 We're back to VMworld Live 2010 Silicon Angles continuous coverage of VMworld in theCUBE with a very special guest, Ed Booneon, who's the CTO of the Business Union within Cisco. But he's known in the industry circles as somewhat a legend, as an early co-founder of VMware. So welcome to the broadcast. So I have a first couple of questions like, how's it feel to look at VMware growing up like this? I mean, just get a little bit of weirdness to you. Is it? No, it's not weird, but it's amazing. It certainly is getting bigger and bigger every day and every year. I remember the first VMworld we thought 14, 1500 people was a lot. I heard the whisper in this morning was 17,000. So I don't know if it's true or not, but it certainly feels like a packed house. Virtualization is certainly center stage right now in the cloud. Would you have predicted it would be this relevant, this fast from when VMware started or yes, or did it take the shape that you had thought it might? Well, I mean, you know, the, if you go back to the early, early, early principles of VMware, it was all about isolation and encapsulation and it was all about mobility. And if you go back to the late nineties, right? Mobility meant you can take a VM and you could basically copy it as a file from a computer to another computer. These were desktop operational models at the time. So the elements that are still foundational to the cloud today or we're present over a decade ago. What's changed in the meantime as we've gone from desktops to data center and service scale virtualization. And then the other thing obviously is the networks and have evolved to the point that you can now think of computing in the cloud, which is not something that was being thought about in those terms a decade ago. Great. I'd like to introduce my co-host for this session is Stu Miniman, senior analyst at Wikibon who recently is now an analyst, was at EMC and knows Ed. And you guys gonna talk a little bit. So Stu, I'm gonna pass it over to you for, I'll jump in if I have any questions, but go ahead. Thanks John, yeah. And I went my 10 years at EMC, I met with Ed back in the Diane Greendaze part of the acquisition and reconnected with Ed when he was the CTO of the Nuova group with Cisco and obviously worked a lot with Ed on the fiber channel over ethernet and his UCS has been launched out. So great to see Ed. Thanks for joining us. Good to see you as well Stu. Great, so I went to how we choose session this morning talking about networking and one of the comments how we made is when you look at VMware and server virtualization a couple of years ago you wouldn't have had networking people showing up here and now there's a good contingency of networking people showing up here. Obviously Cisco's well represented. I'm wondering if maybe you could talk for a minute about kind of the Cisco and VMware innovation that we see going on right now. Yeah, so I think this is an absolutely correct statement and back in the early days virtualization was really a single disciplinary solution, right? It was sold and operated by server administrators very often departmentally and without really having to worry about network policies without having to worry about storage policies for that matter and without having to worry about issues of compliance and security and audibility and reproducibility because it was used as a tool rather than as a core infrastructure and the transition is now vSphere is really viewed as a core infrastructure of the data center, it's part of, it is the fabric of the data center at the software layer and the interaction between the distributed system that is vSphere and the underlying network becomes that much more important. So the thing that we introduced two years ago that sort of puts Cisco on the map into the VMworld community was obviously the Nexus 1000V which basically is a pretty simple observation which is virtualization and it was good if it's not thought through carefully can introduce operational gaps or blind spots. You get efficiencies as far as one set of deliverables of the IT organization is concerned but you create blind spots or issues to other groups and in this case the networking team and it's really important when you think about the problem from a networking angle and I've been now at Cisco for between the new over days and the Cisco days working in this industry for the past five years is it's really about the end to end visibility across a common set of tools, a common set of methodologies and so the Nexus 1000V is the bridge because it's the tool that allows you to have a network classic network methodology extend the topology of the network into the virtual infrastructure while at the same time fundamentally not changing anything as far as server administration is concerned as far as VMware Virtual Center is concerned so yes, lots of people with a networking background at VMworld, lots of people from Cisco not necessarily all networking people because we're much more than a networking company but we have a north of 200 employees at this show one way or another and certainly because virtualization is the center of the thinking and the brainstorming from all of the organizations that constitute a data center and obviously not to mention the cloud. Okay so and obviously when you talk about these cross-disciplinary solutions obviously Cisco's UCS is playing right into that marketplace. I wonder where specifically do you see UCS in the virtualization marketplace, what's next? I mean we talk about kind of scale up, architectures and cloud architectures. How is UCS going to tackle I guess the two biggest challenges I hear from the community are management and security so any comments on the direction there? Yeah and by the way management and security are related, the more complex the management the more difficult it is to reason about the security of it. The more manual the provisioning of the infrastructure the more difficult it is to reason about its security. So UCS is really at the end of the day was a relatively in retrospect straightforward observation which is that there were significant advantages of going from a box-centric view of thinking about x86 computing in the data center which is the classic systems management approach to things where basically at the end of the day a server company sell chassis with a few blades in them and then the customers have to operate them by having these different points of configuration that are independently managed for the same entity in the same box. And the observation we made is let's just flip this over in its head and rather than selling a chassis is to sell a fabric in which you can plug chassis into it and with a single point of management. And one other thing that I think resonates the most by customers who use UCS is the fact that you can really now have this dual view of your infrastructure. On one hand UCS as a system is something that gets laid out on a few racks, right? It can scale way beyond the size of a single rack so think about four racks worth of gear with the networking size. And that's sort of the physical view in all what we assume and understand about the physical lifecycle management of things. At the same time UCS is also from an operational perspective simply a web service, a single web server that exports a RESTful style XML API as a single point of provisioning and control for that entire kit. And it's really the reduction in the number of points of management that is enabled by the raising of the abstractions that we export from the system itself that is making UCS sort of a differentiated offering in the marketplace. And at the end of the day the pain points that customers have is it still takes too long to cable and provision physical gear and once you have it cable and wired up it still takes too long to configure physically before you can even get to the vSphere layer. And then also you end up having an infrastructure where it's very difficult to adjust your network policies and your server policies because those tend to be managed separately. In UCS you have the ability to atomically repurpose capacity and really turn the data center into a set of fungible resources at the hardware level. Right, so absolutely, I like how there's some argument between especially some of the guys that really are helping customers deploy this solution because if you look traditionally, the VAR has done the job of taking all the pieces together, integrating it together, building it for scalability performance, whatever we need. And of course UCS is built and especially if you start talking about a vBlock we put all the configuration together and some would say it's too fixed but rather than having all of the knobs we have something that can be deployed a lot simpler. One of the things when you say this is the configuration and this is what you get but when you talk about roadmap, one of the questions that I've gotten from the Wikibon community and the practitioners is what other technologies are there going to be some flexibility down the road? For example, will there be an AMD option on UCS? Okay, so I think we've got to separate a couple because there are different sort of questions laid into the set of questions. Sure, I know. First is UCS itself is a product with a very strong roadmap including elements that are public and that we're talking about and including elements that are obviously under NDA at this point. But fundamentally the way we think about the way we are thinking about UCS from a roadmap perspective it's really about further automating and further increasing the scale of the systems under management but while keeping sort of a certain set of architectural principles as we've laid them as we introduce UCS because those have been extremely effective and those architectural principles is the unified fabric. The center point around Ethernet as the underlying link technology. Transition from 10 into the next level of standardized link speeds starting with 40 gig ethernet which is actually not far away and then building bigger and bigger systems with richer fabrics inside. Now UCS itself is an architecture but it's not a prescriptive deployment model that you can have different size and shapes and ratios of UCS. There's lots of choice with respect to the blades and the compute factor, form factors. There's a lot of choice with respect to the adapter form factors. And then there's a lot of choices and options with respect to the topologies and the ratios and the over subscriptions levels that you have in your management model. The V block is something different. The V block is really designed to take a lot of the guesswork out of the deployment in an enterprise environment of a system that has a UCS compute backbone and as far on the computing side is concerned. Combined with enterprise grade storage which is either the Vmax line or the Clarion line with all of the associated enterprise storage capabilities that are come with the product. And the V block is a more prescriptive lay down of the combination of those two pillars. With a very clear value proposition is that it takes out a lot of the guesswork as you're deploying virtualization at scale. And I think if you think about where the state of the art is today it's relatively straightforward to deploy vSphere in a small scale. And I think VARs will say that and end user customers obviously will say that as well. Where things become more challenging is as you scale your deployment, as you make them more multi-tenant, as you want to turn them into a service as opposed to something that you consume internally whether it is your private cloud or your online cloud. And that's some of the challenges that the V block is designed to address. And certainly with respect to how VARs think of this they very much welcome the value and the reference design and the fact that the three companies stand behind the solution itself. The issue of support and the complexity of support as we deploy these high-end sophisticated solutions is always something that's front and center in the customer's mind. It's not something that the VAR can directly address. And it actually if anything allows the VAR to focus on some of the much more value added elements of the conversation including how you can take advantage of that new scalable architecture. Okay, great. Excellent stuff, Ed. Yeah, I guess, you know, trying to think of any kind of follow-up. But I think about in the networking spaces one of the challenges we see is that when you look at the next generation speed, you know, 40 and 100 gig ethernet there's only a small number of customers that are really requiring that bandwidth and driving that, you know, if you look at, you know, the Facebooks and the Googles need that bandwidth, need to drive it further. And, you know, you wonder how many customers, you know, don't need that technology and you're kind of pulled along behind it. So just curious your thoughts and kind of macro trends in the networking space. You know, what would be the biggest challenge you see over the next three to five years? Well, so all of us who started in the industry after Bill Gates have learned never to say that you would never need more than X, right? Because unfortunately it gets held against you decades later. So yeah, of course, one day we'll need 40 gig and 100 gig. I think that these new levels of speeds are associated with the fact that the data center is going to be truly managed as a system. When VMware talks about the software mainframe, the distributed mainframe, it is premised on the existence of a high performance compute fabric. The analogy I would use is with respect to multicore, right? You could have made a sort of a few years ago the same argument, how would you possibly need more than one or two cores per CPU? Well, the reality is now you've got eight and 10 cores per CPU as a standard building block of your offering and that there are ways to take advantage of them, namely virtualization. So I'm very confident that the industry will find ways to take advantage of those new capabilities and we are actually already seeing some demand for both four times 10 gig solutions because in some cases the bandwidth for certain types of capabilities is actually the limiting element and also for lower latency solution where you actually take advantage of the 40 gig all the way down to servers. It's still part of the roadmap but the one thing that the industry has done is certainly it's preparing to do this time more smoothly than in the previous transition is to enable a faster adoption of the new technologies. We went from one to 10. That has proven to be complex and complicated because of lots of five issues and transmission issues. The transition from 10 to 40 is going to be a lot smoother. Okay, great. I think, Ed, I appreciate you coming on theCUBE and talking about, are there any sessions coming up at VMworld that we should be especially paying attention to? Yeah, actually thanks for bringing it up Stu. So I will be speaking tomorrow at 11 o'clock right after the keynote from VMware. This is the super session and the big, the theme of the super session is the delivery of scalable cloud services. So really moving from, if you think about what we did in 2008, it was about operational gaps. It was about network awareness for virtualization and the Nexus 1000V. In 2009, it was about UCS and how you can take UCS to solve compute problems at a data center scale. This year it's about scalable cloud services and how you can start thinking about the network and the services that are part of the network on a cloud scale for cloud deployments. So I hope to see some of you at the show. Okay, well thanks so much, Ed, for coming on. We're going to take a break. We'll be right back with the Rob Roy, CEO of Switch Networks, and Charlotte Yarkany, who's with EMC running the public cloud of EMC to talk about what that means. And we'll be right back. Facilicon angles, continuous coverage of VMworld 2010 Live here in theCUBE. Ed, thanks so much, Stu. Thank you, John. Appreciate it.