 from Aluelto, California. It's theCUBE at Pier 2.0. Brought to you by the Pier 2.0 Foundations. Learn, connect, and grow. Now here are your hosts, John Furrier and Jeff Frick. Welcome back everyone. We're here live in Silicon Valley. This is theCUBE Silicon Angles flagship program. Go out to the events, extract the signal from the noise. I'm John Furrier, my co-host Jeff Frick here at Pier 2.0, an organic ground breaking organization with all the internet executives together. Experts and engineers talking about the future of networks. Our next guest is Paul Wunby-Hagan, chief architect of IAEA Networking Co-author of IEEE Shortest Paths Bridging. Welcome to theCUBE. Thank you. So obviously, you know, networks are hot, right? Networks. Some say the networks are the bottleneck with all the advantages in compute, cloud, virtualization, a lot of activities, software defined networking is exploded, different paths, all this stuff going on. So what is the state of the union in networking and what's underneath that's forcing all this change? Well, there's a whole lot of different drivers out there causing change. I mean, everything from the proliferation of tablets and a few years ago, the idea of a tablet and the one standard desk didn't really exist, right? We had the original tablets, but they didn't take off like we had. At the data center, we've got servers who are now very virtual moving around, right? And they're making a single environment look like many, many environments, single, multi-tenant. We're also watching a proliferation of devices with the internet of things, drive things. So it means enterprises have to start thinking like they're service providers because they don't want to put the MRI in the same network, say with guest wireless. And that kind of creates some interesting issues. For the doctor's line. NFV is hot right now too. NFV as well, you try and make sure you got service chaining involved in that as well. This is a catch up, quite honestly, in the networking space. I mean, for the last 20 years, we've been reusing the same protocols to operate a network that's been very static when if you think about what I just said, everything connecting to this is very dynamic. Moving around, and the original networking models, you didn't like things to move around. You wanted things to stay where they were and now you have to deal with things that are mobile. And everything is mobile. Your entire edge is a mobile environment, right? Whether it's VMs or tablets. So what are some of the things involved? Okay, so like provisioning, people think about like providing ports, packet traversing, different networks, SPB, SPB's shortest path bridging is a paper you wrote. What does that mean? Define what that is in context to some of these trends. So, short-spath bridging, there were a whole lot of very talented people I worked with and created a protocol. It was a child of necessity of growth and where we looked at the migration beyond Spanish tree into dealing with the mobile environment. Like for example, for years, we've been deploying data centers where, and I say years, I would say two years. It's been, as we look at the growth this thing. And internet years. And internet years, yes. You have to be very clear on that. We're no longer dealing with Spanish tree. We're dealing with all the issues of, hey, I want to move a VM from rack one to rack 48. That would have been a nightmare in the old world. You know, that's VLAN mapping and all that stuff. And you see other ways of trying to deal with it now with overlays and VXLAN and things like that. But you could also use the native layer in combination with that. So that you can actually go straight on the internet, say I want to move a VM from here to there and the protocol just learns and moves it. Now it happens within a few milliseconds. That is a huge boom to a data center environment where basically the network disappears. It becomes nothing more than a black box and you just do the services you want to do. That's a very important thing is that the network has to stop being a limiting factor, a speed bump in deploying services. And now it's become an enabler of turning on new services. Is that workload driven? I mean, is that becoming the key driver workloads? We hear that buzzword a lot in IT and data centers. Yeah, it's another word of term. I mean, it's over the workloads of driving it, but it kind of comes down to where the pressure points are with vertically integrated applications that have certain requirements. But now in virtualization you can spin servers anywhere. So where's the drivers? The network adapting to the change from the workload app and how's it all connected? Honestly, I think it's even simpler than that. It's money, it's budget. Every environment I know in this world is having the issue of their budgets are shrinking, they're asked to do even more and more every year and that workload means, that term means something completely different from one environment to another environment. If I walk into the hospitals who deploy SBB versus I walk into the stock exchange where it's employed SBB, both have, their workload requirements are different. How we saw the problem is going to be similar because basically their problem was born in the fact that the network is very static, it didn't adapt, it didn't allow them to be mobile, but then you also have now they're free to do the things they put off for years, that workload pressure and healthcare, it's epic. It's all the other things doctors moving around on their tablets and want to look at radiological data as they're sitting there with the patient just after the MRI was done and the stock exchange is quickly delivering quotes, it's being able to move VMs around and virtualize and deal with outages. That dynamic- So you're saying the optimization points have to be dynamic in the sense of the problems are still the same, managing the infrastructure, but to be prepared to be dynamic based upon the optimization objective, moving around mobile or trading high volume transactions. The expectation of the user base has changed so dramatically in the last two years that if we don't adapt the network, we're going to be a stalemate in the environment. That's a big part of the aspect of both the campus side with wireless LAN, it's the same problem as in the data center with virtual machines, everything's mobile, everything's moving and that creates a pain point because to do this usually means, in the previous life, you had to spend millions of dollars to start doing this. This is a different thing now, too. You don't have to spend millions of dollars, right? So that expectations is driven by people, we talk a lot about the consumerization of IT and people's expectations are really driven by what they can do on their phone now and why can I do all this other stuff. But the other thing that always happens in any situation, as you move to your next point of failure, you've got a system with interconnected parts, as soon as you fix one or you invest for improvement here, then it just kind of moves the bottleneck to a different part. So how's that impacting, what's the next kind of big bottleneck or the current big bottlenecks you guys are working on? Automation, there's a whole lot of buzzword in the industry about SDN and what SDN is and quite honestly, SDN is purely an infancy about what it is and what it can be. It means a lot of different things, too, a lot of different people, but the fundamental issue about SDN, what it's trying to solve is automation. So SDN to one person might be open flow, it's another person might be VXL, it's another person that might be automating attachment to the fabric, something like SPV. But the reality is, you could actually use all three together at the same time. What you're really trying to solve is, make it easier to deploy all these things because human beings can't be involved in every single configuration point anymore. What do you think about the consumerization of IT? Everyone's got three to four devices. You've got light bulbs with IP addresses, you've got trash cans with IP addresses. Then you've got an explosion of virtual machines. There are doctors and environments right now that are talking about putting an IP address in a pill for you to swallow. Do you really want to be involved in all this? It's kind of interesting when we think about what Internet of Things are. It's a lot of different things we probably haven't even imagined yet. And the role of big data has been big. I want to ask you the big data question because I think we were with Jay Adelson on earlier talking about the social web and some things just become not categories. They want to become like just native part of the fabric and big data has been involved in networks for a long time, data centers. People are using data for probes and all kinds of network management stuff. So as big data comes in with the tooling, have you seen any cool innovations around that? Because you bring up Internet of Things that's especially a big data problem. It's just another device on the network to deal with in this obviously SLAs and QOS real quiet depending upon the trickle effect coming from the device. So how does big data play to the network? What innovations are you seeing around use of data? So it's an interesting question. I kind of look at it maybe slightly different a lot of other people because I look at it from, okay, oh my God, there's a whole lot of devices coming in now, how are we going to deal with this? I have to empathize with all the guys who actually have to run these environments and they don't live at a console managing a switch. They live in managing everything that's coming into this thing. And it does go back to this automation slant but it also goes back to the fact that when you're moving to a converged environment and you've got real-time applications and real-time means something different to you versus Jeff, that becomes an issue because for example, a doctor's real-time expectation for his voice talking to another doctor in another country while he's doing surgery versus a radiologist trying to download a whole bunch of radiological data which is huge. If you're not doing QLS properly, if you're not doing traffic reservation properly on this conversion environment, you're going to have major issues for the actual user experience. That's the biggest thing with big data that we have to constantly take on its aspect from. So I'm always going to approach it from a network perspective because that's my impact. Because at the end of the day, if the plumbing doesn't work, you know about it. When the plumbing works and it's invisible, your high life is happy. Of course when an application breaks, everyone just blinds the network anyway. So in the old days, it used to be the pager goes off. What's the new pager? I mean, what goes off? Cell phones and text, is it automatic notifications? Well, the fun things actually I've seen kind of grow out of this thing is it's watching social media hit network management. For example, we have switches that will actually IM you. That there's a problem. And you can IM back to the switch and I see all I've commanded to come back and you can do that on a group chat and a group text actually on the cell phone. So that's kind of an interesting way of looking at things differently. There's also so much things we're playing around with where the idea that you could have a tablet and instead of having to have a PC console to a box, as you get close to the switch, you're now consoled into the switch and making life a little easier to actually. I mean, the personalization aspect of data is interesting is now you can not only get real-time information, a lot of network management concept, but also predictive analytics. Yes. I mean, the predictive analytics opportunity is huge. Oh, I love the analytics aspect of this thing. I'm a big metrics guy. The more metrics you have as an engineer and engineers love metrics, it means you kind of geek out on the fact that you can do some stuff with this metrics and that metrics actually in its own case is its own big data. Because now like in a lot of environments, I can tell, hey, what's the delay in general loss between these two endpoints before I do the emotion? What's the performance of that doctor he's experienced in the environment before he calls me, he's complaining that there's a problem in that network? What's all this kind of combination? We now have that available to us. We didn't have a few years ago. Let's talk about virtualization. We're big fans of VM. We're obviously, you know, the beautiful campus in Silicon Valley. Very well, you know, design, nature. It's just like, it's the new HP in my mind in the valley. It's like growing company, but they came from just a single app and now full on platform. And virtualization certainly changed, certainly with cloud open stack of these environments where, you know, the role of the hypervisors kind of blending into the fabric. So that battle for the hypervisors changes. So where do you see VM where going? Or virtualization in general, you mentioned V motion. I mean, that's the nirvana for data centers is to treat geography like it doesn't exist. So that's a lot of speed of light and physics problems. Break that down for the future of virtualization. So virtualization is going to become far more ubiquitous than we ever imagined. It's going to leave the data center as well. The constant of what a data center is is also going to change dramatically for a lot of people out there. I think for a lot of people, they think of a data center as being this big building off to the side. But another thing you're going to start to see is smaller environments out there that are located off site become arms off the data center where I can start moving VMs closer to users for improved experience and having that ability to move it on the fly is something that you couldn't do before we really VM really pioneered that environment, right? VMware pioneering that helped us do some things that are pretty powerful. That's that point, though, is that we couldn't do that because VMware was stuck with the old world of networking. They couldn't move that, because to move that VM, you have to keep your IP address. You have to keep your MAC address. From the networking world, that's a bad thing to do. You never move an IP address somewhere else in the network, right? Curious whole habit. Now, we can do that with protocols like SPV. Grab that VM, drag it across the network. I don't care if it was in LA one day or in New York the other day, right? Or in two different clauses, two of racks. It's all one virtual, one-rated environment. Move the compute where the storage is, move with the data, is all this- We're stretching out. You know, we think about a three-tiered architecture, what VMs are typically used. You have your application tier, your web tier, and all behind it, by the way, you got this database tier. And typically, they're all located directly with each other in a couple of racks. Hey, what happens, though, if I can grab that web tier and actually move it closer to the user? Now, the response from the web tier becomes faster and you can kind of distribute it out there in your cluster and your database didn't have to move. It's too heavy. You don't want to pick that up and move it around too much, but the web's pretty light. You can do creative things now that you couldn't do in the past. So let's kind of break this down. So you're basically saying all this flexibility and agile-like features of moving resource around is now going to be the new normal. So that's a pretty transformative concept. You think about where we've come from and where we are in the market. What's that going to enable? And in your mind's eye, what are some of the things that will be coming up and the fruit off that tree will be what? New apps, new experiences, what do you see? I think it's going to drive a lot of different things and there's a lot of things we don't know, we don't know yet, but one of the interesting things I'm watching in the environment as people start adapting to this is a realization that you have to think differently. Everything you've known for the last 20 or 30 years, you have to adapt your way of thinking of how you use it, right? So we used to think we build colossal buildings and we got to build around this protocol, that protocol now. You don't think about colossal buildings. You do have colossal buildings, but how do we use them in combination? We use them to do more creative things of deploying the services, like I said. You're able to kind of twist your thinking to say, now that I have this freedom, what are the things I put off for years because I couldn't do, they were too expensive or too complicated, which also goes back to OPEX costs, which is a good 80% of costs in any environment is OPEX. A lot of people think that the cost is buying the boxes and the switches and the VMs. No, it's running the environment. When something breaks, it's an incredible cost to normalization. Now when something breaks, we've got that analytics, we've got a lot of information and now that we can feel more comfortable about that, we can do more for these environments. Freeze us. So Paul, I'm going to shift gears a little bit. Talk a little bit about the impact of open source in the networking world. It used to be people would get together, we'd define a standard, we'd define protocols and it's actually interesting that G and Cisco are leading a pretty, it looks like from the outside, successful effort around internet of things and coming up with some industry protocols. But now we have the whole open source movement that's moved from the operating system and continues to proliferate through the stack where it's really the everybody that goes in and starts to define really what the standards are. Talk a little bit about the impact of open source in the networking world. Goodness, badness, and where do you see that going? I'm a huge fan of open source. Having worked in Linux for many years, I love everything about open source and I also have lived through the pain of open source in the early days. And that's a clear delineation here. You got an economy of the ability to quickly have multiple eyes on a problem. It's not led by a single vendor. You've got the actual problem set to find to the real user base. That's a fantastic aspect of the open source. The other issue though and the other side of this economy is, well, who's going to support you when something breaks? Does this require a whole lot of configuration and software development and are you an environment that can take that on? Do you even have a software development team who've got PhDs and master's degrees in software development? Or is there some middle ground where you can grab off the shelf things like open stack is a phenomenal example of something that's got package environment that allows an IT organization to quickly deploy an Amazon Web service that function. It's pretty complicated to set up. It's not for the faint of heart but it's getting a lot easier today than it ever was a few years ago. That's going to help drive a lot of innovation. And to me, that's the biggest thing that we can actually drive out of open source open standards, right? So as we work the ITF and the IEEE and open source environment as a big nice triangle of development, it drives innovation to the real problem, right? Not to a single vendor's proprietary direction. And that's a real strong statement there because the internet of things is no longer driven out of a certain particular vendor market set. It's coming from everywhere. And like I said, I mean, trash cans of IP address. You ever think you're going to be talking to a trash can vendor about your network, right? Oh, by the way, that means that the trash can has an IP address, guess what? There's a server somewhere managing them, right? Because it's important for security. You want to know if a heavy package was dropped in a trash can because that could be a bomb, right? So that has a whole lot of cascading effects to it. Ripple things that we, you know, unintended consequences and reasons. Right, right. So on the future of networking, let's talk about the big players. Cisco, Juniper, they're the incumbents, okay? There's a lot of legacy involved. East, west, north, south, mindset, you know? And certainly Juniper's been under some management changes recently. Cisco, obviously, the whale of all is getting in the server business. You see what they're doing in the server that's been very successful. Where is the converged infrastructure going? I mean, the data center is the future. I'll see how it gets materialized, whether you're a service provider or an enterprise that's going to get those arms out there and using VMotion, things like that. So where is converged infrastructure going in the sense of it's happening? So converged infrastructure is no longer just going to happen, it's kind of happening. Converged infrastructure is going to drive some interesting things. It already has been driven as well as drivers. There's a push-pull effect going on there about how do we deal with this converged environment and its impact on some of the network space and then the network dealing with, okay, we got a whole lot of these package systems coming into this environment and is it prepackaged for one environment? Okay, what's the cost basis of this as well? Because is it cheaper to build it myself or is it cheaper to buy as one package system? And then that cheaper is a very important question because you have a net present value concept you have to do this. Is the cheaper really a question of the capital costs or the cost of buying the stuff, the licenses and the people to know how to put it together? And then after it's deployed, how do you manage it? So converged infrastructure is going to become just a single point you're going to see everywhere. It's going to become the de facto standard and an environment out there because it turns a data center into a turnkey Lego kind of aspect. We can start deploying these converged environments in nice little ways. I mean even in Avaya, we're deploying converged environments. Cisco is, you see it coming from every direction which means that's another problem. It's a, how do you guys think differently? Obviously Avaya, you guys are in the big whale category and you've got companies like Newtonus that have come out of the woodwork, nimble storage, these guys just, I mean quite frankly, knowing what fun Newtonus when they first started, they were talking about thinking differently. D. Rogers was talking to me, he's like, hey, you know what, we had a hard time convincing people that this was going to happen next to you know, they're doing very, very well. So again, that's an example of thinking differently. So how do you guys think differently? What do you guys need to do? We think of it in a lot of interesting ways. So I mean as I was trying to say this to you guys, when you have all these different converged environments and they're all by different vendors, if they don't have an irreparable way, you've got a problem because you don't want to be locked into a single vendor. That's one of the aspects we take by this and we use standard operational procedures, we use open source technology, we use open standards. We're approaching it from the aspect of we know real-time applications and real-time networking better than anybody else in the industry. We have developers running software for iPads and androids and real-time software that's voice and video and we tune the network to make sure it can deal with that. And that was an environment we noticed too is that a lot of the failures in real-time applications is because you haven't configured the environment properly. You might get the QoS settings never properly but did you get all the buffers tuned properly in that conversion environment? Because if you didn't, you're going to have a lot of problems in your real-time application. Now, we look at this, okay, we got real-time because we have UC as a service. We worked with a lot of service providers now and they're offering this as an conversion environment. You have UC as a service, you have Context as a service on our network, on our environment. But when I go to healthcare, another customer of ours who also buys these services, their real-time service is that radiological data I just talked about. Their real-time service is the patient data and I can use that over again in education and in manufacturing. It's an important thing to change that. So our aspect of this today is like, how do we make sure that the environment is doing what they need to be to run their business? If your business is cooking chicken, you shouldn't care about the network. You should be worried about the productivity to make more chicken, all right? I wonder if you can just talk a little bit more about that because real-time is a great word, right? It gets thrown around all the time and what is real-time and it's really contextually sensitive, right? It's really about what's real-time for you to do something important or to make an important decision or to make an important move, right? There's really no single definition of what is real-time. No, no. I think there are certain elements of real-time that most people can agree on, all right? It's been around forever, voice, okay? That's semi-real-time. But actually in reality, you can step back and say, okay, I have accepted certain levels of quality on cell phone that I will never accept on an online. And that's a certain user experience you got to make sure you meet where your users are going to complain, right? So that's real-time. It has absolutely nothing to do with real-time. We're not real-time. It's all about expectation. And that's the subjective nature of the word real-time, right? We will accept different types of real-time on different environments. And if you don't meet that human expectation, you're going to have problems. Your user base is going to yell, they're going to demand their service or say you're not doing your job, blah, blah, blah. The other issue that I suspect is that next one, what is the user? A doctor's got a completely different set of what his expectations of a real-time application and what he would deem real-time versus a stock trader, right? One's worried about human life. The other one's worried about billions of dollars on the line for something not working for a few minutes, right? I think you asked them what real-time is. They're going to have completely different definitions of it. And if you can't meet it, you've got a problem. But the solid actually comes down to very similar problem sets, right? It's quality of the network. It's quality of the application on the end device. It's quality of the experience of the user into it and his expectations, which goes back to your point, John, about analytics. It means you can't be reactive anymore. You have to be proactive and seeing things happen before someone has to pick up a phone or complain that they can't pick up the phone or any other aspect of this thing. You have to be able to react before any problem ever really exists. Paul, my final question to you. What is the biggest game changer that you're going to see in the networking space in the next couple of years? As things like peer 2.0, this community comes together. You have software-defined networking. You have cloud, you have mobile. User experiences are changing up and down the stack. Get big data, internet of things. I mean, it's almost the perfect storm for all the theaters of innovation. Absolutely. There are two aspects. There's the things you won't see and the things you will see. The things happen in the background, like peer 2.0 is going to draw the hole. We've seen the conversations happen here. It's fantastic when you talk among ourselves and where we've been, where we're going, the things we can do now that we couldn't do before will lead us to new business models and new capabilities to deploy services faster to people in residential environments. We've got CDN aspects and all that. It's a pretty cold environment. But then there's the actual user space that you will see things change. We're sitting here right now with you guys having two laptops in the desk. I look around the environments where I look in IT environments, I don't see laptops or even desktops in environments anymore. I see 90% of people walking around of the younger generation it is with just tablets. Their whole world is that tablet. And that's going to drive a whole new way of having to deal with America as a problem. How you communicate and your expectation of the world. I laugh because I watch it with my daughter every night when she gets home from school where when I was her age, all of our friends got together after school. How did they do it? They all get on a Skype. And I look at her, she's sitting there doing homework and she's got six other of her friends or the girls on the Skype session communicating across to their kitchen tables. That is going to drive an expectation into the workforce. Yeah, and certainly mobile as well as driving that even more with app specific integrated communications. Yep. So when you get that user base, it expects one and they're going to expect that when they get to work. So real time in your opinions, table stakes at this point. Absolutely. For all applications. Absolutely. Every application has its own concept of real time. And it differs so dramatically no matter where you go. Paul Unmi Hagen, Chief Architect of IA, Networking Co-Author of ICHIPELEE 8021, AQ, Shoda's Path Bridging. Thanks for joining us here in theCUBE. Peer 2.0 inaugural CUBE event here. Their first event, great community, small kernel of folks and experts here expanding the industry. This is theCUBE on the ground and extracting the signal from the noise. We'll be right back after this short break.