 And we're back. This is Stu Miniman with wikibon.org, with EMC World, live continuous coverage from Las Vegas, siliconangle.com, where we extract a signal from the noise, try to find the smartest people we can find and share that information with our community. While we're at EMC World, EMC World almost every year competes with Interop. Interop, if you're not familiar with, it's been around for many years. I used to go to it when it was NetWorld Plus Interop and was a heavy networking show. It now also covers mobile and cloud and many of the developing topics. And I'm very excited to bring to our audience Ivan Pepelnyak, who has come all the way from Slovenia to speak at many of the networking sessions at Interop. Ivan is with NIL, and he's the chief technology advisor. Thank you, Ivan. See, Ivan wears many hats. I've met Ivan through his, he's a prolific blogger. He's on Twitter. You can find him at iOS Hints there. He does lots of webinars. And he is just one of those people that shares with the community. He's helped me understand a lot of these, you know, new waves of networking because networking is complicated and it takes a special talent to be able to simplify some of those things. So, Ivan, thank you for taking some time out of your busy schedule and joining us here on the field. Thanks for inviting me. It's a pleasure being here. So, you know, Ivan, you know, storage was kind of boring for years. And we think over the last couple of years, storage has kind of heated up. We think virtualization on the server side really kind of hit this ripple effect through some of the infrastructure pieces. So, storage is changing. And now networking is changing. When I became an analyst three years ago, you know, everybody in networking said, it's just Cisco. I mean, I look at your blog used to be, you know, iOS Hints and your iOS, you know, Hints on Twitter. It's Cisco. It's much more. It's a blog to IP space. So, you know, tell me, you know, what's the big wave, the themes that you're seeing in the networking industry? Well, the big theme is network virtualization. So finally, the hypervisor vendors got sort of involved in networking. And it's amazing what we are seeing in the last years coming, for example, from VMware or from Microsoft with the new Hyper-V. So they actually, you know, I'm sort of joking, but I'm not joking. What I'm saying in the past, they were lazy. They took the easiest way out. They implemented the simplest possible thing they could, which is a VLAN switch. And they tried to keep it as simple as possible. It didn't even run spanning tree, for example, which of course, of course, cost everyone else a huge amount of problems, but they just ignored that. And in recent years, things really heated up. They started figuring out that networking is important. And they're rolling out solutions that actually work and work well and work at scale, which is amazing. Yeah, so if I hear you, you're saying, you know, while the networking hardware guys are the ones that get all the discussion, it's VMware and Microsoft that are really critical to this next generation of networking. Well, you see, I always compare this to the telephone network. So in the good old days, you have, you know, the rotary telephone. And you had the ladies sitting at the exchange office and you would call them up and they would patch the cable. And so you would get your connection. And that's how we are doing VLANs today. Actually, the virtualization guy is calling up the networking guy and asking him for a VLAN. And the networking guy takes a virtual patch cable and patches the ports together on the server sides. And that doesn't work. And so what the networking industry is trying to sell us all is, you know, well, let's move from the manual patch cables to the relay exchange. And eventually we'll get, you know, to the SS7 protocol and everything else. And what the virtualization guys are really doing is they're bypassing all that. They're saying, let's go to Skype. So what do you want to have? A telephone exchange or Skype? Yeah, that's, I totally love your analogy. You know, I think about the, you know, the storage world, you know, all of these administrators, you know, they're former, you know, UNIX administrators, they love just, you know, they don't use GUIs. They do command line interface. And networking is a lot of the same thing today. It's just very manual and moving. So don't mix manual with CLI. Okay. Because everyone who has to do something eight hours a day prefers CLI over GUI. I mean, I had great fun watching my poor storage admin when I wanted to set up a lab with VMware ESX hosts. And we were using Cisco's UCS. So my UCS guy prepared all the conflicts in advance. So in the morning, it was cut and paste of CLI commands. He had like four servers, plus fiber channel, plus FCOE, plus interconnects, plus VLANs, servers ready to boot, done in 10 minutes. And then we called the storage guy. And he came and he locked into his console. And he only had to create four LANs. And he had to map those four LANs to the worldwide names for the servers. And he spent half an hour clicking, clicking, clicking, clicking, clicking, clicking around the GUI. His amount of work was like one-tenth of the amount of work the other guy had to do. And GUI was just blocking his productivity. So. Yeah. No, I think that's a great point, Ivan. It's too often we look at the tools. And that's why my next question I have for you is so much of the SDN discussion has been about some of those tools, specifically things like OpenFlow. You've talked a lot about network virtualization, but you hadn't said SDN yet. Can you break down for us? We talked to Brocade yesterday. And their emerging technology group talked about network virtualization, SDN, and NFV. And was trying to help give our audience just what are these technologies? Because, to be honest, you asked at this show here at EMC World, there's probably 10 people that have heard of NFV. I'm sure more over at Interop. But can you kind of walk us through where do those fit in your mind? OK, so let's start with OpenFlow. It's really easy. OpenFlow is a very low-level protocol that allows you to download forwarding entries into the switches. Right. So OpenFlow is part of the toolkit for SDN, correct? OpenFlow is a very low-level screwdriver. OK, and so what about SDN? SDN is whatever you want it to be to attract VC capital. And how does SDN and software, I'm sorry, network virtualization match? OK, now let's correct the SDN joke. So SDN, in its strictest definition, is the idea that you decouple the forwarding that's done by the physical switches from the control plane, which is communicating with the outside world, and from the topology calculation. So the idea is that you don't look at the network one switch at a time, but that you take a slightly bigger picture. Let's say right now they got as far as 50 switches and then they hit scalability limits. Yes, I told them they will. And so you manage those 50 switches as one entity. So it sort of reduces your management pain, and it sort of allows that controller to do something smarter sometimes, because it's looking at the bigger picture. See, it's like if you have 50 small switches somehow connected together, and you're trying to manage all of them, or if you have one giant switch like Nexo 7000 or the new Arista box, or the new Brocade box, or the new Whoever's box, and you connect all the servers to that one switch. You know that that is simpler, but that's expensive. So what if you could have that same model of one switch, but actually have it implemented with pizza box switches, which are cheap, and have that controller be like the abstraction layer between you thinking it of one switch and the reality being that there are 50 switches there. So that's the SDN in the strictest definition. And then, of course, every vendor is saying, well, SDN is software-defined networking. So whatever we do, we have an API that is SDN, and we have load balancer. It can actually be configured. That's SDN and so on and so on. Yeah, so just like cloud, everybody in CloudWatch, there's SDN watching. Yeah, SDN has as much meaning as cloud. Okay, so we're so many topics we want to cover. I wonder if we can kind of go a little bit more rapid fire. So NFV. That's load balancing, yeah. Okay. Well, it's routing functions. Well, the idea is that, you know, once you go above simple packet forwarding, once you go into things like even DSL termination or load balancing or firewalling or anything that actually has to touch more than the packet headers, those things are computationally complex, which usually means it makes sense to do them in the CPU, not in the A6. And actually, if you open a box, if you open a load balancer or firewall from almost any vendor, you will see there's an Intel CPU inside. Only, you can't buy their solution as a software package. You have to buy the box because their business model is, we sell you the box. And major service providers got really upset with the business model. And they said, we want to virtualize those things. We want to run them on commodity Intel hardware. We want to have those things that we can spin up on demand. I don't want to buy another load balancer. I want to spin up more instances of the load balancer. And that's the idea of network function virtualization. So whatever functions we have in the network that are computationally expensive, let's move them into the VMs. So Ivan, you've got a huge resources of blogs that you've written, your seminars that you've done. So we're not going to have time to go into too many more of the technology. So I want to just switch for a second and talk about the people. So you know network administrators and you talk to all these guys. I had Martin Casado on at VMworld and he's banging on the table. We are going to transform networking, networking and falling asleep. There's no innovation. What is your advice to network administrators today? What's their job function going forward and what should they be doing to still have a career that lasts a few years? Well, let's put it this way. Some people are afraid. And some people are rightfully afraid because they haven't invested into themselves. We will not fix that. It's their problem. And they're getting the rewards for their past actions. There are other people that would love to move forward but they're doing simple things like they're configuring VLANs today. And so they're afraid that when I stop configuring VLANs because I don't know, network virtualization will take over. What will I do? And the reality is there's so much else to be done but we're not doing it because we're busy configuring VLANs. In data centers, I've read somewhere, of course I forgot the source. You spent 80 to 90% of your people resources on operations. Yeah, you've seen that quote from everywhere. Well, we talk about this wave at Wikibon. We said we need to eliminate the undifferentiated heavy lifting. Yes. So if you spend 80% or 90% of your time on doing nothing, on keeping the lights on, that's waste. So if you get rid of that, hopefully you will have time to focus on things that actually matter to your business. So the first advice is don't be afraid. There will be plenty of work. It will not be the same work that you're doing today but we've been experiencing that for the last three decades at least. Great, so the final question I have for you, Cisco has dominated this marketplace and if the, you know, not 80%, 70% to 90% somewhere that is wasted on keeping the lights on, you know, is that Cisco's fault? And does that mean that we have to change horses if we are truly going to change the market? Cisco's not listening, so just give us your advice. I want to phrase it correctly. It's not Cisco's fault, it's the industry-wide fault because what you're seeing is that every single vendor is jumping at niche problems that don't make sense if we would step back and re-architect the applications. Now, someone, you know, things are flowing down usually. So the application people don't want to change anything. The server people don't want to change anything. Virtualization people don't want to change anything. It all lands in the network. And instead of pushing back and saying, hey guys, this is not how networking is supposed to work. VLAN is not supposed to be 3,000 miles long. It wasn't designed that way. Instead of that, the networking vendors are crazy inventing technologies to solve the problems that shouldn't exist in the first place. And then of course, once you add all that complexity in your network, the network is so complex that you're not doing anything but desperately trying to keep it up and running. So it's not just Cisco, everyone is doing it. No one can exit this game. That's the problem. So Ivan, I want to give you the final word on the interview. You know, what's exciting you in the networking space these days? Well, SDN is definitely exciting. Overlay networks are exciting. Virtual appliances. So I'm really focused these days on three things. Virtualization, scaling. So how do you scale to build bigger? Let's say private clouds because the guys building public clouds know what they're doing. And the third thing is how can we fix the applications? Because if we as the networkers can help the application guys realize what the problems are, maybe they will listen. Maybe they will change some things. Yeah, so Ivan, I really appreciate you taking time to hear. As we know, these transformations take many years to go through. There's lots of pushback on change in the industry. But absolutely, hyperscale, moved to modern applications, and some of these just really transformational things. I really recommend that our listeners check out Ivan's blog post, ipspace.net. He's iOS, hints on Twitter, and sign up for one of his webinars. And if you have questions, he's actually really good at responding to those, even sometimes writing blog posts out of it. So Ivan, really appreciate all you've done for the community. Appreciate you coming over and sharing with us. And this is Stu Miniman with Wikibon, siliconangle.com's live continuous coverage from EMC World, but we're right back after this quick break.