>> Announcer: Live, from Boston, Massachusetts. It's theCUBE! Covering Red Hat Summit 2017! Brought to you by Red Hat! (electronic music) >> Welcome back to theCUBE's coverage of the Red Hat Summit, I'm Rebecca Knight, your host, I'm here with Stu Miniman, our guest today is Jason Hoffman. He is the vice president and global head of cloud infrastructure at Ericsson. Thanks so much for joining us here. >> Thanks so much for having me. >> Thank you. So, I want to start out by talking about the cloud market today, and where we are. We've heard here at the summit that the number one thing on customers' minds is cloud strategy. Is that, does that resonate with you? And what are you hearing from customers? >> Cloud's an interesting topic. Because it's both an infrastructure approach, sometimes people use it for a delivery model, sometimes people use it to describe a business model, sometimes people use it to describe a number of things. For me, for most companies, cloud's not strategic. Cloud's tactical. Cloud may be strategic to Amazon, cloud may be strategic to Microsoft, we've yet to see whether it's truly strategic for Google, and that, but for most people it's tactical. And I think it's a fine distinction to make, because tactical means there's stuff to do, and if you look at the thing to do, it's pretty clear that that approach to infrastructure is on highly accessible, industrialized infrastructure. And I anchor that in infrastructure because in the absence of that type of infrastructure then your delivery models and business models and stuff don't apply, basically. >> Jason, you bring up an interesting point, we heard in the first day keynote, they said, "One of the top priorities for end users is, "to build a cloud strategy." Because when I talk to most users, they've been very tactical, it's like, well, we moved our scene around, to SaaS. Everybody did. I'm starting to move certain business apps to SaaS. I've, everybody tried Dev, you know, on Amazon, and played with a number of other clouds, but you look at that, I had a guy who wrote for me, he was like, "We don't have hybrid cloud. "We have composite cloud." Because I've got all those things there, it's like, "How do I shop?" It's like, well, I stop at a convenience store where I need, I go to the store, and I get lots of stuff from Amazon. Sounds like how we do cloud! So where do you see, from the users that you talk to, I mean, you've been involved in this since the early days, so it feels like we talk a lot about cloud stuff, but it feels like we're getting a little bit more real, I'm heartened to see that users are starting to, you know, think about strategic, how they put things together, where do you see the customers you're talking to? >> Well, I mean, I think a lot of the customers are educated by the market. And the market tends to use a lot of these marketing terms, like composite, rather than getting down to what the actual sort of issues are. The issues, even if we go back to the highly accessible, industrialized type of infrastructure, I mean, the issue there is. Customers need to simplify things. They largely would like to do that through some type of, quote unquote, "automation". And they would like everything that they do under one governance model. Essentially, what you'd call life cycle management, or governance, or whatever. But these types of things that both the infrastructure, and the nature of the things that you do on it, are too complex. And they tend not to be, you know, once they're basically put in place, they tend not to be touched again. You know, so in fact, if you look at a good definition of like, legacy. Legacy, for me, is when the cost of keeping something as is, is basically cheaper than the replacement process. And actually if you look at that sort of cost curve, the day that something is best to leave as is, and it's very expensive to replace, is the day it goes live. >> So the legacy curve-- >> It's like driving a car off the lot. >> The legacy curve is actually a U-shaped curve. You know, in there. And in fact, the day something goes live, it's like a sieve and at some point, three, four, five years down the road, it's not legacy, and then about three, four, five years after that, it's legacy again. >> Reminds me of what Jeff Baso said, "There will never be a day two for us, "we are always going to stay day one, "focused on creating new stuff, "because otherwise once you go kind of steady state, "the decline is coming soon after." >> Yeah, and I think, I mean, if you look at now, myself having started I dunno, 20 plus years ago in scientific computing, and then, you know, meeting HPC and these types of efforts, and then showing up in sort of what one would call the quote unquote, "cloud world", the effort there has largely been the same, I mean, it's one of these ones where, infrastructure by its very nature is always risky to manage. You know, meaning things like, continuously deploying things into infrastructure, always runs the risk of making infrastructure inaccessible to people. You know, you don't instantly like, if a new operating system image comes out, you don't reboot 100, 000 servers. And then the applications that sit on top of it, and have historically inherited that risk from the infrastructure. And the effort has largely been separating those two. So that applications get freed of the risk from the underlying infrastructure, and then you can start applying a certain approach to the infrastructure, not really caring about what's on it. And cloud is pretty much this emergence of infrastructure, as its own practice. And it's going from traditional IT environments, where that sort of operational model, and that economic model, is not the center of anybody's product designs, to an operational and economic model that is at the center of everybody's sort of product designs. And the KPIs around what you're doing there, and whether you're successful, are relatively simple. And that is, you know, are you continuously improving, in terms of capacity, capabilities, and unit economics. You know, if you're not, if you don't have an exponentially decreasing unit economic curve, within, you know, a five year period of time, then you're probably not doing, you know, quote unquote, "cloud" right. >> Jason, we know with Ericsson's background, you've got strong positions in the telecom space, working a lot on the NFB solutions, expect that's a hot area of working with Red Hat, can you kind of sketch out for us, just where Ericsson sits in the marketplace today of where customers are coming to you, with kind of key partnerships? >> The element for us is that we're a 141 year old Swedish multinational. And we've been in the same business for 141 years. And that we provide the communication backbone for the world. And so, on one part of the business, is that if you see 2G, 3G, 4G, LTE, or stuff on the phone, or if you use a mobile phone in somewhere like North America, 100% of all that traffic goes over Ericsson applications. And so you know, we're a very dominant provider of that radio edge, if you will. The other thing we have, from an application portfolio perspective, is everything you need to be a telecom. So everything ranging from the network functions, through OSS environments, PSS environments, customer front ends. And the like. And the telecom space, has been undergoing some pretty dramatic changes, in some ways. Or not dramatic changes, depending on one sort of attitude over the last few years. The first one, is even though that far radio edge, if you will, has tens of millions of points of presence globally, that talk to billions of devices, that's a very algorithmically driven, event driven, very industrialized type of infrastructure. It's fair to say that if you look at the typical definition one would apply to hyperscale, it applies to the radio edge. It also applies to the sort of hyperscale cloud providers. In between those two things, is roughly about one to two trillion U.S. dollars of investments that sits in tens of thousands of facilities globally with pretty much the computer history museum of hardware and operating systems and everything else inside of it. It's effectively this whole middle mile infrastructure. For us, we sort of view the virtualization of everything that's in the telecom space, and uplifting it to a common infrastructure approach, is pretty critical to get a handle on that middle mile infrastructure. >> Jason, you know, when I think about the telecom space, we're talking about 5G now, these roll outs tend to take a while, we measure things in years, if not decades, before they roll out. Jim Whitehurst got on stage this morning, said, you know, "Planning is dead, "we don't do a 10 year big data plan, "we won't do a ten year AI plan." How does, you know, your set of the market live in this greater changing world, how do you look forward, can you predict like we did in the past? That's a, so I think yes and no. I mean, we, you're correct in that if you take a market like India, India's going to hit 200% 2G penetration in 2020. And so, we're talking about something that's going to happen 20 years after it started in other parts of the market. You have spots like Papua New Guinea, that are, have 2G networks throwing 200, 300% year over year. And so when you look at this globally, the roll out of these technologies are on, typically, a 30 year time schedule, overall. And they result in people owning assets that they have for at least 30 years. In many cases as well. So on one aspect, there is very long term planning that has to exist, and we have some customers we've had for a century. From that point of view, they have to know those types of plans. On the other end, is if you look at what we have to start doing from a technology standpoint, is enabling just a tremendous amount of flexibility, to be able to just get out of this legacy hamster wheel. This idea that the second you go and deploy something somewhere, you can't touch it again, and you can't continuously improve it again. You know, we're trying to sit down and say, "If we go out and do these types of deployments, "we need the ability "to actually continuously improve these." In fact, that's really what the industrialized word means for us. And so, for many of these things, you know, we've been taking certain approaches. I mean, interestingly enough, if you look at the actual mobile edge of things, the radio edge of things, that is one where there is one hardware platform, it comes in tens of form factors, we'll go and deploy out that infrastructure, and whether it's a 2G, 3G, 4G, 5G, narrow band ILT infrastructure, it's purely unlocked in software. And so that is a good software to find, sort of infrastructure. And you do it that way because at the end of the day, somebody's got to go into a jungle and put a cell tower up, and you don't want to go out there again. And we want to take that same sort of thinking throughout the rest of the infrastructure. >> Jason, it always fascinates me, if you look at this space, I mean, I remember we interviewed you back at Joint, we were talking a lot of the show about containers, and there's things that we've tried in the past, and now it's coming at it again on edge computing, is something we're starting to talk about, it's a little bit difficult to predict how long some of that will take, and some of the cycles, but maybe to wrap us up, talk to us a little bit about this, you know, Ericsson's planning for the future, what are some of those big challenges that you see, that we as an industry need to tackle, and maybe some of the things that will earlier wins and stuff, that you know, might take a little bit longer. >> Yeah I mean, a little bit longer, and if I look out really, probably the next 15, 20 years, I mean the type of middle mile industrialization they're doing today is a prerequisite for all the IOTUs cases to show up on 5G. Because there's simply not enough middle mile capacity in the world, or capabilities there, to even have it go to an amazon type perspective. So even if that all ends up there, so, now it's going to be this very large effort of I think, bringing these two worlds together, and on the hyperscale cloud providers, you have a very industrialized approach in infrastructure, and it's very supply chain driven, and it's very sort of, easy to cookie cutter out. On the extreme end of the infrastructure, on the radio edge, it's the same what. And, but all the stuff in the middle's not. And if you look at the channelage from a cloud perspective, it's about taking those sort of learnings from the book ends of the infrastructure, bringing it to the middle, if you will. And starting to think about what, because edge computing's a great example, it's like, what does it mean when you start using that, from an end user perspective? You're not going to go to a pull down menu, and pick 12, 418 availability zones from a list of 120, 000 global availability zones. You're not going to automatically have issues around data sovereignty and everything else like that, just sort of be tackled. So I think in many ways, when we start looking at some core concepts that live in hundreds of facilities, and sort of large cloud providers, and some of these concepts where we manage to do things in tens of millions of base stations, talking to billions of devices, the reality for me is when you look at the redo, and now the more global use of that infrastructure that's in the middle, it's going to require developer accessibility, and a certain degree of programmability that doesn't actually exist yet. And you know, of course it's an open question of how much you do it. I mean, you know, you're going to have to scale topologically, rather than sort of scale in other ways. And a lot of that stuff's just not done yet. >> Jason, thank you so much for joining us, we really appreciate your time. >> Anytime. >> We'll be back with more of theCUBE's coverage of the Red Hat Summit, after this. (electronic music)