMs. So, 20 out of the 29 workloads, that's 70%, are virtualized. But from a networking perspective, that's one port out of the 10 ports that are connecting to vSphere environment. So, even if I can go all the way adopting VMWare NS6 technology, let's say for a second, I cannot fire any network admins. That's the reality today. So, I don't save any OPEX by adopting NS6. Of course, NS6 is not free of charge. So, I'm not saving any CAPEX. So, what is the real saving here? Now, not to say future, it cannot get there. But that's today's state of the art. Okay. So, you're saying you're looking to fire network engineers then? Well, it's not about firing networking engineers. The entire industry movement, IT industry movement, is always about money shift. Where's the money shift from one place to another place? All I'm saying is the money shift has not happened yet. That's all I'm saying. Yeah. Absolutely. One of the biggest discussions is when I did server virtualization, I could have an immediate CAPEX savings because I could consolidate there and the story on whatever the networking SDN network virtualization is not as clear cut from a CAPEX standpoint. Okay. So, you know, VMware has talked a lot about open source in their keynote this morning. Pat Gelsinger got up, talked about open compute, talked about open stack, talked about Docker. You know, I've spent a lot of time over the last year talking with Cisco about some of the open sources initiatives. You have things like OpFlex involved in open networking. You know, what's your take on how much is open source influencing networking? What's your position on kind of the networking and open source? Yeah, you're opening kind of worms here. Open. I mean, people want to open because they don't like vendor locking. That's the sort of rationale behind the people are so enthusiastic about, you know, going for the open solution. Yet, at the same time, open doesn't mean maturity, right? It doesn't imply maturity there. You know, VMware announced open stack distro this morning. It's interesting to me that this is probably, I don't know, 17th or 20th open stack distribution on the planet. Yet, when I count the number of real solid open stack install base, it's actually, it's possible less than 20. I don't know, you know, there may be more distros than the number of, you know, really large install base out there. This means that open stack is still in its, you know, early stage. So in the meanwhile, I think it makes perfect sense for VMware sort of to grab the market to sort of, you know, coming, I have open stack, you know, I manage VMware environment really well. So I think that's a very good move. Docker is a very, another interesting thing. There are so many bloggers talking about blog, you know, dockers, whether that's, you know, disrupting VMware technology. My personal take is VMware has to say what they say at a marketing wise, right? You know, their marketing language seem to be very clear, better together. Now, the way I read them better together is if you ask a Microsoft guide 10 years ago, when it comes to Windows versus a VM, they probably want to say better together too. But they are better together as a very different from VMware's view of better together 10 years ago. I think today VMware had this, you know, has the same sort of the issue where fundamentally VMware's number one object is a virtual machine container is just a sort of a together partner at a marketing wise. So it will be interesting to see. I personally, I feel like Docker can disrupt the virtualization space tremendously. Yeah, and how we have an agreement. I attended DockerCon. It was a fascinating show. I didn't hear Yahoo or Google or Facebook talking about, you know, VMware when they were using the containers. I mean, you know, it's a very different, you know, way to look at your applications. Of course, they're doing applications that span, you know, will be on a single server many times hundreds or thousands of servers, as opposed to traditional VM is putting, you know, an operating system with an application inside of it. So very different from a networking standpoint, you know, how does kind of the containerization and Docker impact networking? I mean, we spent years talking about how virtualization changed us from north south to east west. How do things like Docker impact the network? I think the in terms of the networking primitives, Docker has a lot of the, you know, equivalent of the Linux bridge, OBS in it already. So that's not a problem. The problem is really the provision, the lifecycle management. I think in terms of the Docker, you know, both of us are in violent agreement that it has the, you know, disruptive potential because it's, you know, potentially 10x more efficient, 10x, you know, more faster. However, at the same time, the lifecycle management, the provisioning, you know, all the sort of the things you need to do on the day-to-day thing doesn't quite exist. It applies to CPU memory management. It applies to storage management. It applies to networking management. So I would say that the core technology is, you know, right there. But, you know, there is, but that's only 10% of the product. 90% of the product is to deal with the boring, you know, the day-to-day management. That thing is still, you know, someone needs to take that, you know, and take that challenge and then do it. Okay. So I wonder if we can, you know, discuss a little bit cloud kind of the public versus, you know, private hybrid cloud discussion. You know, how does, how does your role kind of fit into the whole cloud conversation? Well, cloud, I think VMware also understands this issue and the Cisco sees the same issue in that enterprise customers wanted a public cloud. They wanted to move things into some public cloud. At the same time, that's CIO view. But if you look at the CSO view, the security guys view, no, I don't have enough security. I don't have enough compliance. I don't have enough visibility. I want to slow down the move. That's really the struggle between CIO and the CSO today. And that problem is not solved today. VMware wanted to solve this problem one way and Cisco has, you know, a slightly different approach. But that problem is yet to be solved. So that's sort of the one angle of the issue. The other angle of the issue is, you know, we were discussing the Docker thing, Docker and the container thing. Docker and the container is built, you know, in some ways, is very much suitable for this inter-cloud kind of things. Because, you know, at Francisco's perspective, we wanted to not just treat the VM movement migration between on-prem versus off-prem, treating VM as the first class citizen, but also the container and the Docker kind of the objects as the object we wanted to migrate. Okay. So, yeah, I mean, one of the questions you talk about kind of that inter-cloud world, and we believe that, you know, there will be, you know, quite a few clouds out there. It's not one. And of course, VMware looks at everything needs to be a VM. You know, a VMware VM. However, the NSX group, from a networking standpoint, they do support, you know, more than just, you know, VMware. You know, what's Cisco's position on this? You know, where do you see the cloud world? I know you've got your inter-cloud solution very much tightly partnered with service providers. You know, how do you see this evolving? Yeah, I think this is where the future battle is. You know, I actually run the engineering team of the inter-cloud of fabric within Cisco. And the way I look at it is, again, it's more about life cycle management. It's more, it's, you know, the core technology. Cisco has that technology. Cisco needs to move up the stack to make sure that people can just not just have the layer-to-layer three connectivity, but also, you know, entire life cycle management. And for NSX or VMware or another technology company, that's where I think that where a lot of the gap is today. Okay. Essentially meaning that the technology cannot be a technology. It needs to be a turnkey solution that I can use right away. Yeah, you know, really good point. So, you know, one thing when just looking at the overall market, SDN has been much, very much a discussion of various tools. If I look at NFV network function virtualization, it feels like they have end-to-end solutions a little bit easier and therefore they've been going to market a little bit faster out there. You know, what's your take on that? You know, how do we get SDN really moving forward or, you know, is NFV still going to lead for a while? Yeah, I think SDN fundamentally is a component of a solution. That's why, you know, it's sort of the wavelength is a little bit different. Yeah, you get both Cisco and VMware don't really say we're SDN. You've got application center. Well, no one on the planet says SDN the same way anyway, but that's a different story. But NFV is sort of, there is a very solid concrete definition. Okay, here is my networking functions before and now it's after. So, it's sort of leveraging the virtualization platform, leveraging that maybe SDN layer underneath that. Here is a very solid solution, you know, before and after. SDN, you know, what is before? What is after? You know, it's a little bit, you know, more fluffy than the NFV side. Absolutely. So, just like we discussed, you know, what is the OPEC saving? What is the CAPAC saving? Can you define that? Is there a matrix to define that with NFV? It's a lot easier to define with SDN. You know, it's less, less, less, you know, easier to do that. Yeah. And one of the, one of the things I look at how we is, you know, if you're, you know, really large web scale company or if you're a service provider, it's much easier to be able to build out that new data center or adopt these technologies. How far down, you know, is the, you know, do the mid-range companies need to even worry about this whole SDN discussion? Or if you reach a certain point, do you just say, I'm going to go to some hosted or service services out there because, you know, I don't really need to need to go through that transition. I mean, heck, just the transition from one gig to 10 gig has taken us over a dozen years since the standard. So, you know, where is that demarcation line and who needs to pay attention to this? That's a very interesting question. You know, Pat Geltinger mentioned in the keynote speech this morning that, you know, the one example in the keynote speech was this Golden Gate Bridge. I look at a virtualization as a bridging technology between the old guard IT sort of technology versus the new guard stuff. But that's a transitional technology. That's a bridging technology. But that doesn't mean it's the, you know, the new territory technologies. To answer your question sort of more specifically, I think the, I look at the transitional technology has fundamental limitation. You still have to deal with a virtual machine as the first class citizen. The scale limitation, you know, it's still there. For the web guys, web scale guys, Google or the sort of the Google offering service for the SMB, they don't need to think about all of this. So in some ways you asked me a question, you know, Cisco versus sort of the VMware, whether that's perceived or that's just the endless like to make this up. The real thing is both VMware and Cisco has a challenge to deal with, which is the public cloud is where the future, you know, might be and the workloads are moving there and how to be relevant in that, in that new era. I think it's a challenge for both Cisco and the VMware potentially takes both company to work together, you know, to come have a real relevant solution so that this hybrid cloud thing is there. Or, you know, some other startups or, you know, Cisco can reinvent itself and there are all sorts of possibilities. Yeah. So, you know, how we, you know, you've been on both sides of the table between the VMware and the Cisco piece. We had Sony on last year right after the keynote and she talked about, you know, the tight partnerships, the many things that Cisco and VMware are doing together and where the vision differs, you know, the last year, what have you seen, you know, where are the, you know, tight engagements and where are the areas where, you know, just to give us a final word on that. So, I cannot speak for VMware, but the way I look at it from outside of these days is, you know, it's really how to treat the physical network. From VMware's perspective, physical network is not an important commodity. But if you read NSX manual, it says I need existing network. But what's the definition of existing network? Its definition is this highly scalable leaf spying, you know, low latency. But that network doesn't quite exist for majority of the enterprise customers. So, it's sort of like Cisco's take is I give you a solution that deals with both physical and the virtual. In my personal sort of opinion, that makes sense because people are, there are two transitions here. One is dealing with this networking virtualization. The other thing is this three-tier networking to this leaf spying architecture. Cisco is leveraging both motion. One is I'm migrating from the three-tier network to the leaf spying and then ACR gives you that. And at the same time, you want SDM functionality, I give you that. From VMware's perspective, it seems like, you know, the physical network transition, it's all, you know, you can take it for granted. I don't need to do anything. But I think people need to do a lot of work on the physical side in order to, for the entire solution, holistic solution to make sense. All right, well, Howie, really appreciate you coming, helping us unpack all the discussions going on with the networking space, physical and virtual, and how we're trying to solve those problems to make it easier for customers to be able to, you know, take advantage of their infrastructure, deliver on the applications of the future. We'll be right back with lots more covered from VMworld after this quick break. Thanks a lot.