 Live from Barcelona, Spain, it's theCUBE. Covering Cisco Live 2018. Brought to you by Cisco, Veeam, and theCUBE's ecosystem partners. Welcome back, I'm Stu Miniman, and this is theCUBE's coverage of Cisco Live 2018 in Barcelona. You know, I'm a networking guy by background, but there's certain people in the industry that really I've gone to to learn, been really thrilled that I've had the opportunity to get to know, and every once in a while I get to bring them to our audience, and really happy to bring back to the program Yvonne Pepelniek from Slovenia, blogger, author, webinar, generally, you know, that network guy that, you know, those of us that watch the industry knows. Yvonne- That grumpy networking guy. Yeah, you know, aren't most networking people, at least online, a little bit grumpy? And when you meet them in person, though, it's a slightly different experience, so thanks so much for joining us. Thank you for inviting me. All right, so 2013 was last time that we actually got to do one of these in person. That's true. So, networking, it's all the same, right? I mean, we're probably still working on 10 gig rollout, maybe 25 or 50 gig, you know, speeds and feeds, and, you know, oh, okay, you know, IPv6, I think we're kind of getting there, and lots of other acronyms we could talk about, but really, what's been some of the big things that you've been looking at? What are customers actually doing, and what are customers thinking of that you've been playing with? Well, it's amazing how little has changed. People are still talking about SDN, like that's the big thing. No one has delivered on that apart from some point products like VMware, NSX, or Cisco ACI. Cloud is still the thing that will happen next year to most companies. We hear how 90% of all the companies that participate in some survey are using Cloud. And then the next question I ask is, well, is this Office 365, or is this something more? And they go like, well, the survey didn't differentiate on that, so thank you. Yeah, but yeah, look, the SDN, you know, front of mind said SDN stand for still does nothing. That being said, ACI, NSX, there's customers using it. Oh, absolutely. It has not totally transformed the industry, like they said, Cloud, I've yet to find a company that's not doing some SaaS, and unless you have some regulation or things like that, you usually have some sandbox that you're doing, some public cloud. Absolutely. But absolutely, people, they still have data centers, despite the, you know, they might not want to, it might not be their own building anymore. I was just talking to a service provider and the like, but yeah, so I mean, the more things change, the more things stay the same, right? Absolutely, yeah. Well, yeah, we do see people moving to Colos, or they would build their stuff somewhere else, or whatever, but it's amazing how much interest I'm still getting in data center design courses. So there are still zillions of people who think that that is important, and yes, we all know we'll go to the cloud, but everyone has his own hurdles. And so, I think that eventually everyone will get to some sort of hybrid cloud where some stuff will be there, and some stuff, legacy, whatever, will be here, and we'll have to live with that forever. Yeah, I mean, those of us we think back, I remember when this wave came, it was like, well, how about the XSPs in the 90s? There were two reasons why it failed. Number one, there wasn't enough network, and number two, security. Well, you fast forward to two decades, and the network's gotten way better. I've got great speeds and things like that, but physics is still a factor, and security is even more of an issue today than it was 20 years ago, I think. Although, this started as a joke, but it's becoming more and more true. If you move to the cloud, your security actually improves, right? Because they have some security and you had none before. I at least get to rethink my security when I make some transformation. They have the basics right, like physical access control, multi-tenant separation, encryption, strong authentication, they get those things right because otherwise they would be out of business. Okay, so we spent more than a decade with how virtualization and networking, have we gotten most of that at least reasonably well now? There are still people who don't get that Ethernet was designed to be used on a single cable. So they still think that stretching a single Ethernet across wide distances is a great idea, and everyone is still letting them get away with that. Fortunately, the cloud vendors aren't buying. So if you want to move to Amazon, Google, whoever else, you have to redesign your applications and make them work correctly. So eventually this thing will die, but it's like cobalt and mainframes. It will be there forever. Yeah, I mean, we've been saying for a few years on theCUBE now that the challenge of our time is really distributed architectures. And of course that a huge impact on networking. So how's the industry doing? How'd you rate, say we're here at Cisco Live, how are they doing helping customers with these challenges? Most of them don't. I mean, if you look at a typical enterprise application, it still isn't developed for a distributed environment. Yeah, they use three tiers of servers, like always, but then they try to cope by solving all the problems in the ops phase when they deploy stuff. And that's the biggest problem we are facing today. We are not changing the development processes and paradigms. Well, we're actually here in the DevNet zone. I mean, I gave Cisco Kudos, last time I came to Cisco Live was 2009. There weren't, we didn't talk about developers. Everybody was, you know, doing plug-fests and getting their latest certification, but they're trying to embrace the developers more. There seem to be more of them here. That boundary between network operator and developer, do you see, you know, is there communication or the network guys still stuck in a closet somewhere not talking to anybody? Well, there are two types of challenges. The first type of challenge is that the network guy in particular, but ops teams in general, are still not invited to the table when new stuff is discussed. So the application developers dream up something based on their best knowledge. I mean, they're not evil or anything. They just don't know the operational impact of their decision. And because the networking security virtualization storage, people are not at the table, then they have to cope with whatever these guys dream up in isolation. I'm never blaming them because, you know, we should educate them and we are not doing that. So anyone who manages to bring security networking storage people in, when the application architecture is being designed is my hero, but there are only a few of them. All right. The other challenge is that the networking people don't realize that their world has changed, that they can't manually provision VLANs the way they've been doing for the last 20 years. And it's amazing once they get it, once they start the simple automation stuff, how creative they become, what types of problems they solve that they don't have the shackles of CLI anymore. I shouldn't be saying that. I'm the old CLI jockey, but it's amazing how much can be done once you realize that you don't have to do everything manually. Yeah. Cisco's, you know, not shy about putting out strong visions, you know, marketing is definitely part of what they do. And the keynote this morning said it's a new era and new infrastructure powered by intent informed by context. I sounded like a nice message, but this whole intent-based networking, what's your take on it? Is this, you know, are we going to come back five years from now and talk about intent-based like we did SDN? Well, let's keep in mind that this is all hype. What we're really talking about is an orchestration system with an abstraction layer. Because first, it's really hard to define what intent-based is because there's no good definition. But there is a definition in programming, which goes, differentiates between declarative programming and imperative programming. And if we use declarative programming as something which could be intent-based, that thing says, well, I don't tell the machine or whatever the system, how to do things, I just tell it what to do. And if you take a look at that from that perspective, then you figure out that every device configuration is an expression of your intent. You never tell the device how to work, you just tell the device what to do. It's a dream, Yvonne, I think back, we used to manage individual boxes, then we kind of created a little bit more pools, and the challenge they see right now is with the explosion of vice, we're not going to have time to talk about all the IOT, edge piece and everything, but there's no way an admin or a team of admins are going to be able to help there. So I need to infuse, I hate that, the ML, AI, choose your buzzword of choice, though. The machines need to be able to manage that a little bit more autonomous, networks are something they take deep, too. I understand, you're skeptical, so how do we get there, or, you know, otherwise, this whole label crash. There are three totally different things here. The first one, I totally agree with you that we should view networks as a single entity. Configuring boxes is stupid. It's like, CIS admins don't do that. Well, some still do, they get the results they deserve. So we should start thinking about network-wide data models, which are then translated into device intent, which is really device configuration, and that makes absolute sense. But remember what I said, this is just a glorified orchestration system with an abstraction layer. The second problem is machine learning. Some of the things we are dealing with have physical limitations, like the speed of light, or the number of things you can put into a hardware forwarding table. And once you're faced with those physical limitations, it's like, you know, self-driving cars. Yeah, they are self-driving, but they cannot go 300 miles per hour because of laws of physics. So it's one thing to say, well, I have these infinite resources and I can learn how to play Go in eight hours. And it's a completely different thing to say, well, now I will figure out how to deal with my network, which has these physical limitations. And also, you know, whenever I hear about this autonomous distributed thing is, we have routing protocols. They have been autonomous and distributed and self-healing for decades. And we didn't call them machine learning or artificial intelligence. And finally, once you get to the bottom of it, then you're faced with all those physical limitations. And now let's say you want to solve a simple problem, which is how do I optimize the use of my network? You do some research, you figure out that this problem has been solved 20 years ago. There are companies with commercial products that have solved this problem. It's just that no one is using them because they're too expensive. Because what you can save by using them doesn't offset the cost that these people had to invest into R&D to make this work. So machine learning, yeah. Can you make it cheaper? I don't think so. All right, so Yvonne, I want to give you the last word. Grumpy networking, what do you look forward to the most at this show? Any final anecdote you want to share before we have to wrap? Well, the one thing I am looking forward to see people to start automate their networks, to jump over that mental barrier. And when they break through it, it's amazing how many success stories you get. So I know a number of networking engineers who were on my automation course. And six months later, they write me saying, now I have this thing in production. And we cut down the site deployment from three days to five minutes. When I read emails like that, it's like, you're my hero. Excellent, I love it. For a grumpy person, sure sound a little bit of an optimist about what some of the people come in and get this. Maybe your realist is more right. Yvonne Pepelniek, really appreciate you joining us. We'll be back with lots more coverage here from Cisco Live Barcelona 2018. I'm Stu Miniman, you're watching theCUBE.