 Live from the Moscone Convention Center in San Francisco, California, it's the Q at Oracle Open World 2014 brought to you by headline sponsor QLogic with support from HGST, violin memory and MarkLogic. Now here is your host, Stu Miniman. Hi, this is Stu Miniman with wikibond.org here with SiliconANGLE TVs live coverage of the Cube, Oracle Open World 2014 here in San Francisco, California, fifth year of coverage at the show. We go out to all the big enterprise show, extract the signal from the noise, find the best guests that we can and excited to talk a little bit about networking in this segment. Joining me first time on the Cube is Steve Garrison, VP of marketing from Pika 8. Steve, thanks for joining us. Stu, I'm really excited to be here. Yeah, so Steve, you've been in the industry for a while, well known to a lot of us in the networking space for a long time. So why don't you give us a little bit of background as to who Pika 8 is and what brought you to them? Well, that's a great question. So Stu, you're right. No one wants to ever talk about their age on camera, but I have been in the industry long enough to know where there's a good trend and where there's a good opportunity and I think software defined network is one of those trends. So let's talk about what Pika 8 does first. So first of all, you have to understand the networking industry is vertically integrated today. That I mean just like a mainframe, you buy an entire system from one vendor and Cisco Systems is the poster child for that model. Now what we want to do is break that model apart and sell a modular system where you have the switch itself, the physical box separated from the operating system, separated from the configurations or applications the customers are using. And guess what, Stu? We've seen this model before. We look at the IBM mainframe to PC model Microsoft became one of the dominant operating systems. Right? On the server side, Red Hat became one of the dominant operating systems. In the mobile phone, we can argue whether we're Android or iPhone bigots, but they're both operating systems. And the same thing is happening in the networking. So our first claim is that we're going to be the dominant operating system in this new mantra of what we call open networking where the customer buys hardware and software separately. And then the second idea is we're trying to usher in this idea of programmability or policy-based usage. And again, your mobile phone is a perfect example. You click on an app and things happen. You don't know how they happen, you don't care how they happen, but in today everything's very manual on the networking side and you have to still basically program the switch for every application. It's very time consuming, very costly. All right, so Steve, I think one of the main things let's talk about is how the supply change is changing for networking. So yeah, absolutely. You said Cisco has dominated this market for a long time. And of course, most of its chips, Cisco makes. The operating system, Cisco makes. They build their entire stack and they've done quite well for a long time. And Cisco's brought a lot of good things to the industry. As I've said in many times, Cisco solves customer problems and then they try to make that the standard. Now, of course today, things are changing. Open source in many ways is replacing what used to be done in the standard. You've got chipsets out in the marketplace that many of the networking vendors have taken over the years to build their own boxes and operating systems. So tell me, what are you seeing in the supply chain that's changing in networking? I tell you, Stu, you said it right. First of all, I see a lot of excitement. And why is that? Because everybody looks at what Cisco has built and said, holy cow, there's a 30 or 40 billion dollar pie that I can take a piece of and call it my own. So everybody looks at Cisco from a foundational point of view. And by the way, I grew up on Cisco Internet, right? I think we all did. And there's nothing wrong with the technology looking backward. But looking forward, it's stuck. And before I answer the question, I want to kind of give an example to the users who may not really understand networking at the level that you and I do. So let's just say we're moving a file from our laptop to Dropbox. Or we're actually using Oracle since we're at Oracle World. And we're using their SaaS product and we click on the screen. And instead of something happening, a phone call comes to us and they say, hey, I want to accomplish your goal that you've clicked. But I need to move some things around on the network and reconfigure the network. So just wait a few minutes. Well, that's not a good answer as it's due, right? That's called cloud is broken. And storage, guess what? It works that way today. Compute, guess what? It works that way today. Networking, no, you got to wait for a phone call. So fundamentally, we want to change the way people program the network. And because the supply chain now is mature, now we can get back to the point you have ASIC manufacturers like Broadcom, Cavia, Melanox, Intel, and Marvell all supply very mature silicon. They can actually produce it faster than at homegrown silicon within a company. So there's a real go to market advantage by adopting commercial silicon. There's a whole group of people in Taiwan over a $200 billion manufacturing base that make very similar switches that Cisco makes. But at a high volume and a better price point. Yeah, so Steve, if I could say that it's what's known as the ODMs out there are original vice manufacturers. I could tell you really opened my eyes probably about two years ago. I started hearing a lot about them. So the first one that kind of creeped up was Quanta, but there's Acton and Foxconn. A lot of people probably know Foxconn because they also make iPhones. And so there's a number of these companies, many of them based in Taiwan. And one of the main things that they did is they build the switches for the big companies. So Cisco's and HP and everything, some of them there's the OEMs, which means I design it and kind of brand it myself. But the ODMs just build your design and then ship it to you. So how are the ODMs changing their business model and how do they impact what you're doing at Pica 8? Well, I think it's just like you say. When I was at 410 Networks, by the way, a couple of shops ago, we used those same manufacturing companies to build our products. And we learned five years ago that they were still green. They didn't understand some of the requirements of the data center, for example. But guess what? When you work with a lot of companies, you learn. And they're now realizing that the product they're being asked to make is the same over and over and over again. So they're saying, hey, maybe I actually have something I can sell to the market directly. All I need is an operating system. And boom, that's where Pica 8 comes into the puzzle. So we're finding that we're a catalyst for these new manufacturer or legacy manufacturers now want to turn as an end user supplier. They just need a piece of the puzzle to compete. We provide that piece of the puzzle. And now we have a complete solution, if you will. All right. Yeah. Can you help us unpack this a little bit, Steve? Sure. So the ODMs, does this mean that they now go to market with your solution? Are you doing the solution? How is the supply chain going from today to what you guys envisioned it to be? Yeah. It's still early stage, too. But a lot of them have come forward and are actually reselling our solution with Broadcom solution and some of the other startups in the space. So one of the things a customer wants to see is choice. And one of the things a customer needs to feel comfortable about a new mantra, a new market segment is stability. And so by having one or two suppliers do this with one operating system, that wouldn't be enough to trigger the domino effect. So we want more people to jump into the open networking mindset. So there's actually four or five ODMs now that are very aggressive. There's four, by my count, operating systems available on the market. So I think we've got critical mass to really turn this thing over and really cannibalize Cisco's market share, by the way. I think that's really the end game. All right, Steve, you've used the term open networking a bunch. Can you help break it down for us? And how does open networking and SDN fit together? All right, well, they do fit together in our worlds, too, and I'll get you there. So first let's define open networking. So it really is a marketing term to define this modular model that I mentioned before. You have a bare metal switch, just like you'd have a bare metal server in a Red Hat environment. You need an operating system. That's where Red Hat comes in in the server model. Peacate comes in on the switch model. And then again, you're seeing with SDN, which we'll get to explaining more detail, there's a whole new app store developing there where you have different applications to perform different network functions. Just like in the PC, people became experts at each of their layers. You have metal makers like Dell. You have operating systems like Linux or Microsoft. And you have multitudes of applications available for multiple suppliers. That's the freedom that open network should bring on the networking side. Okay, and so Steve, I've had the opportunity to attend the open networking user group. So there is kind of a groundswell from some, especially the big enterprise guys. And one of the impetuses that I've seen is they look at really the hyperscale players, the Googles and Amazons of the world and see what they're doing and how they tend to buy straight from those ODMs or they build something by taking the chipsets from Intel on the server, Broadcom and Intel on the switches and build their own piece together. So does that fit together, kind of looking at what the hyperscale guys play? Yeah, actually they prove the model. And I think that's a lot of people look at the hyperscale guys as people who are willing to break the mold. They have the mass, they have the muscle, they have the depth, they have the patience to figure it out. They've got a team of PhDs and one of the things we talk about at Wikibon is those hyperscale guys, they will spend their time to save them money, whereas opposed to the enterprise and even some of the service providers, their IT staff doesn't have necessarily that team of PhDs and they don't have time. So they want to spend money to buy a solution that works. So I'm guessing that's where you guys help close that. Well, no, we're not there yet. Actually, I think SDN isn't even there yet, but I think Stu, you're on the right direction. The past starts with the hyperscale guys because they have the depth and breadth and understanding and they can teach us all what works. The service providers and the carriers are actually the beach we're focused on because they have a huge pain point in service delivery. Imagine a telco that makes a phone call for every customer deployment. That's a truck roll, that's expensive, that delays a service, that creates a branding taint, not a branding promise, right? That breaks the promise of good services. So we're experimenting a lot with that group because they're like the hyperscale guys and that they have time and resources and a real business driver. I think you're right, Stu, the enterprise wants that point and click and go kind of app store model and I think we'll get there over the next couple of years but we're still building, we're still understanding. So obviously it's a journey to move to this new model. That's right. Talk to us a little bit about Pika8. Where are you in the company life cycle? How many people are you at? What's the funding? How many customers do you have? Well we just, we recently raised series B by the way, so the way you raise series B is you show VCs that you have customer stickiness and you can do it again. That's the faith, that's the model. So we've passed that hoop and we're well on our way to grow and do more sustainable differentiation, building, acquire customers, expand to different regions if we choose. So- Great, can you share how much the series B is for? Oh sure. Is it 12.5 million? So it gives us a little over 20 million totals too. Great, congratulations on that. Thank you very much. It's never easy to raise money, but when you get it you feel great, so we feel great. But the story isn't over, right? I've already said this is a journey, you've said this is a journey, this is going to take time. So where we really spend time is customers who want to partner with us over a six, 12 month period to figure out what are the milestones of success? What is that app for them? What's the business case for them? What would be the challenges of retooling their organization? So we get very specific about a very individual client, and that's where SDN is today because we're still trying to figure out what is a complete problem and a complete solution that sells. And then we ask ourselves can we replicate that? Are there more people like them? And that's back to again, good startup theory. You pick customers you can work with that replicate and from then you figure out how to go after the mass market. So where we are today with SDN is we've got a lot of proof of concepts in the trenches. We've got a lot of ideas around three things and if you want to spend the time we can go through those to actually four application ideas. And that'll help I think people understand where the beach heads are. Yeah, I remember one of the first times I met with you, I knew a lot of people that had a box in the test lab that they could play with this because people are still trying to understand what this SDN thing is and how it fits. I mean you listen to VMware and Cisco dominated that market. When I interviewed Cisco they said there isn't any SDN solution that's fully baked today. Now VMware's position of course is that Martin Casado banging on the table. It's here now, we've done all the work, we've got 200 customers that are doing it in there. So yeah, let's walk through some of the applications you're seeing in the use cases. Well let's start with that one because that's the one most people talk about. This is the whole model of an overlay which basically says can I virtually create domains of interaction or assign virtual machines to certain logical domains so they can have them communicate or share resources. That's a heck of a lot better than re-engineering the network every time you want to do that. So it's a time saver, it's an agility story, it's meant to solve a problem of the network being sluggish. So we're actually trying to figure out how to make that network not sluggish underneath those overlays. So we're looking at technologies that allow us to look at traffic patterns and reassign bandwidth and resources on the fly. So it's kind of a traffic engineering story. It actually happens within the data center but it also can happen between data centers where you have massive bandwidth flying between two data centers or three data centers and you need to load share or make decisions which data center is actually able to take that data or which bandwidth path is cheaper because some of these paths are actually still governed by telcos versus dark fiber. The third application is attacking an actual device. A typical edge router today is a well-known device that sits between the carrier network and the enterprise network. It's actually a subset of the ethernet market. It's about a five billion dollar space but the reason it's attractive, it's a point solution, it's a product that can be canalized by commodity hardware and new software thinking. And then the third or fourth app, I'm sorry, is something that a lot of people talk about. It's not exactly exciting but it's really important. It's called network tapping and this is the idea of monitoring physically or virtual assets on the network to understand what's working what's not. Again, part of that SDN need is to have real-time feedback in the solution. So network tapping being a beachhead application makes a lot of sense because that starts to build the foundation for real-time data that will feed into some of these Uber apps of the future to tell them what to do. Yeah, so Steve, some great points there. I'm wondering if you can help parse something for us. When I look at SDN, obviously it makes sense for the Googles of the world, for what they're doing and the service providers in many cases have growth patterns and geographical dispersion where many of the pieces that you talked about make sense. When I get into the enterprise, there's a point where I say, okay, I might not have the growth pattern or I might not have the sophistication of my network that I will need to do this. And where I'm trying to line up is many of these same customers are trying to figure out what do I do in-house versus what do I go collated into the service providers. So are you having those discussions with customers that are sorting out where their applications live in-house versus out-of-house? Absolutely. And how does SDN fit into that discussion? Absolutely. How many companies really need to worry about going through this transition and how many of them can just say, look, I'm going to do this little piece of it for myself and that whole SDN and everything else, I'm going to let a rack space or some other service provider or big cloud take care of it for me. Right, well it's a great question and I think there's a misnomer out there too which is the Hyperscale guys basically have one giant app. It's a giant mainframe designed to do one thing well whether it be search or social networking or video distribution. And you go into the enterprise, it's exactly the opposite. They have thousands of apps and in many cases they don't even know what apps are in the data center. So it's not the same level of complexity. In fact, it's orders of magnitude higher complexity. And this is one of the reasons I think the enterprise generally does sit back and kind of wait for these trends. Sometimes to sort of nurture and help me understand I don't want to get too bloody stew. I'll cut off a finger now but I don't want to cut off my hand, right? So what we're seeing with the enterprise is that the more they use their network to generate e-commerce revenue, the more they're touching SDN today to learn. And it might be a simple application like a Hadoop cluster. Hadoop is something everybody uses today for data modeling and data mining. Big data is the buzz but it really is a algorithm that runs on high performance servers to help a decision maker see data presented in a different way. It's a contained app. That's really the point. It's controllable, it's containable. It's usually in a set of racks all by itself. So that would be a perfect example of where to try open networking as a starting point where maybe you want to try some tapping to understand are the traffic flows optimized for the computer algorithm. Maybe I can squeeze a little bit more performance out of the cluster. So that's the way people think these things through stew. They kind of put boundary conditions around it and dig in there. And they don't just throw it at the overall operation because they know things will just break and fall over the floor. And Steve, I've spent most of my career working with the enterprise. So you know, there's a reason why they can't do anything. They buy networking storage on risk. And if something goes wrong, they're out of business. The service riders and the hyperscale guys, if they don't adopt these new methodologies, they can't deliver on their business. So as people said, there's that pain there that they have to do this because if I'm Google and use your example you said before, if I have to wait to move something or make sure it's in the right data center, forget it, this internet scale that we're talking about here. So it's really different. So where are we on this journey? What are the conversations you're having with the customer? How much do, are they listening to what Cisco and VMware are saying versus that messaging? Either of them have a claim to be open networking? I mean, their job is open daylight and open stack. And you know, they're doing pieces. You know, take off the gloves a little. Tell us what you think of these guys. All right, well first of all, open doesn't necessarily mean open to everybody, but let's first talk about where the journey is and we'll come back and hit that one. So I think you can kind of tell by the way we're talking that the hyperscale guys are really the beach head. They actually are doing SDN today. Amazon's cloud environment is very SDN-like in how they can provision and assets and move things around and give customers the environment they want. Google by default is an openly a leader in the SDN environment. It goes to standards bodies, publicly talks about what they're doing. So they're the poster shot for, hey, it works, try it, I'm pushing the technology. The service providers and carriers, so you know, Rackspace or your Telcos are the second beach head. They're deep, they're doing POCs, proof of concepts, limited trials, NTT, Japan for example, actually was first to launch real services about two years back. The next beach head will be the enterprise. Financial services are dabbling, e-commerce based organizations are dabbling. What I would call a traditional print and file brick and mortar company. Those are the guys who are still sitting in the theater, literally in the peanut gallery waiting for someone like you Stu to say it's ready to go. And I will admit, it's not ready to go yet because there's too much learning that has to be had. Now let's go back to what's open networking. Open networking by default means that you're willing to open up your system entirely, not just an API. So a lot of our competitors like Cisco like to say, hey, I've got an API now and now you can program our switches using Python. That means we're open. Well, my friends, that doesn't mean we're open. That means we're letting a customer use a script. That's not open. That just means we're being clever with APIs and remarketing them. How about that for gloves off Stu? What open networking really means is we're really ready to delaminate the stack and let the customer go to a website, pick the metal they want, drop in the ASIC and CPU they want, select from four or five operating systems, and then here's the joke of the day, have the Amazon Droid drop it on your doorstep the next day. That's open networking. And I can tell you today, Stu, Cisco has no interest in this model. In fact, they're terrified of this model because it's going to cannibalize their business. Yeah, I mean, look, Cisco has been doing a lot in the open source community but it needs to run on their gear. That's right. So Cisco came and asked me, what does it mean to be open? And I said it exactly, if I have to buy your hardware, you don't fit that fundamental definition. That's right. Look, not to say that they're not doing good things in open daylight, open stack, and lots of movement there. And obviously they've got a huge install base. That's right. And boy, is there a huge army of CCIEs that are probably going to follow them. That's right. But if you're talking about following the open source path, they've got some work to do if they wanted to be really involved. And a little bit of humility for myself from maybe not somebody I wrote a blog recently on the history of open so you can come to the PGA website and read that. But open doesn't mean, or actually open means multiple things to multiple people. And one definition of open that we actually appreciate Cisco being a part of is as you said, we all want people to participate in these communities because the community gets masked when everybody's playing. And that's good. We all learn, we all share. So Steve, wondering, speaking of that, open daylight just came out with their helium release. That's right. They've got a number of players now that have various flavors of the open daylight controller. What's your take on how that fits into the story? So two answers. One, we appreciate that and we think open daylight is actually the open source control of choice this year. Last year, Rio was one of the leading controllers from NTT Lab. This year, it's Rio and open daylight. That's actually good because it shows people are learning. And again, the whole point of choice, right? People don't move in a new trend unless there's critical mass. Our value prompt is tuned towards nine controllers. We actually believe that by being open as a system greater in this early stage of our journey, we get more cash with customers by showing them that we can cobble together a solution that matches their needs by asking them a few questions. What do you need? What do you like? What's your team like? What kind of services and tools do you use? And then we kind of piece together something with, hey, let's start here. That really helps people get started. In fact, we launched the starter kit last year that did exactly that. It was a very simple kit. It's meant to help you get from zero to Nirvana in 15 minutes. You actually build a very nascent network tap on our solution set, but it lets people play and understand the value of combining the pieces, which again, it was what they have to learn. This is the new part. I have to combine pieces. Oh, it used to come together. And then the second part is now that I've combined them show me what I do differently with the provisioning and management, which is the speed and agility story. All right, so Steve, if we're talking about the maturity of this, what milestone should users be looking for? You know, for them to be able to say we're making progress, you know, looking for various places where they can jump into this market. What do you expect the next kind of 6, 12, 18 months out? Well, I think it's actually monthly progress is what I'm seeing, Stu. And I think it's just back to people's pallet of deciding when they want to jump into the hot water, the cold water, the deep end or the shuttle. And so I think you're going to see more momentum around open compute project. There's projects like the open network clinics that are trying to be driven into open compute, open network install environment, which is an open source tool to help people onboard operating systems on the bare metal. So these are components that should be open, should be free, should be things that everybody can use and understand. I think there's more ASICs coming to market next year, so we're going to have yet even more choice and more demand to pick the ASIC you want right now, not wait for your OEM to put it on the roadmap and you beg and plead for that OEM to accelerate that. That's ridiculous, right? Well, I want it now, it's available now. Why can't get it now? You're going to see more bare metal switch providers. So let's call them that, be aggressive from Taiwan. I think you're going to see big system integrators. Dell's made this move to jump on open networking, but you're going to see other big system integrators say, hey, this is cool. This helps me compete against Cisco. This helps me show customers my value. My value is not building a box. It's building solutions that matter. And I think that's the right direction. And all this is going to happen, just believe it or not, it's due in the next three to five months. In terms of next year, I think we're going to see people be bold and come out and say, I'm actually making a lot of money on this application. Step one, step two, here's the five customers who will actually talk to you about it. And I think that's the curtain that has to open up. Customers needs to come forward and say, hey, I'm actually using SDN. I'm using it from new vendors. It's working. Look how much my business is now being driven better. Hey, this is win-win. And I think that's what I'm waiting for too. All right, so Steve, so last question I have for you is the practitioner to themselves. So it sounds like it's pretty easy to get your hands on some code. You can get a box. What is the role of the network administrator become after we go to open networking? Has everybody become a programmer? Do they all have to be coming to puppet comp and understand all those pieces? And we're running low on time. Maybe we'll have to pick this up next time. But tell me what you think of kind of the, what should the network admin be thinking about his role going forward? Well, first of all, I think they should, first of all, not be afraid. And they have a job for life as far as I can see, because IT people in general really scarce right now, no matter what your skill set is. And I think that's one of the misnomers out there. The second misnomer out there is, is that we don't all have to become chef or puppet heads. Network engineering is an art and a science. And understanding how to build a network, whether using open networking tools or Cisco gear, those skills still matter and will matter for as far as I can think into the future, because we're still going to have ethernet packets. We're still going to have internet protocol. So what I see though is a convergence of a mindset. How do I program the switch is going to be as important as, how do I enable other people to program the switch? And I think that's where people get a little lost in the conversation. You're still going to need an architect to design a topology, configure the basic network, but you're going to have a whole host of both admins who will actually be allowed to build applications or click on applications and program the network, where they are now offloading that challenge from the one guy or two gals and young guy who when I get up at 4 a.m. to do that, now those people can sleep, actually have eight hours of sleep still, wouldn't that be nice? And because they've offloaded and packaged what they do into applications, that's really the dream of SDN is scaling the operations team, whether you're an engineer, an architect, or an admin, it doesn't matter. Spread that well, love. Allow cross-functional training. Allow teams to actually bond on a management framework. So network compute and storage are all managed the same way. That's really the vision. It's not to disrupt people's jobs. It's to get more horizontal sharing and cross-training. All right, well, Steve Garrison with PK8, I appreciate you coming so much, coming on. You're now a CUBE alum, so we'll be getting a LinkedIn invite soon. I love it. And, you know, big future for open networking, major trend that we're keeping close eye on and we'll be talking about through many of the shows. This is Stu Miniman with the CUBE's coverage of Oracle Open World 2014. We'll be right back with lots more coverage after this break. Thanks, Stu. Great to be here.