 The CUBE's live coverage is made possible by funding from Dell Technologies, creating technologies that drive human progress. Hi everybody, welcome back to the FEDA in Barcelona. My name is Dave Vellante, I'm here with my co-host, Dave Nicholson, Lisa Martin is in the house. John Furrier is pounding the news from our Palo Alto studio. We are super excited to be talking about cloud at the edge, what that means. Charles Sy is here, he's the Senior Director of Product Management at Dell Technologies and Peter Federolf is the Chief Technology Officer at ACG Business Analytics, a firm that goes deep into the TCO and the telco space among other things. Gents, welcome to the CUBE, thanks for coming on. Thank you. Yeah, good to be here. So I've been in search all week of the elusive next wave of monetization for the telcos. We know they make great money on connectivity. They're really good at that. But they're all talking about how they can't let this happen again, meaning we can't let the over the top vendors, yet again, basically steal our cookies. So we're going to not mess it up this time, we're going to win in the monetization. Charles, where are those monetization opportunities? Obviously at the edge, the telco cloud at the edge, what is that all about and where's the money? Well, Dave, I think from a Dell's perspective, what we want to be able to enable operators is a solution that enable them to rule out services much quicker, right? We know there's a lot of innovation around IoT, Mac, and so on and so forth, but if they continue to rely on traditional technology and way of operations, it's going to take them years to enable new services. So what Dell is doing is now creating the entire vertical stack from the hardware through CAS and automation that enable them not only to push out services very quickly, but operating them using cloud principle. So when you say the entire vertical stack, it's the integrated hardware components with like, for example, a red hat on top, or a wind river. That's correct. And then open API so the developers can create workloads. I presume data, data companies, we just had a data conversation because that was part of the original stack. That's correct. So through an open ecosystem, you can actually sort of recreate that value, correct? That's correct. So one thing Dell is doing is we are offering an infrastructure plot where we are taking over the overhead of certifying every release coming from the red hat or the wind river of the world, right? We want telcos to spend their resources on what is going to generate them revenue, not the overhead of creating this cloud stack. Dave, I remember we went through this in the enterprise and you had companies like IBM or the AS400 and the mainframe saying, it's easier to manage, which it was, but it was still assumed by the open systems trend. Yeah, yeah. And I think that's an important thing to probe on is this idea of what exactly does it mean to be cloud at the edge in the telecom space? Because it's a much used term when we talk about cloud and edge in sort of generalized IT, but what specifically does it mean? So when we talk about telco cloud, first of all, it's kind of different from what you're thinking about public cloud today and there's a couple of differences. If you look at a big hyperscale or public cloud today, they tend to be centralized in huge data centers. Okay, telco cloud, there are big data centers, but then there's also regional data centers, there are edge data centers, which are your typical access central offices that have turned data centers and then now even cell sites are becoming many data centers. So it's distributed. I mean, like you could have, like even in a country, like say Germany, you'd have 30,000 cell sites, each one of them being a data center. So it's a very different model. Now the other thing, I want to go back to the question of monetization, okay? So how do you do monetization? The only way to do that is to be able to offer new services like Charles said, how do you offer new services? You have to have an open ecosystem that's going to be very, very flexible. And if we look at where telcos are coming from today, they tend to be very inflexible because they're all kind of single vendor solutions. And even as we've moved to virtualization, if you look at packet core, for instance, a lot of them are these vertical stacks of say Anoki or Ericsson or Huawei, where you can't really put any other vendors or any other solutions into that. So basically the idea is this kind of horizontal architecture, right? Where now across not just my central data centers, but across my edge data centers, which would be traditionally my access COs, as well as my cell sites, I have an open environment. And we're kind of starting with packet core, obviously, and UPFs being distributed, but now open ran or virtual ran, where I can have CUs and DUs and I can see they could be at the cell site, they could be in edge data centers, but then moving forward, we're going to have like Mech, which are new kinds of services, could be remote cars, it could be gaming, it could be the metaverse, and these are going to be a multi-vendor environment. So one of the things you need to do is you need to have this cloud layer. And that's what Charles was talking about with the infrastructure blocks, is helping the service providers do that, but they still own their infrastructure. Yeah, so it's still not clear to me how the service providers win that game, but we can maybe come back to that because I want to dig into TCO a little bit, because I have a lot of friends at Dell, I don't have a lot of friends at HPE, I've always been critical when they take an x86 server, put a name on it that implies edge, and they throw it over the fence, and it's not going to work. Okay, we're now seeing, we were just at the Dell booth yesterday, you did the booth crawl, which was awesome. Purpose-built servers for this environment. That's right. So there's two factors here that I want to explore in TCO. One is how those next-gen servers compare to the previous gen, especially in terms of power consumption, but other factors, and then how these sort of open-ran, open ecosystem stacks compare to proprietary stacks. Peter, can you help us understand those? Yeah, sure, and Charles can come in on this as well, but I mean, there's a couple of areas. One is just moving the next generation. So, especially on the Intel side, moving from Ice Lake to Sapphire Rapids is a big deal, especially when it comes to the DU, and the DU, you know, with the radios, right, there's the radio unit, the RU, and then there's the DU, the distributed unit, and the CU. The DU is really like part of the radio, but it's virtualized. When we moved from Ice Lake to Sapphire Rapids, which is third generation Intel to fourth generation Intel, we're literally almost doubling the performance in the DU, and that's really important because it means like almost half the number of servers, and we're talking like 30, 40, 50,000 servers in some cases, so, you know, being able to divide that by two, that's really big, right, in terms of not only the cost, but all the TCO and the optics. Now, another area that's really important, when I was talking, moving from these vertical silos to the horizontal, the issue with the vertical silos is you can't place any other workloads into those silos, so it's kind of inefficient, right, whereas when we have a horizontal architecture, now you can place workloads wherever you want, which basically also means less servers, but also more flexibility, more service agility, and then, you know, I think Charles can comment more specifically on the X or 8000 some things Del's doing because it's really exciting relative to what's happening there. So, you know, when we start looking at putting compute at the edge, right, we recognize the first thing we have to do is understand the environment we are going into, so we spend with a lot of time with Calcos, going to the south side, going to the edge data center, looking at operation, how do the engineer today deal with maintenance replacement at those locations? Then based on understanding the operation constraints at those sites, we create innovation and take a traditional server remodel it to make sure that we minimize the disruption to the operations, right? Just because we are helping them going from appliances to open compute, we do not want to disrupt what is having a very efficient operation on the remote sites. So, we created a lot of new ideas and develop on general compute where we believe we can save a lot of headache and disruptions and still provide the same level of availability, resiliency and redundancy on an open compute platform. So, when we talk about open, we don't mean generic. Fair, fair, fair standard. Open is more from the software workload perspective, right? A Dell server can run any type of workload that customer intends. But it's engineered for this environment. And so, what are some of the environmental issues that are dealt with in the telecom space that are different than the average data center? The most basic one is in most of the traditional cell power, they are deployed within cabinets instead of racks. So, there are depth constraints that you just have no access to the rear of the chassis. So, that means on the server, everything you need to access need to be in the front. Nothing should be in the back. Then you need to consider how labor union come into play. There's a lot of constraint on who can go to a cell tower and touch power. Who can go there and touch compute. So, we minimize all that disruption through a modular design and make it very efficient. So, when we took a look at the XR8000, literally right here sitting on the desk, took it apart, don't panic, just pulled out some sleds and things. One of the interesting demonstrations was how it compared to the size of a shoe. Now, apparently you hired someone at Dell specifically because they wear a size 14 shoe. So, it was even more dramatic. But when you see it, and I would suggest that viewers go back and take a look at that segment specifically on the hardware, you could see exactly what you just referenced. This idea that everything is accessible from the front. So, I want to dig in a couple of things. So, I want to push back a little bit on what you were saying about the horizontal because there's the benefit, if you've got the horizontal infrastructure, you can run a lot more workloads. But I compare it to the enterprise. Because that was the argument, I've made that argument with converged infrastructure versus say an Oracle vertical stack. But it turned out that actually Oracle ran Oracle better. Is there an analog in Telco or is this new open architecture going to be able to not only service the wide range of emerging apps, but also be as resilient as the proprietary infrastructure? Yeah, and before I answer that, I also want to say that we've been writing a number of white papers. So, we have actually three white papers we've just done with Dell, looking at infrastructure blocks and looking at vertical versus horizontal and also looking at moving from the previous generation hardware to the next generation hardware. So, all those details, you can find the white papers and you can find them either on the Dell website or at the ACG Research website. ACGresearch.com. ACG Research, yeah. If you just search ACG Research, you'll find lots of white papers on TCO. So, you know, what I want to say relative to the vertical versus horizontal, yeah, obviously in the vertical side, some of those things will run well. I mean, they won't have issues. However, that being said, as we move to cloud native, you know, it's very high performance, okay, in terms of the stack, whether it be a Red Hat or a VMware or other cloud layers, that's really become much more mature. Now, it's all CNF-based, which is really containerized, very high performance. And so, I don't think really performance is an issue. However, my feeling is that if you want to offer new services and generate new revenue, you're not going to do it in vertical stacks, period. You're going to be able to do a packet court. You'll be able to do a ran over here. But now, what if I want to offer a gaining service? What if I want to do metaverse? What if I want to do it? You have to have an environment that's a multi-vendor environment that supports an ecosystem. Even in the ran, when we look at the RIC and the X-Apps and the R-Apps, these are multi-vendor environments that's going to create a lot of flexibility. And you can't do that if you're restricted to, I can only have one vendor running on this hardware. So, it's- Yeah, we're seeing these vendors work together and create RICs, and that's obviously a key point. But what I'm hearing is that there may be trade-offs, but the incremental value is going to overwhelm that. Second question I have, Peter, is TCO. I've been hearing a lot about 30%, where's that 30% come from? Is it up, is it from an OPEC standpoint? Is it labor? Is it power? Is it, you mentioned, cutting the number of servers in half? If I can unpack the granularity of that TCO, where's the benefit coming from? Yeah, the answer is yes. Okay, you want to find that in terms of where the big bang for the bucket is. I mean, so you really need to look at the white paper to see details, but definitely power, definitely labor, definitely reducing the number of servers, you know, reducing the CAPEX. The other thing is, is as you move to this really next generation horizontal TELCO cloud, there's the whole automation and orchestration. That is a key component as well. And it's enabled by what Dell is doing. It's enabled by the, because the thing is, you're not going to have end-to-end automation if you have all this legacy stuff there, or if you have these vertical stacks where you can't integrate. I mean, you can automate that part, and then you have separate automation here. You need to have integrated automation and orchestration across the whole thing. One other point I would add also, right? On the hardware perspective, right? With the customized hardware, what we allow operators to do is take out the existing appliance and push an edge-optimized server without reworking the entire infrastructure. There's a significant saving where you don't have to rethink about what is my power infrastructure, right? What is my security infrastructure? The server is designed to leverage the existing what is already there. How should TELCO's Charles plan for this transformation? Are there specific best practices that you would recommend in terms of the operational model? Great question. I think first thing is do an inventory of what you have. Understand what your constraints are, and then come to Dell. We would love to consult with you based on our experience on the best practices. We know how to minimize additional changes. We know how to help your support engineer understand how to shift from appliance-based operation to a cloud-based operation. Is that a service you offer? Is that a pre-sales, freebie? What is, maybe both? It's both, yeah. It's both. Yeah. Guys, just really quickly, Dave loves the TCO discussion. I'm always thinking in terms of, well, how do you measure TCO when you're comparing something where you can't do something to an environment where you're going to be able to do something new? And I know that that's always the challenge in any kind of emerging market or where things are changing. Well, I mean, we also look at, not only TCO, but we look at overall business case. So there's basically service agility and revenue. And then there's faster time to revenue as well. And actually, ACG, we actually have a platform called the BAE, or Business Analytics Engine, that's a very sophisticated simulation cloud-based platform where we can actually look at revenue month by month. And we look at what's the impact of accelerating revenue by three months? By four months? By six months? So you're looking, you're not just looking at TCO. We're looking at the overall business case benefits. Yeah, exactly right. There's the TCO, which is the hard dollars. CFO wants to see that. He or she needs to see that. But you can convince that individual that there's a business case around it, and then you're going to sign up for that number. And they're going to be held to it. That's the way the world works. At the end of the day, Tocos have to be offered new services, because look at all the money that's been spent on investment on 5G and everything else. 0.5 trillion over the next seven years. All right, guys, we've got to go. Sorry to cut you off, but we have wall to wall here. All right, thanks so much for coming on. Fantastic. Donate for Dave Nicholson. Lisa Martin's in the house. John Furrier in Palo Alto Studios, keep it right there MWC23 Live from the FIDA in Barcelona.