 Live from San Francisco, celebrating 10 years of high-tech coverage, it's theCUBE. Covering VMworld 2019, brought to you by VMware and its ecosystem partners. Welcome back here on theCUBE. We continue our coverage. We're live in San Francisco, Moscone North, day two. Wrapping up day two of our three days of coverage here at VMworld 2019, Dave Vellante, John Walls. Glad to have you with us here on theCUBE. And we're now joined by Kevin Schatzkamer, who's a Vice President of Service Provider, Strategy and Solutions at Dell EMC. Kevin, good to see you this afternoon. Thank you, you as well. And Ihab Tarasi, who is the SVP and Chief Technical Officer at Dell Technologies. And Ihab, good to see you. Thanks for taking the time to be with us. All right, a couple of telco guys. And we've had a lot of telco on and talking about it in terms of progress that you made. This was an area that you got into with a major commitment, some probably three years ago. Kind of benchmark it for me then, from where you were there on day one to where you are now today and the progress you've made and maybe the services that you're about to provide. Yeah, sure. So I think if we look over the last three years, our opportunity that we defined early on in the telecommunications space was that the virtualization and software defined everything was leaving the data center and we would see the software defined architecture extend all the way from radio through the core network through the cloud over a period of time. And it started with technologies like network function virtualization. So if we flashback three years ago, where our entire strategy was built on a premise that relationships with the network equipment providers like Nokia and Ericsson were our primary path to market and our primary opportunity. I think what we've realized is we've emerged in this space to a greater detail is that our expertise and experience in building IT networks and building cloud has led to the first wave of conversations in the telecommunications industry directly, not through the network equipment providers, but that carriers want to engage directly with Dell EMC for the lessons learned and how to deploy IT architectures. And now as we extend towards the edge that they want to engage directly with Dell technologies in terms of how we build cloud architectures. We've had a number of big announcements over the last several years. We've announced partnerships and engagements with NTT. We've announced partnerships and engagements with China Unicom. Just in the last three months, we've announced partnerships with Orange around network edge out of France. And then most recently with AT&T on the automation of edge infrastructure related to their airship project. So I think from a benchmark perspective, it's just been a continued growth opportunity for us and recognition that the more we engage and the more we contribute as a productive member of what is a very complex and changing and transforming industry, the more success and relationships that we'll build and the more it will translate into opportunity to sell to. When you think about, you have the modernization NFV, for example, as a former technologist inside a large telco, what were some of the challenges? Is it, it's taken a long time obviously. When you talk to some of the telcos they say, well, you know, it affects our infrastructure, we still get this application mass. I mean, maybe you could add some color and describe for our audience why it's been so challenging. Yeah, I think that's an excellent question. Going back to my days at telco and data centers, even SDN and the software defined tools were just beginning to show up. So the biggest challenge is where you were basically having to work with predefined operating system, predefined hardware, the hardware was not exposed for programmability, the ability to take advantage of it and then you had to integrate multiple players of technology in a way where it took significant time to not only for software development, but for product development and user experience. Since then, many of those walls have come down and some of them have come down very hard. When you look at what we're doing at Dell here and we led for with open networking, not only do you have the choice of operating system, we're also pushing hard on new open operating systems for networking like Sonic with Microsoft and Broadcom. And then we're taking industry leading steps to expose the silicon chips themselves for programmability. These are all the components that are critical when you talk about 5G, for example. You really have to have those capabilities. I also would say that the software evolution have made it to infrastructure. The DevOps and the modern applications we talk about here is also available for infrastructure, which means you really can develop a capability in weeks instead of years and months. Five people can do in amazing progress. All of this was not possible before. So we talked to Shikhar about this in the earlier segment, challenges in the telco business. I mean on the one hand you got these quasi monopolies in some cases, real monopolies that maybe just chug along and do pretty well but at the same time you've got the cost per bit dramatically coming down. You've got the data growth doing this. You got over the top providers taking advantage of those networks. And so new infrastructure allows them to be more agile but there's a workforce component to that and there's a skill set and it's got to transform. I wonder if you could maybe talk about that a little bit Kevin. Yeah, I think that's exactly right. I think when we work within this industry it's not just a technology conversation. It's the ability to consume and operationalize technology and I think that comes down to a number of different things. It comes down to the processes that exist and it comes down to the skill sets that exist to be able to build these new processes around. And I think if we flashback several years ago the model of how we built networks was that the team that operated it needed to understand networking. Now if you look at the team that needs to operate it they need to understand networking. They need to understand compute. They need to understand virtualization. They need to understand APIs. They need to be able to script and program. They need to understand some level of data science so that they can close a loop in the operational models eventually with AI and machine learning technologies. So I think that the teams that are getting built look very different than the single soul capabilities that they've had in the past. These are smaller teams. They're more agile teams that can develop and have their own more unique processes in each part of the network. And even if we think about the organizational structures we've always built vertical organizations. When I had an appliance that was an EPC I had an operations team that was focused on an EPC and I even broke that into an S gateway, a P gateway, an MME, et cetera. If we look at the world now, that S gateway, P gateway, MME consists of a server, consists of the networking that connects that server, it consists of a virtualization layer, it consists of a stack of a software application and all of those need to be automated and orchestrated and programmed to work as an EPC does. So I think that the skill sets have just really expanded in terms of what's expected from our operations. And this is really important because the processes used to be pretty well known and hardened. So the infrastructure could be hardened and now it's every month the market changes, right? What kind of challenges does that bring to the telco provider but also to the infrastructure provider? Yeah, actually I have a really good way to describe what I think is happening. We heard it from a lot of our customers and not just telcos but enterprises. I would say the last five to 10 years everybody's been dealing with hybrid cloud. The move to cloud was the big challenge. While this remains a key challenge, a new challenge showed up which is how to succeed in this new modern software development model. Are you able to move at that speed which means you have full stack engineers? Can you develop the app beginning to end? It's not an IT model anymore. Also, you no longer have an operations team. You really have to have SREs who are able with software and also the customer service change to a software driven. So we're starting to hit from a lot of our customer. That's the next journey they need help with. If you think of infrastructure, those challenges are even bigger and this is where it's important to lean on technology partners who can help you with that. And you hit on 5G a little bit ago, you have in your initial statement and I think we've kind of touched on the impact that it can have in terms of you understanding they're going through a transformative time, right? I mean telcos are with new capabilities and new opportunities and this whole edge is going to be crazy, right? So you've got to, I wouldn't say some learning to do, but you've got to get up to speed on what their new fundamentals are going to be, right? Yeah, I think that's true. I think we've understood their fundamentals because it's the same transition that the IT world's gone through and to a large degree the cloud world has gone through. I think that the challenge we've been working to break through collectively as an industry is the paralysis at the rate of adoption of new technologies. Because there's so much change so quickly because we talk about virtualization and then we're talking about Kubernetes, we're talking about cloud native, we're talking about bare metal services, we continue to talk about micro services architectures. We see this progression of technology that's happening so fast in various segments of the industry. I think that the telecommunications industry has been somewhat paralyzed in terms of where do they jump in and which do they adopt and how fast do they migrate between them? And which of them can be capable of being hardened to be telco grade and fit into their requirements that they have for being able to offer regulated services? Paralyzed because it's just too fast. It's too fast for a big, I mean it's a big decision to make but things are evolving too quickly. That's right, it's evolving too quickly and they also sometimes have a concern that they get stuck on a dead end path, right? Because things change so quickly, do I jump here, then here, then here, then here, then here or do I follow a logical path? And what we tend to find when we work with the telecommunications industry is that yes, Dell technologies can define a strategy. Certainly VMware and Dell EMC can define our individual strategies. Our operators can define their strategies but there's just not one strategy for this industry. Reality is that when you get together with an ecosystem of partners and you work at a particular telecommunications company that is a strategy and you start from scratch when you go to the next, right? Because their ability to consume technology is just so different. The end game may be the same across the board but the path to get there will look different. So every customer's different, you get that but clearly some patterns must be emerging. So my question is where do you start? I mean if you're sitting down with a, what are you seeing in terms of common starting points and advice you'd give to your customers? I think the two main areas everybody's starting with. First of all, the physical infrastructure, compute, storage, networking is moving to X86 model of some sort, which means many, many parts of their infrastructure today that is not based on X86 needs to transition. So we're seeing big RFPs, significant discussions of how you take compute and this new programmable networking and put it everywhere, like in thousands of locations. So infrastructure wise that is a known specific thing to be solved at early stages and given the capabilities we've delivered for enterprises we have a lot of tools and capabilities to give them. And the second one is that a lot of people are approaching this as a network issue. In reality it's a cloud decision, not a network. You heard Shaker talk about it. So the tools, capabilities you need to build a cloud is completely different. This cloud may not be generic cloud, it needs to support the different specific platforms and therefore they want cloud and they need it to support this specific capability. So that's the two, a year ago nobody even could articulate that was the challenge they were facing. But I would say that's what we are today. I would add to that, that as we kind of think about the infrastructure and then that cloud decision that there's abstractions that exist between those. At the infrastructure layer there is the need to have an automation system that has the ability to support multiple different cloud platforms that sit on top of it and that's work that we're doing on the Dell EMC side and then secondary to that at the cloud layer it's the ability to support a multi virtualization environment. Virtual machines do exist and will continue to exist. Kubernetes and cloud native containerized applications do exist and will continue to exist and the challenge becomes how do I orchestrate an environment that allows those to exist simultaneously and be layered on top of a common building block of infrastructure and I think that's really the power that the broader Dell technologies has is that we have all of these entities and capabilities in-house. How long does this take a telco to transform? Is this decade, is it more, is it can, I mean obviously certain parts can happen faster but when you sit down with customers and they put together their plans, I mean what's their time horizon? So I would argue that we define the first NFE standards in 2012 and if we look globally and even within the vast majority of the industry and carriers we're somewhere in the 10 to 15% range. Yeah, the two compelling areas that are gonna maybe be a forcing function for making some of those decisions are the economics on moving to X86 are very compelling. It's 10 times the speed to deploy and it's a massive order of magnitude and cost. Therefore, it's not something that you could wait on as you continue to build capacity. So that is forcing the infrastructure decision. The second forcing function is that what 5G is starting to look like is not network and wireless independent from enterprise solutions. You really have to collapse to a single infrastructure to offer services and wireless embedded and that's another forcing function in terms of enterprises is starting to ask for those capabilities. You know, you mentioned X86 a couple of times and when you think about the telco cloud, it's generically what we're talking about here, in the commercial cloud, not the telco is not commercial but the mainstream cloud, you're getting a lot of offload, hardware offload, alternative processing, ARM, GPUs, FPGAs, even, you know, custom A6 coming back. Are you seeing the same thing in the telco cloud? For sure. I think if you look at what we've done over the last several years, we've seen this dramatic shift and almost a pendulum swing away from A6 and proprietary hardware towards everything on X86. And I think what we've learned over the last several years at X86 is a platform that has its value but it's just not for every workload. So we've seen things like network slicing and control user plane separation and technologies that are first moving user plane, very high IO applications back onto SmartNICs and FPGAs and eventually onto merchant silicon with programmable silicon and the network switches. But I think that even if you look at what's happening in public cloud with things like GPU virtualization, they're still largely virtualized in the time domain which means that they're used by a particular application for a period of time and then the next application schedules it and the next application schedules it. That doesn't work for network workloads. So I think that what we're finding as we go to this telco cloud model, especially with offload and the virtualization of acceleration technologies, is that it's an entire set of problems that just aren't solved in public cloud yet. Yeah, I would say based on experience, the vast majority of network workloads have to be X86. I definitely think ARM cores and GPU offloads will play your role at some point in the future, but that's not the heavy duty that you need to offload those functions. Because most of these network applications were written for custom ASIC that's very high performance that has high throughput, security built in, ability to build services directly into the silicon. So that kind of transition, over time you'll see a lot of distributed applications hidden in container formats all the way at the edge, but that transition to that kind of distributed model from what we are today is probably not possible. And I would argue you'll always have the mix of high performance, high throughput. I mean, think about it, if you're trying to activate 20,000 IoT devices instantly, you really need a high core density, you know, X86 chip with significant memory. You really worry about the data plane and how much data you can put through, et cetera. We didn't even hit IoT, did we? Yeah, that's... Another day, another conversation. Hey, thanks for the time. We certainly appreciate it. Been a good show, I hope for you all too, right? For sure. Good energy, good vibes and good business. Thanks for the time, we appreciate it. Thank you guys. Thank you very much for your time. Thank you. You're watching theCUBE live coverage here at VMworld 2019 in San Francisco. Thank you.