 Okay, we're back here live inside theCUBE. This is SiliconANGLE's flagship program. We go out to the events. Extract the signal from the noise. I'm John Furrier, founder of SiliconANGLE. I'm joined by my co-host Jeff Frick of SiliconANGLE. Jeff, we're back here with Big Switch. Okay, Andrew Harding, senior director of product marketing of Big Switch. Was in Palo Alto, moved to Mountain View. Got a new team over there growing fat. Well, not new team, expanded team. You guys were in that wave of startups really from the DNA of the open flow. First of all, welcome to theCUBE. Thanks, thanks, good to be here. You guys were part of that, the early DNA of open flow. And your founders have that academic background, doing some seminal work around open flow, which became the catalyst for SDN, software defined networking, open flow is still around. And then Nacira got bought by VMware as you guys were getting your funding. And some just great stuff happening around SDN. At that time it was just like, okay, there was a couple of horses on the track. You guys were one of them. And all of a sudden Nacira thing kind of woke up everybody outside of the vertical of SDN. So give the folks out there a quick overview of Big Switch, kind of where it is, where it's come from, where it is today, and what you're doing here at OpenStack. Sure, I think you covered it pretty well. It's Guido Oppenzeller and his work heading the open flow work within the Clean Slate Lab. And that transitioning from university to commercial product over the last year is really the exciting stuff that's happened recently. And last summer was crazy times. And since then we've launched the open SDN suite, our product suite, and had, I think, more exciting product announcements than work in research in the last six months. So talk about what the Nacira acquisition has done for your business. Obviously, you guys, that's a competitor to you guys. Now it's VMware. And how has that changed your business? And how has that changed the market? Because OpenStack really got a big lift from SDN. A lot of code came in, a lot of awareness around where this market's going. The data center obviously has one big impact for OpenStack. I think to folks who are already inside of SDN, the Nacira acquisition was almost a noise that woke everyone else up. So we- Like Cisco. Lots of folks got a snooze button going off. And I think that was a point in time that really saw the incumbents or the legacy networking vendors waking up and saying, okay, this is real. For Big Switch, it was exciting in that the world is now recognizing SDN. And honestly, nice to have clarity around who's the leader in standalone SDN companies at this point. So what are you guys doing here at OpenStack? What demos, if any, you're showing? And what are some of the things that's happening right here? So there's so much good stuff going on this week. We've got a demo of our Switch Lite software going on down at another conference. And today here at OpenStack, we're demoing our proposed blueprint for service insertion and service chaining. This is actually based on something that you can do today with Big Switch's products. We've taken that and added it as an OpenStack blueprint for adding services like load balancers or firewalls and chaining them. You can chain them as Layer 2 service. You can chain them as a bump in the wire or as a network tap or as a Layer 3 service. We can do this in a way that enables the network operator to provision service change and then allow tenants in their network to choose which services they want to cobble together into a chain of services that allows them to do see an application server or a web server and then load balancing and firewalling right in front of that in a very simple manner. And we're excited about this contribution. Okay, so Andrew, break it down for us here. Okay, on the open source contribution. Obviously you guys have a big presence and a big part of your company is open source based. So one, break that down for us for the audience. How much is open source of Big Switch? And then the big theme here is vendor neutral, right? So people want choice. That's a big part of the OpenStack Foundation. So talk about what you guys are doing with open source, what part of your company's open source, if all of it and what's not open source and then talk about the kinds of choices people can get with Big Switch. Sure, so this is a bit of a long and winding tale. So we started an open source with Floodlight which was based on Beacon. These are open source SDN controllers. Project Floodlight has clearly been the open source controller that emerged as a leader and that's the core and that's what our big network controller, our commercial product is based on. We recently proposed that as a contribution to the recently announced Open Daylight project. And that Open Daylight Consortium is one place where open source is happening. OpenStack is another and I think our most recent contributions here is this blueprint around service insertion and service chaining. And the really exciting thing about service chaining in OpenStack and with Big Network Controller is that it's something you can do today. Service chaining and service insertion, people have been talking about for, I mean, years. We've heard announcements from other folks that they'll maybe ship it next year or the next year. In an open system, it can be done today and users can just choose the one that works. So we had a comment on Twitter 22 hours ago. Scott Lowe tweeted, NTT agrees with Big Switch. More plugins creates customer choice. What is he referring to there? You know? I think he may be referring to our quantum plugin that enables folks to provision virtual networks within Horizon and have those networks rolled out on our Big Virtual Switch product. I'm guessing here. There's been a lot of action in open source and in shipping software here in the last couple of months. So one of the conversations out there in the keynote, which we found compelling, David Floyer was, our analyst was on earlier from Wikibon talking about it was Bloomberg. One of the examples was talking about the innovations happening underneath and above OpenStack. Obviously you guys are playing down underneath OpenStack. What innovations are going on right now that you can point to that you're involved in that's lifting up and filling in the white spaces of OpenStack? I think putting service insertion or service chaining into kind of OpenStack proper is a critical innovation. This has been a tough problem in networking for years. You know, it's one reason why people have been in networking or network security have all these boxes sitting around the network. These very, you know, the hot box here, some box from one vendor, and maybe only one box is from a single vendor because you can't mix and match anything with proprietary systems. Doing multi-vendor service insertion that can use the best in class security device. Why has it been a hard problem? It's tough to get the incumbent networking vendors to cooperate. You know, I maybe have some opinion there, but to focus on the positive. Well to focus on the positive, what's the delta now? So what did they have to do before and what was the scale and scope of the effort? And now what does this enable? I think it's communities like OpenStack. Smaller communities like Project Floodlight is focused on a problem in networking and SDN. It's communities where code is the currency, where technical innovation that goes into the project is the currency. I think that's the real change. It's no longer an environment where you can write a spec, propose a protocol, and then just wait it out because if folks want to get it done, if they want to really solve the problem, they can do that with code. Yeah, no, it's probably a different twist on the question. In terms of the business problem and the issue that the service insertion solves, what are those guys doing today? Or what were they doing before they had this option and how's it going to change their world in terms of the guys that have to actually implement and execute all these processes? In the past, service insertion was complex and fragile cobbling together of different devices and so you would need to build up a static system and in one place make one change and then test that change and then replicate that change and there's a very static and manual system by putting it into an open stack and by putting security devices or application delivery controllers in the network and doing that programmatically, we can take what was a manual process and automate it and that's really the goal because in networking specifically, around networking there's been a lot of automation that ends with a manual workflow on the network. So all the benefit of automation that's occurred in compute is diminished when you get to a legacy network where the workflows have to be manual. By adding service insertion as an example as a part of open stack, that can be automated so people continue to get that business benefit all the way to load balancing and firewall or network tapping, other services they might be putting on the network. Okay, great. So talk about the O&S summit that's going on right now, the O&S 2013, you guys are also down there. Your CTO gave a talk, we just saw that come up on the radar. What's going on down there? Can you talk a little bit about what you guys are showing at O&S? Yeah, so Rob Sherwood just gave a demonstration and a talk of our Switch Lite software running on physical switches. Actually I haven't seen it yet, I saw it in a dry run but I've heard that it went very well. I think it's really exciting. It was called on Twitter a fluffless demo. There you go, it was a real fluffless demo. I don't know if that's a compliment, it sounds like fluff is not a compliment. Sometimes I'm worried that networking folks and compute folks still don't talk to each other because we have these things the same week and so you guys know more about this demo we just did than I do. I saw it as we finished it up last week. It was... We don't miss a thing on SiliconANGLE, we know we're all the actions, we got our lackeys out there getting all the data but in all seriousness there's two shows going on. I talked to Jason Metloff over there, his friend, his VP of marketing and he was supposed to be here but you have a contingent down there. We swapped, I came up here. I'm a fan of Portland. A better beer. Good beer here in Portland. Beer, beers and bikes. So given the compute obviously converging infrastructure is a big theme that's been kicked around, even though HP kind of coined the term but David Floyd from Wikibon, we call it software-led infrastructure is certainly not defined. Compute storage and networking is the core tenets of the OpenStack Foundation here, they're rallying around those three elements. Our SDNs, a lot of work needs to get done there. Software defines being kicked around. We saw HP talk about it with their moonshot announcement, software-defined servers and then you got software-defined data centers, it's kind of like the moonshot, a little bit marketing hype right now but what is your take on that? Because now the interaction between all elements need to be in place. What is software-defined, how do you define software-defined these days? There's a little bit of software-defined washing as we would say going on and people slapping some stuff on it but for the most part, is it just software-programmatic software? How are you guys looking at it? I know you guys have more of a purist view being at the networking level. You know, I think we're the purists and the pragmatists coming together. The purist view that came out of the research has evolved as it's inserted into production networks and software-defined networking is really about making the network programmable and that's different than having an AP on a device that can make that single device programmable. Software-defined networking is about having a physical infrastructure and then a logical error on top of that that enables tenants and operators to see two different things and hopefully enable network operators to do less things manually, to do less things in a static way because that increases costs, that increases the likelihood you'll have an outage. So I think software is in all of these systems. It's how that software is used that's the critical difference and one change in the last six months is that there's a clear contrast between the SDN washing and the folks who can get up on stage and do a fluffless demo of real switch firmware working today, the Switch Lite stuff. I don't want to make too much news here, but the Switch Lite stuff is doing great in production. We had planned beta in June and it looks like we may come in a little early actually, so we'll be shipping that before the planned beta in June. It's not a free download yet, but it'll be there before the beta in June. Okay, so let's talk about the competition. Obviously, Nasirah got sucked into VMware. I interviewed Martin at VMware. He was very candid and he was talking about his vision and what he wanted to do to put a dent in the universe with the innovation. How are you guys competing against, say, now VMware? And what are you guys doing in particular? And you don't have to talk about their negatives, if you can, if you want, but you could also talk about what you guys are doing differently than them. You know, I think when I look at MVP, MVP made the leap from academic research. We started the same place in Stanford, same lab, to being inside of a big company and that happened in one step. There's some stuff that has to happen in between those two things to make it a real scalable, deployable product. This is the stuff that we're focused on and we're focused on these things without the distractions of being inside of a company that does things in addition to, in addition to SDN. So we've maintained focus on open flow. Martin has different opinions on open flow in different years. We've held the focus and stayed on target with open flow and we're working through what has to happen with the product to take it from university research to real production deployments. There's issues with tunnel scalability that have to be worked through. There's issues with gateways in the network that are software based that have to be worked through. This is the normal course of any product as it goes from these very early deployments to large scale production deployments with the very early adopters to the folks who are wanting to get the advantage of SDN but also wanting the safety of a product they can go to and count on. And I think a focused SDN product can address these natural issues of any evolution of a product more effectively. So Cisco is pumping a lot of investment now. Obviously, it was a wake up call for them. They were one of the ones that hit their bell rung a little bit on that one, especially with the VMware and the VCE relationship and that's had its own little shake out separate conversation there. I don't want to rat hole into that politics of it but obviously they're the networking leader and some are saying a laggard in the sense that they've kind of rested on their laurels and SDN has looked at as an innovation opportunity. Now they're gearing up. What's going on with Cisco and how are they being perceived in these communities with open source is a big part of it and how do you guys compete against Cisco? Yeah, I think we compete against Cisco by having an open SDN offering that folks can get today. If folks want to get network monitoring via SDN they can get it today. If folks want network virtualization it's available today. And if you want switch firmware that can run on a Broadcom platform switchlight for Broadcom that goes into beta in June I think we're actually going to have it in May. And we are going to change the network and get the same kind of automation and time savings that comes to automation the same kind of operational cost savings that folks have gotten in compute. It's a very kind of core of what OpenSec is doing. We're going to bring that to the network. And I don't want to trash Cisco. You don't need to. I think they're licking their own wounds a little bit but the signs we're seeing from Cisco is kind of like as we say the white smoke coming out of the chimney. The pope has been elected it's called SDN and that's cool for them and that's a good sign for Cisco. The question is can they move there fast enough and with the open source movement scale out open source really dominating the innovation and the developer community. It's a challenge. It's a challenge for a large company that has a legacy business to preserve and for any company that doesn't really have a history of true innovation in software and the way that it needs to be done in SDN. This isn't implementing yet another- How does it need to be done on SDN? It's not implementing yet another protocol with OpenFlow you don't need to implement yet another protocol in the device. With OpenFlow you implement the protocol via OpenFlow and extensions to OpenFlow and some other protocols as well but this focused effort to deliver an instruction set for the switch so that we can extend the network from a controller is a fundamentally different model and I think we see the degree to which folks get it or don't get it by the degree to which they have embraced OpenFlow or try to have yet another proprietary interface to every device be the panacea. One of the themes here obviously at OpenStack is because you got the pure cloud guys is DevOps. Which we love we have a site called devopsangle.com. We started two years ago before anyone kind of knew what that was we saw that trend. So I don't want to get your opinion on DevOps I think that's pretty much a done deal people see the benefits of abstracting away the complexity and giving developers freedom to push buttons and know through all their greatness above on the top of the stack. What I do want to ask you though is infrastructure as code is a term that's being kicked around more in mainstream on mainstream in enterprise and service providers as a viable philosophy right? Prior to this year infrastructure as code was kicked around by geeks like super geeks in the trenches who understand programmatic software based systems. What's your take on infrastructure as code and where do you see it going? You know I think the old way was to have devices as infrastructure and golden fixed configurations as infrastructure or parameters to that infrastructure and it's a natural next step to think about wait these configurations could be dynamic and so the thing that enables that dynamism that could be the configuration of the system underneath and if these configurations are dynamic and the infrastructure becomes that enabling piece that changes the configuration you can stop thinking about the world in a device by device manner and that's the fundamental shift and I think that really is a generational shift or maybe it's a shift of folks who you have to have the mindset that came out of software and not out of network protocol and network protocol configuration. Static devices, firewalls, switches. Because the safety was in the old way in fixing the configuration and knowing that fixed configuration wouldn't change. And the scale equation changed too, right? I mean so you had X number of machines, thousands of machines, now you got millions and you got virtualization. And no matter how smart the network engineer is and no matter how fast they type and how well they happen to know ILS they can't type as fast as code can execute and that's the fundamental change. That's a good one. Well the other thing that begs is especially with the scale, can an individual guy or team trying to innovate compete with kind of this open model where you've got a large group of contributors coming at the problem from all different ways. And coming to like yours it's so leveraging the open source community in the open stack and your other new foundation. Can the legacy guys keep up with that innovation? I think our contribution to the recently announced open daylight consortium. Open daylight that was the second one. Project for Light being the place where our open source controllers developed. I think that is a recognition that a single vendor's contribution can cause a whole industry to move to a more collaborative model. And this is again I think a generational shift rather than being hyper competitive about this protocol or that protocol in a device we can collaborate in a community that's contributing code and have it truly be a meritocracy. Whether it's project floodlight or the open stack community or something like open daylight that's brand new. If it truly is a meritocracy then code wins, the best implementation will win. Code wins. I mean I love that philosophy here bring code to the table. Let's talk about use cases. Obviously one of the things that we talked about this morning that in looking at yesterday, reviewing yesterday was the vendor-less hype here and really focus on the code and also the users, right? You see you have a lot of great presentations from people putting out implementations, proof of concept and actual deployments of stuff. Open stack related projects. Yeah, still a lot of work to get done. Talk about big switch. What do you guys can point to right now where you can share in terms of people implementing big switch, open switch. Big switch and what your code and if you can do it on an open source developer basis then take it to a user. So there's, we recently announced there's a nice sort of momentum announcement around project floodlight and the folks using that from the Stanford Research Institute to some folks in the Midwest using it for a security application. Our customers right now are seeking to gain that strategic advantage of SDN and have been focused on rolling, making these deployments roll out successfully. Well, what are they doing? Can you name names? Yeah, we're not naming names right now. There was some news yesterday where someone leaked a large deployment of BigTap but our focus right now is helping these customers gain an advantage in their business and we're not talking broadly about who that is. What did they leak? BigTap? BigTap was leaked. I think Nick Lippis leaked it yesterday. So it might not have been Nick. I shouldn't paint him with a brush. I think it was on his website though. He linked to it. He amplified it. So I think the advantage that folks are getting from SDN is something they wanted to hold close to the vest. We have very large deployment of BigTap with a technology provider and that's the one that if you keep typing there you'll find. And the folks that have been on our testimonial page, those are the folks we're working with. So they're there on the website. So what is the big use case? I mean if you could point to like, obviously where you guys have a group swing on it can you point to a specific use case or case? There are two very clear use cases. The first one is network monitoring. Network monitoring in a traditional environment is very static where you do port mirroring or you have taps and you aggregate them in one place and this is kind of a fixed system which means that if you need to see traffic that's over here with a tool you have over here there's not a good way to bring them together. By putting an SDN layer right in the middle you can now get the traffic to the tools and kind of break the silos that have emerged in large data centers because as traffic goes more east-west and less north-south leaves the data center it's actually harder to get universal visibility unless you put every tool everywhere and you have taps everywhere and aggregate them all it's harder to get visibility. So BigTap enables folks to get the kind of visibility they need in a very large scale data center especially when you've got east-west traffic. So that's one, using SDN to give visibility into the network. The second one is network virtualization and that's really the focus of our work with OpenStack. With service chaining you can actually add tapping in BigTap. Network virtualization enables a provider to spin up services or to enable users to spin up those services themselves and with service chaining users can even now as long as the blueprint's accepted we get it into OpenStack. Users can with OpenStack can say hey I want to have a service here and I want to spin up a web server or spin up an app server and then I want that to be firewalled or I want that to be load balanced and they can do that on their own. So it's not just automation with network virtualization but it's delegation of that all the way out to the end user. Great, well you guys have been following Big Switch for a while. Obviously when you guys were in Palo Alto on Caliber Street, I think I walked in and sat down with Kyle there and one of the founders was Kyle. Yeah, and just, hey I was just visiting EMC Ventures and I popped over and walked in and there was no admin, I just started talking to him and at a great constant I was, that was two and a half years ago when you guys just moved in. Well we have bigger offices now but we still welcome visitors. Big Switch and you know Jason's a great guy over there our kids go to the same school and your VP of marketing, great guy. So I want you to get the last word on this segment before we end and just share with the folks bottom line them your presence here at OpenStack and what you guys are going to do as you go to market and how you guys are going to differentiate going forward as a company. So I think the specific to OpenStack one milestone on that journey is I think we actually have more folks from engineering here than we do from the commercial side of the business. So we've got folks here who are really really contributing and proposing blueprints that will drive this forward with real work. There are those of us who wear golf shirts with a logo and we do like a little bit of hype but we want this to be about the contribution to the community and then the contribution that community can make to the overall improvement of how folks can run data centers. For the company, I think it's really about taking us on this journey from we're way past research, we're in that time when the product is meeting the needs of the innovators and the folks who really want to gain a huge advantage from something new in their network and the next step of the journey is to get it to the folks who want the advantage but also want the safety of knowing that it has been rolled out before that the product is baked and that's the journey that we're on with Big Tap and Big Virtual Switch. There are products folks can get today and Switch Lite is going to be in beta in June maybe a little bit earlier because it looks like we're rolling a little bit better than expected there. Okay, Andrew Harding with Big Switch here they're also got the ONS in San Francisco and you guys are doing great. Let's see how the market shakes out you guys are competing with Unopened Source and that's a good thing obviously that market's working. This is theCUBE, our flagship program here at OpenStack Exclusive Coverage of OpenStack Summit 2013 Portland, Oregon. We'll be right back with our next guest after the short break. Thanks guys.