Hello, we are very lucky to be talking to doctor Russ Clark today. Doctor Clark is a researcher at Georgia Tech, he actually has a unique position of being both research faculty. And has an appointment in the Office of Information Technology at Georgia Tech. So he ha, he has both research savvy and operational knowledge particular how things work on a, on a real on a real campus network. Russ actually [SOUND]. Received his Phd at Georgia Tech so he has quite a history at Georgia Tech and in between that and now he helped start a network management software company. So he has quite a bit of experience in [SOUND] network management so we'll talk to him a little bit about that, both particular focusing on network management problems on the campus. And more recently he's been very involved in the efforts to deploy open flow technology on the Georgia Tech campus and in connecting open flow networks at Georgia Tech to other campuses across the country as part of the national science foundation/ [SOUND] global environment for network innovations or genie. Effort, so we will ask him a few questions about that as well. So, thanks a lot for taking the time, Russ, it is good to see you. >> Yeah, good thank you for having me. >> Great, so, so thanks, I wanted to basically start with one question about open flow deployment on the Georgia Tech campus network. You have been really, really heavily involved. In getting an Open Flow deployment on the Georgia Tech campus network, as, as well as making efforts to connect that Open Flow network with different campuses across the country. And I was wondering if you could, could talk about your experience in, in getting that deployment going. You know, what did you learn about Open Flow as you, as you got deeper into the deployment efforts. Both in terms of like, the challenges involved in getting it rolling on a single campus. You know, efforts in connecting open flow networks, etc. So, what's been your experience so far with, with open flow on the campus side? >> So, it's it's it's always fun to be a part of, of something completely different and completely new. And, you know, they don't call it bleeding edge for nothing, right? We've we've learned a lot, we have experienced in a various stages of implementation and support by the vendors. We've worked with a lot of different vendors and a lot of different, you know iterations of releases and have. You know, learned a lot along the way. I think probably, you know, to start us off the, the most important thing that we've sort of learned and, and come to deal with, with SDN in general, and with Open Flow is, is kind of that expectation around and education and sort of demystifying it for folks Probably the most common, mistake that I'll hear people make, is when someone describes a particular, you know, network management, or traffic management, or policy issue. They'll say, well, but, SDN is going to solve that for us, right? Or Open Flow's going to solve that for us, right? And while we'd like to. I'd, you know? I would obviously like to say, yes, absolutely. The right answer is no, not by itself. It doesn't solve the problem. SDN and Open Flow give us a great new tool, so that we can work on solutions to the problem that we didn't have access to before. Basically when, when we have a. You know, historically when we have had a problem that, that needed a, a new solution, whether you know, scaling to the campus or dealing with a new protocol that has appeared that the students have, you know, some new technology brought to campus we, we, we've have had our hands tied and we have had to wait for vendors to help us solve the problem, and now. We actually have, have some tools that we can go start thinking about you know solutions ourselves and, and trying things out. >> Yeah, that's a really good point I guess like just Open Flow as, is, is really an enabler it, it gives you the, the ability to think about solving the problems in new ways but doesn't. Doesn't necessarily give you the solution, it gives you a way of, of coming up with new solutions. >> Exactly yeah, yeah and it, I think once we help people see that that's what we are trying to do I think. Okay now everybody can, can appreciate what the goal is and, and, and you know, sort of help us identify the, the, the both the problems and the solutions. And turn people, turn people loose working on things rather than you know, being afraid of this new thing and not really understanding what it is. >> Yeah that makes sense. So I guess, certainly you've had a chance to, to sort of get familiar with open flow, see it in many different vendor switches, et cetera. In your efforts to [INAUDIBLE] whether or not it can solve problems in the campus I'm wondering what you think. [INAUDIBLE] Of course it's probably worth mentioning that we don't have Open Flow widely deployed across the campus yet and I'm wondering sort of what you see there in terms of like futures, so. What are kind of the major hurdles that you see in terms of like, Open Flow deployment in a large campus network? [INAUDIBLE] How real is it and what, what do you see [INAUDIBLE] are sort of the obstacles and also, I guess to the flip side, what are the major opportunities, I mean where are like, the real pain points in the campus, where. If we could sort of clear these hurdles you know, we could really do some really you know interesting, or good stuff in, in campus network management. >> So, ultimately, so you know what have we done so far, [INAUDIBLE] it's not wide scale. It's in several buildings, it's in several labs. we, for the past couple of years, I've referred to it as, as lower case p production. That means that we have, we have real projects, you know, real production use of it. People want, you know, people use it for their day to day work. It, they use infrastructure that's based on Open Flow. We have ongoing installations of, of different t-, different projects and different, you know, like video conferencing and things that use it. And we also have some, you know particularly work in, you know, security and, and using flow, Open Flow for, for monitoring types of things. Rather then, then you know in the heart of the, the traffic stream if you will. The heart of the infrastructure. As to you know, why is it just that, why you know, why haven't we gone further with it? Definitely maturity of the. Of, of the offerings from the various vendors, has been a impediment to that. We can't, you know, we have several hundred thousand ports on campus. We can't, you know, and thousands of switches. We can't go in and change those. All at once. And, so as we you know, as we make new investment as we move forward. We're looking to adding the, the capability in, in new switches that we buy. Looking [INAUDIBLE] working with the vendors to, to, to get that in place so that we can start to do. To take some of these, you know, things from lowercase p production to capital, capital P when, where, where we can, you know, where we have the opportunity. >> Yeah. That, that makes sense. You mentioned a lot of different, you know, places where. Where Open Flow's been deployed in, in a lowercase p production. You know, security and, and, and, you know, video conference applications, etc. Another one that seems certainly like a huge network management problem is the, is the management of the wireless network. On the campus that I know you and, and, and your colleague Matt Sanders have, have designed and have really extensive knowledge of the, of the wireless network on the, on the Georgia Tech campus. So I was wondering if you could just talk briefly about what problems come up in wireless network management, in terms of like performance, troubleshooting, security, and et cetera. Do you see SDN playing a role in, in wireless network management? Certainly we hear about it but, but, it's another one of these things were you kind of just hear about but you don't hear, like what's really going on in the trenches. So? >> Yeah, absolutely. So the first thing is, you know, just recognizing the reality that. Wireless is becoming more and more the, the primary way that, that our user's access the network. You know, the, the, the bulk of new laptops that people are purchasing and bringing to campus, don't even have wired ports anymore. They want portability. The, you know, we have lots of classrooms with wired ports in them that no one uses. Because they want wireless. Right? They bring their, their tablets. They bring their, their phones, smartphones, etc. and that, and of course, you know, it's not one device. Right? People are carrying two or three. Not to mention, you know, gaming platforms and those kinds of things. So,. And it's not just students. It's faculty it's staff. You know, it, it's, it's, more and more of the, the end user side of things. You know obviously servers are still you know, in racks that are wired together. But, but from the user side, wireless is the, is the, the norm now. And so, what that means is we have to support. We have to do everything we can to support that user experience. Including things like dealing with, with policy you know, how do I access, how do I get access to my network and my my lab resources and my printers, and,. All those sorts of things that, that people want to be able to do. That they want it to work just like it did on the wired network. But, but now it has to be wire, no now they want to get it to it from wireless. And so, challenges are things like how do we support the, the portability that want they know, they we want. We basically have. Taken the approach that the wireless network should work the same across campus. That means that my, you know, essentially today that means multiple VLAN's that span the entire campus. It's that can be a, a painful process. >> Right. >> You know a painful thing to support for you know, you know 5000 access points and. All on, all on, a handful of VLAN's, so you know dealing with broadcast, dealing with multicast, dealing with the protocols like Bonjour and what have you that. Work great in the home, you know, people expects you know, to be able to, to plug in an Apple TV and, and just broadcast on it,. because it works at home, but the reality is that, those things don't scale to the enterprise. >> Mm-hm, mm-hm. >> And because that's not how the protocols are designed. And so, when we're dealing with, you know, we're, we're, we're looking to SDN as an opportunity for us to build some of those,. Solutions to fix some of those things, and make the you know, both, still have the benefits of, of portability and movability, you know, roaming all across campus, yet you know, give us some, some isolation so that, that we aren't, don't have to broadcast every packet everywhere in order to do that. So that's kind of a, a, you know? A starting point, I guess. >> Yeah. That, that, that makes sense. I mean, one of the things that, that, that SDN gives you, of course, is the ability to kind of like see the, a lot of network states and managed network policy from one place. Like, you know? [INAUDIBLE] A logically centralized controller. That basically can control like many switches or devices etc. across, across the campus and it seems to me like certainly some of the things that that you talk about involve you know, or, or maybe could be easier if you could see the state of the network from one place or manage a lot of distributed things from, from, from from one. Or a few central controllers. I'm wondering if there are other kind of network management problems that, that you think are worth mentioning. You know, such as like, you know, campus firewalls, or, or VLAN's, or anything like that. Where, you know, the, the types of, or the architecture that SDN gives you could make some of those network management problems easier. >> Yeah, so I think, our traditional, so, our, our sort of traditional problem right now is, is, you know, policy management issues, things that arise from policy management, and the fact that. You know, our, our tools that we have today are pretty blunt instruments. VLAN's and Firewalls and, you know. >> Right. >> Sub-netting of networks it just it this is, this is a legacy right? You know, using what IP sub-net your on as, as your policy container is, is, is just not practical in a world where. >> [LAUGH]. >> People don't want to think about what IP address they have, because they want to use wireless for instance or, or what have you. >> Right. >> So the, you know, we don't want to be limited to that. So it's, it's, [SOUND] I'm you know, we, we're very much interested in,. Things like how do we reproduce the same really auditable security solutions that we have so that we, you know, we can still meet the, the policy and the audit requirements for things like. You know, protect the data. And you know, whether it's health records or, you know, student records or what have you. or, or, sensitive, you know, secure research. All of those things have. You know, procedures in place today, whether it's physical isolation or VLAN isolation or various forms of, you know, authentication and what have you. It would be, you know, as we get into more and more demand for this flexibility and, and mobility. We really need again, you know finer grain control. Better tools that let us describe policy and then implement policy. And then actually go back and be able to audit it and say yes. The policy we'd, we, we you know, declared here is actually what's being inforced. And yeah, it you know, those are the tools we need. And you know, we're working on some of them, but we need a lot more. >> Yeah, it, it definitely seems like auditing and testing and things like that are, are something that certainly, you know we've talked about it's network management problems for a really long time. And you know, maybe there's some hope that, that, that, that STN might have something to say. There that it seems to me, like, one of the things I think is interesting is that, STM turns a lot of, sort of, domain specific network management problems into software problems, or, or, distributive systems problems. And maybe by turning [INAUDIBLE] the, you know, the. The networking problems into things where, you know, people have studied the, the problems in other domains of computer science, you know, maybe there's a glimmer of hope. >> Yes, I think so. Certainly it's, it's, what, you know, keeps us working on it, we think we can solve some of these problems now. >> Yeah. So, I wanted to talk a little bit about things you've been up to lately. I mean, Certainly, there's been a lot of attention of [INAUDIBLE] of SDN and data centers. But there are also other opportunities for, you know, SDN to take Hold in, in, in a variety of other domains. And, and one of those is in the wide area. Like, we've seen a little bit of. You know? Google saying that they've deployed SDN in the wide area. I know you've also been sort of exploring the deployment of SDN at, at an internet exchange in downtown Atlanta. And I was, I was wondering if you could just mention, you know? Talk a little bit about that. I, I know it's, it's, it's fairly early stages. But what do you think are the, the applications of SDN. An internet exchange, and, and also what are the challenges? What, what, what types of things make that a new and interesting place to look at SDN? >> So yeah, we have started to and building out a actual deployment working with some folks at CoderAtlanta, and Hunter Newby at CoderAtlanta And building out a, a real peering location we've called the Snap and based on you know, leveraging SDN as part of, part of the solution. All right? It's, it's, it's a traditional peering operation, but it also. Has this goal of exploring SDN as a tool for peering. And I think the, the starting point for us is, is we're not trying to replace BGP. We know we're not trying to start with let's do a feature for feature make sure we cover everything that BGP does. In fact,. We're kind of interested in doing things differently from, from BGP, and from perhaps taking a different view of what it even means to do peering. And so some of the applications, I think that it, at this point, having the most traction are perhaps the most interesting, is looking at the the you know, grilling needs for cloud. Enterprise cloud usage, basically, as, as companies are looking more to leverage the cloud, not, [INAUDIBLE] running all their compute and storage infrastructure with themselves. How do they do that in a way that they can, you know, feel comfortable not just about. You know, and basically not turn it over to the, the wild west of the internet, in terms of, you know, where do my packets actually go, and where does my data end up. And what priorities do they have? And ultimately. People want to be able to, control where their traffic is going and, and you know, directly you know on a per application basis. And I think that's the, the tricky part and that's where, you know BGP doesn't understand applications. And we, we want to be able to look at, you know, you know whether it's, you know, video distribution or you know, access to my HIPAA data. How am I going to control, who, who, who has access to it, where it goes, and, and what the priorities are? and. For it to be real, for it to be truly a multi-vendor, multi- you know, multiple cloud providers, multiple customers. we, we have to have some, some better tools for managing policy at the exchanges. And I think that is probably you know, there's other applications like time of day routing, and other things that we sort of started with, but I think, I think the one that sort of has the, the staying power and really some interesting problems to solve is around this, this enterprise cloud space. >> That's pretty interesting, yeah, I mean definitely there's been a lot of talk of, of SDN for, for you know for enterprise networking and also for integration with, with the cloud so, so that definitely seems like, that's a, a right space to look at. How do you see the. The exchange as playing a role in, in terms of like the enterprise linking up with the, the cloud. I mean, obviously, it's a place where, where networks come to, to connect with one another. But, do you see that the exchange is ultimately going to become a place where there are additional value added services you know, you know, that might be offered by the exchange itself? In support of these types of cloud, applications for Enterprise. >> Yeah, so certainly you can think, imagine things like you know, caching and [INAUDIBLE] distribution that goes on you know, you know, some of the architecture's the, the Netflix, and others are doing a pushing caches into the. The ISP's and and I think, you know, the exchanges are potentially a place to do that. But I, you know, I think I'm starting with the fact that you know, right now an exchange you know, there is no API's if you will for an exchange, right? >> Right. >> You know, I, I have a contract, we, we can figure some BGP stuff and then we, we, we keep watching to make sure it's working, but. There is no you know day to day or you know, certainly hour by hour, minute by minute control I can, I can ask for in my relationship with an exchange. You know, it's certainly, it's certainly has the potential to, to greatly change, The you know the business relationships between partners at an exchange. If, if I need to, not just share traffic, but share access to an API of some kind or controlling flows, right? >> Right. Right. >> but. So I think you know that's kind of where I'm thinking things are going to get interesting pretty quickly in order to make some of this real. >> [INAUDIBLE]. >> So in terms of just, I mean you mentioned also the role of SDN and supporting the higher applications you know maybe in this exchange cloud context but, also I knew already you've been. Involved in using SDN to enable certain types of applications like immersive video conferencing. To give something called the Magic Window. And I was wondering if you could talk a little bit about, that, the Magic Window project. Like what is it, how does it do you know, what does it do, how does it work? And like how does SDN help you solve some of the challenges involved in, in developing something like immersive video conferencing? >> So selfishly, so [INAUDIBLE] Magic Window's a fun project. And selfishly the one of the main reasons I like it is I've been in networking my entire career. And every time I've demoed any cool new project I've worked on, the main tools I've had for demoing were ping and tracer out and that's pretty lame. >> Yeah, yeah. [CROSSTALK]. >> Yeah so this is kind of fun. To have something that, that people can relate to the goal. People can and, and people can see when, when we make a change in, in the flow table, or what's, you know, what's happening in the network, you can see, have a visual you know, indication of, you know, improved performance, or. Enhanced augmentation. So Magic Window is a, a immersive video of ex- experience. It is, we use a, you know, a big screen fifty inch or over fifty inch TV in most installations and a camera and, and some sensors that allow us to do. A gesture based interface, that lets you you know, using your hands, using your body movement. As you get closer this, the, the, the image zooms in. As you turn your head left and right, you pan around the room on the other end. And. A big part of what we're trying to do is turn the control of the experience over to the viewer, not the presenter. >> Sweet. >> So all you folks out here watching this video, you know, conference right now. Maybe you'd like to look around the room and see what the heck else does he have on that wall. >> [LAUGH] >> You know, that Georgia Tech thing. I wish he'd move his head. You know, in the Magic Window you could do that. and, we're using lots of, you know, different, you know, technologies Web RTC and, and a, you know, Forte HD cameras, and things like that, but the The sort of interesting part from the SDN side is, the fact that the other things you can do as a viewer are, you can ask for different augmentation on the videos. So you can, a, ask for you know, turn on highlighting of different features or, you know if you're looking at a,. A room full of different technologies like when we have a setup in a home. You can highlight. You can image in a home repair. You know, some, somebody helping you troubleshoot yer, yer home stereo system or something. You're wi-fi or, you know, you've called the. The refrigerator repair person, and their, their trying to identify which component is, is not working, >> I see, yeah,. >> and so you want to do overlays, you want to, there's some augmented reality aspects to [SOUND] it, but also, you know, sort of other, other metadata, other enhanced content that goes with it, and so. Because we're, you know, including all of that additional content, the question becomes where does it come from, and how do you actually scale this to a conference size? You know, or, you know, even a conference call with multiple endpoints. And SDN is, is basically the technology we're using to. Explore how do we actually do that? So putting con-distributing content around the network, distributing users around the network. Dynamically changing flows. In real time changing, moving a video stream so that it goes though as additional processor, to add an overlay. >> Yeah like middle box integration. >> Exactly, right. >> Yeah,. >> So those are the kind of things we are leveraging SDN. And again, you know, you can, you can, you can go through the entire demo, and the entire process without even realizing that's what's going on behind the scenes. But for those of us networking folks it's fun to actually be able to show off what the technology, what the network technology. Can do to make a user experience better. >> Yeah I think that's a really awesome example actually because i mean it's clear that there are many moving points right? I mean there's, there's as, as you, you mention's [INAUDIBLE] redirection through middle boxes. Integration with, with end points and cameras and sensors and things like that and. I think that's like really really interesting and, and sort of important to note. Because we, when, you know, right of the beginning, as you mentioned like misconceptions about SDN and Open Flow, you know, we think about, you know, Open, SDN is more than just Open Flow. But SDN also has to be more than just controlling routers and switches. I mean, we think about a network. You know, there are, almost as many or in some networks more middle boxes than there are routers and switches and those things are really like defining the behavior of the network and the performance of the applications that you've seen and so forth. It seems like the magic window's like a really cool example of. How can we use like, in order to use SDN to solve these problems, we have to think beyond just SDN as a controller for, for routers and switches. It has to do with all kinds of other things as well. It's not just enough to like have something that, drops some flow table entries in a switch. It's gotta route traffic through middle boxes and so forth. >> Yeah, so we're actually, you know we're, we're, we find ourselves actually using the term SDI more now. Software Defined Infrastructure. And recognizing that the real value is, when you you know include the, the middle boxes, the, the, the content, you know the, whatever else. And you know to your earlier point about. You know, never management challenges, and the you know, what are the, some of the hard problems we deal with. All of those middle boxes, right? >> Right. >> As network people, we focus on layer two and three and we make sure all of that infrastructure is working. And then, the application still doesn't work. And a big part of it is the ghost in the machine that people aren't aware of, >> Right,. >> That is causing, you know, perturbations of the, the stream, it, you know whether it's, its security appliances or. Caches that are misconfigured or not doing what we want them to do, you know. All of that kind of stuff ultimately affects the end user experience. In many cases, you know, more than, you know, are what kind of packet loss are we we seeing on the link. >> Yeah, yeah. That, that definitely makes sense. I was wondering, actually, just jumping back to the campus Open Flow stuff a little bit. Like that's, it relates a little bit to the National Science Foundation's, genie project. This global environment for network innovations. And I know you've been really involved in Georgia Tech's, efforts, to get Genie. Deployed on the campus. And, and, and, sort of just that, you've been very involved in the Genie effort in general on it. I was wondering if you could talk about the relationship of Genie to, to Open Flow and Software Defined Networking. You know, how, how does this sort of like. NSF's desire to create a future internet or an experimental test bed, what does STN has to do with that if, if anything? >> yeah, and, and it's another one of those demystifying things right where there seems to be some confusion about. So is Genie Open Flow? or is it the same or what, what is what? and. Genie is, so the goal of Genie is to build a platform for people to do experimentation at scale. All right, so get it out of just your lab, or get it out of just simulation, and let's, let's actually test out our new ideas and our new. You know, our new protocols, our new application architectures. And actually do it at, at internet scale, at really global scale, with real delays and real loses and, you know, real users and all that sort of thing. So, The goal, and the goal of course is to be able to support more than one experiment at the same time, right? >> Right, right. >> And in order to do that, you have to be able to do, slice it some way so the term we use is slicability. Of this, this experimental infrastructure so that I can create this you know, what it, what, you know might think of as an overlay. A, a topology. Some subset of the larger set of resources and I can allocate that and create a slice and, and call that my experiment. Platform. My experimental test bed and I want my experiments to not be impacted by, you know I want to have those resources allocated for me. I want it so that you know other people's experiments don't, don't you know interfere with mine because i want to build, you know every producibility and, that kind of thing. But also, I want to be able to do things that might cause damage, right? That might be bad, and that they won't let me do on, on a production network. And so I need my traffic and my stuff to be isolated from other people as well. So it turns out that SDN. And Open Flow is one of the primary tools that Genie is using, to implement that. To pull this off. To, to create an environment that is physically shared by all of the researchers and deployed across the country at university's and network backbones and, and exchanges. And then to slice it up for people to do experiments on it. So Open Flow, and SDN, and specifically Open Flow are keys to, to pulling that off. And if nothing else, you know, SDN plays a critical role in, in, in making Genie work, right? And then, on top of that, the, the, the researchers, the experimenters that want to actually do their stuff, right, that they actually want to run their experiments. >> Yeah. >> They may or may not want to use OSDN slat and Open Flow as part of that, you know, in their experiments. So, in a sense. What we, what we're, we're creating is an environment where we use Open Flow and SDN to create this infrastructure. And then we allow people to, do their own thing with Open Flow and SDN on top of that. [SOUND] you know, it's a little bit tricky in implementing, how you actually pull that off. But and you know. Which switch and which controller is doing what, but, >> Right, Right. >> It, it's, definitely a big part of Genie, both in making Genie work and the things people want to do, you know, SDN is a hot topic right now on networking research. Lots of people want to do SDN related, you know, projects at scale. And, and Genie is a place to do that. >> That's, that's kind of cool. Because it, you know, the, the same technology that's being used to support the experiments, You know, is, is also being experimented with, so >> Yeah, exactly, it's also, you know, obviously you know, it's, it's a little frightening, too. To To try and figure out how to make it work. >> [LAUGH] Yeah, if you want it to actually work right it's a big target so. >> Yeah, yeah, yeah. But that's why we call it research. >> Yeah, exactly. So, just, I just want to close with one final kind of big picture question which is, you know. Certainly, you've had a lot of experience with, with network operations and management. And also, with, with SDN and Open Flow I thought, you know, what do you think are the biggest problems that, that need to be solved in SDN before it can really take hold in, in the enterprise in the white area? Or other parts of the network. Do you think it's an issue of just. You mentioned earlier, just immaturity of, ofvendor support, but also there are other issues, like there's really no standardized northbound API, there's not a lot of management application, etc, seems like there's a lot of stuff that's pretty green. So what do you think are the, what's the lynchpin? If we can only solve these [SOUND] two or three things or one thing then, then SDN would, would, would And be deployed sooner rather than later. >> Yeah, I don't know if there's one thing that the, so, I. A part of you know, I'm, I, I talk about the education and sort of demystifying. I think, you know,. Probably the biggest challenge right now is, is, cutting through the hype you know? Every, >> Mm-mm,. >> vendor that calls me, you know, you know if they're selling, if, if someone's calling to sell me a car right? It, it supports SDN now right? >> Right, [LAUGH], yeah. >> I mean, just, you know totally unrelated you know, and, and. You know, it, they're, people are using it as a way to you know get me to take their call essentially right? And, and so I've, I've you know, any number of presentations where people weren't even, you know, it, selling of a network. You know anything below layer 4 right? >> Right. >> And, and there's this SD enabled stuff they've got. >> And it's basically just ramming more lines of code into whatever they already have. [inaudible] like we have it. >> Yeah And so it's I think you know, cutting through the hype, the crap if you will, I think is right one of the biggest challenges with SDN. And, and you know, helping remind people what it's really good for. What are the real problems that we're trying to solve and and doing that, I think you know i, i am encouraged by the fact that you know, this notion of SDI and the fact that you know, hosting virtual cloud. services. And, you know, virtualization of, of hosts and storage and all those things are really are a part of the story. You know, it's it with. It's a whole listing set of problems that we're, we're trying to look on here. You know, people are changing the way they think about compute and, and access to those resources. And you know, you know? I carry my laptop and then I leverage lots of other things that I don't even know where they are. And so, I think, you know, I think, I think that. That looking at it sort of more holistically and not just as a, we're going to replace your switches. >> Right, right, right. >> Right, okay, then what right? And I think that's kind of part of the challenge to you know, people saying well why haven't more campuses or more enterprises deploy Open Flow across their network and part of the answer is cause no one's asking for it right? >> Right. >> No one is really saying. >> Open Flow will, will solve this problem. What we need is is too look at it more holistically and say, you know, this new security model or this new application deployment model. >> Yeah. >> I'll leverage SDN across the infrastructure and- >> Yeah, here's like the killer app that you want that, that you can only have if [CROSSTALK]. You know, you don't sell someone a hammer unless they want to like, you know, build a house or something, right? I mean, you first have to figure out, you have to, first have to like figure out that they want to build a house. And then, and then SDN kind of comes along with it. >> Yeah, yeah, yeah. So it, you know, we're getting there. and, and we're. We're solving you know, problems a little bit at a time and, and building up the toolkit, I think that ultimately is what's going to you know, get, get us to the steamroll where, where its everywhere. But. We gotta, we gotta, we gotta cut through the hype, and, and, and not just buy SDN because it's the latest buzzword. We got to >> Right,. >> we got to make it work. >> Makes sense. Sounds like we have our work cut out for us. >> Yeah, hey. It's job security. Awesome. >> Cool. So thanks a lot for for taking the time to, to, to, talk to me today about, about SDN. We look forward to, to seeing what's coming next. >> Indeed, thank you.