 Okay, we're back here live in Silicon Valley in San Jose for the Open Compute Summit. This is SiliconANGLE and Wikibon's theCUBE, our flagship program. We go out to the advanced extracted cylinder from the noise. I'm John Furrier, founder of SiliconANGLE, my co-host Dave Vellante, co-founder of wikibon.org. And our next guest, JR Rivers, the CEO of Cumulus Networks. Welcome back to theCUBE, Cube alumni on at VMworld this past year. Welcome back. Thanks, thanks for having me. So what's new in your world? VMworld was just half a year ago and lots of change in this tech world. We're in a bubble, I think an innovation bubble personally, but I love bubbles, I think innovation comes out of it. Any update on traction? You got some news you want to share with us? Give us a quick update and then let's jump into the news. Yeah, so business is moving along, like in the context of any startup, we've got a lot of customer interests, customer engagements up and down the spectrum from production to trials, all the way down to proof of concepts, kind of the classic, this is how you go to market. The real interesting evolution or change in the industry is we just announced today that Dell has, or in conjunction with Dell, they've opened up their networking hardware platforms for third-party operating systems and Cumulus Linux is the first third-party operating system for Dell networking. And we, Dell and the customers, look at this as kind of a pivotal point in networking, especially around the enterprise. In the past, if a customer bought a platform from an incumbent vendor, they're stuck. Once it was wired and cabled, that's all she wrote. Now, with Dell, they can buy a Dell networking platform and pick the software that runs on it. They can wire the platform up and they can put any software that they want on the platform and move forward. So it's pretty awesome. So take us through what's the evolution of Linux for networks. Obviously, you guys have had a little angle on that. And why is it such a big deal for Dell? Obviously, Dell is one of the incumbents that Dave was on stage talking about today around the quote, proprietary, as Frank called it, proprietary bullshit, still billions and four hundred billion dollar market. Dell's going to, obviously went private. We spoke to Michael Dell about that privately and he said, hey, I can innovate. So is this their version of separating out and trying to get a network OS to get into this new build your own cloud game or is it just fundamentally a product whole fill for them? What's your take on it? What's your view? I think it's Dell's trying to step them, say, hey, we're not about the proprietary bullshit. So if you look at Dell on the server and storage side, you can buy a Dell server, you can drop down and pick any operating system you want. Yeah, you bought the hardware from them but it's industry standard hardware, X86, arms coming on for them. Same thing's true on the storage side, you can buy storage solutions from Dell, which includes software, or you can buy storage hardware from Dell and put your own software on it, various software angles on it. On the networking side with Dell, it's in the past it's been proprietary bullshit, right? They're on OS and that's what you're stuck with and Michael and team looked at it and said, hey, right now we make people like our software to buy our hardware, why are we doing that? We should just open this thing up and people can choose our hardware and then pick the software that makes sense to them. Maybe they'll pick Dell OS, which they got from Force 10, maybe they'll pick Qumilis. There's going to be other people on that list too. So I'm sure Marius Haas was involved in this deal. He was. And Marius is three comm, sold out to HP. So Dave, this is an interesting chess game, interesting battlefield kind of lines being formed, converged infrastructure, how that's morphing, what we talked about this morning around this. Disaggregation. So Jerry, I'm interested, how do you compare to say the typical white box, right? How do you differentiate from what's going on there? It's got software management. So define white box. I mean, that's a tough question, right? In the context of networking, there's kind of, we'll call them little known name brands with their own little pieces of software on it that come from various places. We separate that from something we call bare metal networking. And that's really what Dell's kind of stepped up to. We're going to sell you a bare metal piece of network just like you can buy a bare metal server and put whatever software you want on it. And then it changes the discussion. So you can have a differentiation around what the hardware is, you know, is it quality? Which chip it is? Is it Broadcom or Intel or that kind of thing? Then you can also have differentiation around the software. What does the software do? What are its characteristics? Is it an open flow software solution? Is it traditional? You know, everybody gets to play separately. So you see differentiation around both layers. A lot of people say, oh, hardware is non-differentiable. We've heard that a lot. Doesn't matter. You would disagree with that. Absolutely. I mean, I think if you paid it, you know, Frank's keynote today was a leading indication of that. You can take an X86 CPU chip, you know, from Intel, let's pick them for the sake of argument and package it up, you know, thousands of different ways. And that's part of what OCP is all about is, you know, if I need, you know, a DualSocket X86 server and I need it sideways or diagonal in this big of a space, I can go to a set of providers that can get it to me without having to go through a traditional OEM channel, right? And so that is hardware differentiation. So give us an update on, so Dell obviously a huge deal, giant distribution channel, where else are you seeing momentum in the marketplace? So we have some, you know, some reasonably large, actually some extremely large outlier customers that we are either in production with or working with on heading towards production. We're seeing a lot of traction in the smaller web presences, content delivery networks, small social media companies, that sort of stuff, private hosting environments. And those are kind of production engagements. We're seeing a lot of trials and proof of concepts in enterprises, you know, pharmaceuticals, financials, that sort of thing. For a lot of those guys, the procurement channel actually still does matters, which is OCP is about freedom of choice. A lot of them have a approved vendor list and they want a name brand supplier like a Dell. And that's what inspired this relationship on both sides of the coin. The customers wanted to see Cumulus plus Dell together and help make that happen. So you're essentially trying to take the model, the Linux model for servers and apply it to networking, right? Precisely, okay. So why did it take so long? Because I mean, you've got the major networking company has a, you know, two-thirds market share, you know, dominant position. It's hard to even find another market where there's an entrenched player that, I mean, you look at the server business that's pretty fragmented, the storage business is fragmented. Cisco dominates the networking business. Why did it take so long for that open source, you know, ethos to come into networking? Well, I think it's been an evolution in a while coming, right? Now we've, excuse me, we're at a point where the hardware ecosystem around Silicon is you basically, it becomes industry standard. So a company like Broadcom has the de facto chipset that people use for most of the networking components. Even Cisco with the NCMA Nexus 9000 launches also based on Broadcom tried into, just like everybody else is. You know, Cisco makes a lot of revenue off of other platforms based on Broadcom platforms. And customers are starting to recognize that. They're starting to take the tops off of these platforms and look inside and say, hey, wait a minute, that's the same chip in this box, in this box, in this box. Isn't this all kind of the same hardware? Well, wait a minute, you just said this, said this differentiation, right? But that's an example of less differentiation, right? So you got to differentiate differently, I guess. Exactly, differentiation often times is supply chain, form factors, those sorts of things. But customers want to understand what to expect. Like, you know, if I came around and said, hey, here's my new networking chip, it's the best thing in the whole world. Nobody knows what it is, how it works, it's a big mystery. You can't build a system around mysteries. You have to build a system around things that you know and understand, right? Now the fact that you took a chip that's industry standard, like an X86 or a Broadcom networking chip and stick it in a form factor that's suited to your needs, like the open rack form factor for OCP, that's a different discussion, right? Right. So, Jay, I want to ask you about the headline, Wall Street Journal wrote a nice article for you guys, Dell embraces bare metal networking. So obviously, bare metal's back. Or bare metal is back again for the 10th time. But that's a real fundamental, you know, nuance, right? Bare metal has been the preferred approach for many people, building out data centers, stack and rack, now you got a big data center operating system concept going on. Cloud obviously, software defined, kind of everything from data center to internet of things. Still there's a software layer involved, right? Network virtualization can help that. But a lot of people forget about the actual bare metal. There's still bare metal involved, there's still data centers. So what do you mean by that? What does bare metal networking mean? Is it fully differentiated than cloud? Is it kind of interplays with it? Cause here, it's open stack on top of open compute. We saw a great demo from io.com, CEO. That's cool, nothing wrong with that. I love that, I'll take that all day long. But underneath that, what's bare metal networking? So mutually exclusive, are they work together? They're kind of mutually exclusive. Kind of take a step back from it and if you're a consumer, any level in the market, you should be able to go fill out a bill of materials and buy that hardware and then come in and put your software on top of it. And if you change your mind on the software, you should be able to put different software on top. So Greg Slushman, when he was talking about this thing, they have a platform that's almost like that except for their networking component. The networking component is branded and they're stuck with who that supplier is. They can't come in and- Branding you mean Cisco, Juniper or whoever? Somebody, yeah, I'm not liberty to say. The networking guys. Right, exactly, it's an incumbent networking provider. And so in the context of bare metal networking, it basically says you should expect your networking elements to be the same as you would expect a storage or a compute element where it comes in, you can buy the hardware, you can cable it up, you can get it from a whole bunch of different suppliers depending on the supply chain and form factors of choice. And that's what you want to disrupt. You want to disrupt, you want to disrupt, there's no real licensing fees, it's more per port or per box, right? You want to disrupt those guys, right? That's what you're, that's- The networking incumbents, right. The networking incumbents come in and say, hey, I'm going to go in and sell you my box in the rack. And if you, you, Mr. Customer, once you've cabled that thing up, you're never taking it out, I own you. Yeah, yeah. Right, and it's not so much, I mean, part of it is we want to disrupt it, but realistically, customers want that disrupted. They want to be able to say, hey look, I bought a piece of hardware, I cabled it up, I shouldn't be locked into Cisco because I chose Cisco at that point in time. If I don't like those guys, I want to be able to switch to another provider software. Share with the folks, what's the sons of Cisco? What does that mean? I saw that kicked around the register blog. What does that mean? What's sons of Cisco? That's Jack. Jack has the most awesome taglines ever, right? So, I was part of the Cisco crew for a while. I worked there about 16 years and it's kind of, I would say, part of the problem. And now I'm becoming part of the solution. Well, so I want to- You're disrupting the guy that you knew all where the traps are, all those barbed wire, trip wires, land mines. When I started, I was there for 16 years and when I started, we manufactured everything in-house. We wrote the operating systems from scratch. We did all the ASICs from scratch. It was really, really hard. And technology has matured over those 16 years to the point where we talked about industry standard silicon. We have operating systems like Linux that are very full-featured on their networking functionality. We have supply chain, manufacturing people that can build the box and ship it anywhere in the world. And so, a lot of the systems engineering pieces that we started off with at Cisco have disappeared and this is like the new Cisco. This is what it's going to become. 12 years ago you had to do it yourself and it helped Cisco's ascendancy, but so given your perspective and you're a partner with VMware, we had you on theCUBE at VMworld, do you have a perspective? I'm sure you do. I'd like to hear it on NSX versus Cisco's application-centric infrastructure initiative. Right. I think the easiest way to talk about it is to ignore the pluses and minuses of either side in terms of their features and functions but recognize the grip that any one of those solutions has on a customer. In the context of VMware plus NSX and let's just call that vSphere for the sake of overall discussion, I can build a data center and I can choose to run vSphere on top of that. And if I don't like VMware, for some reason they make me mad, I can throw that all away and switch to OpenStack. I'm not locked at all. My hardware is my hardware. I could have put new software on top. That's software-defined data center, right? Now VMware NSX, they can control a hardware switch, cumulus linux, or RISTA, a bunch of other people can be controlled by VMware NSX, but still it's a software problem. Cisco ACI is a vendor lock-in. You go and you build that data center around Cisco, the Nexus 9000 and the ACI architecture, you don't ever change your mind, you're stuck. It literally is a 15-year lock-in that customers are getting ready to walk into. Okay, so you're leaning toward the open approach. We're redefining open in this day and age, John, right? JR, I want to ask you about the DevOps killer environment because what you're basically saying here is, Dave, we're just talking about DevOps guys eat glass and spit out nails. They're a unique breed, right? And the networking has been a bottleneck. We've asked everyone on theCUBE from executives, Bethany Mayer at HB, network virtualization is kind of that first step towards changing the game. But in a DevOps environment, like a Facebook, for instance, or like George at IO, right? You mentioned, they don't want the standard gear, but that's all they can get right now. So you guys want to disrupt that. That makes total sense to where we're coming from. How do you guys compare in that environment with, say, Arista, who's had a good name with folks do large-scale stuff like the Cloudera's of the world, the new school guys coming up? Are you different than them? Are they not open source? Are they one of those guys on that side of the street of name brand? Or are they frenemy? We're definitely competitive. So we're sort of that way. You know, I'm not going to throw too many nails at those guys. I like them as a group of people. I think they're in the incumbent category and that they sell hardware and software together. So therein lies one of the main differences. The counterpoint to that is we both have recognized that having a platform that's open for people to orchestrate around in meaningful ways, whether it's leveraging an orchestration tool that exists like a puppet or a chef or a CF engine or a monitoring tool like Collecti, Nagyo, Scraphite, or some homegrown thing that a DevOps might decide to pull together out of mixtures of all that stuff, that's what makes modern realistic operational environments fly. So for the folks out there that might not be in the weeds on the networking, so what you're basically saying, correct me if I'm wrong, you're essentially allowing people to separate hardware and software environments so that they're not dependent upon each other so that they can mix and match the best solution. Is that what you're saying? That's part of it, and the other part of it is, cumulus specifically, our product is called Cumulus Linux. We hardware accelerate the Linux kernel, so Linux kernel data structures are intact, so anything that manipulates those things, anything you can do to orchestrate a V-switch, you can orchestrate cumulus Linux with. So it makes it a lot easier for people to leverage those environments and tools. I just added that on my list of knowledge, thank you very much, and sharing that with everyone there. So it brings my next question, which Dave and I are interested in, is when you talk about Linux kernel, you can't help but talk about Flash, persistent RAM. Where is that coming in all this? Because that plays interesting, because you start preserving the Linux kernel, Linux kernel based off, is that going to change the addressability around data, data management? Does that affect anything that you're doing? Are you, are you? The bottleneck, put more pressure on you, right? Does that help you? The storage, the spinning disk, no longer the bottleneck? Well, you know, for what we do, we're not fording packets in software or anything like that. We know these, the Broadcom ships, they do all the heavy lifting terms of capacity. So we, as a company, and we see the industry transitioning this way, we're trying to add, help customers add more capacity in an affordable, easy to deploy way. So if you look at a modern server, modern X86 platform, Intel platform, and you look at tiered flash infrastructure, there's no fundamental reason why customers shouldn't be running 40 gig ethernet to their servers today. And if you talk to like a Facebook or any of these guys and ask them, could you use 40 gig? They'll say, yeah, hell yeah, I could use it. And then they're going to say, well, do you use it? They say, hell no, I don't use it. And they say, why? Because it costs two freaking months, right? And so the reason it costs so much is because the supply chain is such that it doesn't allow people to get to where they need to go. Well, so what's next for you guys? Where are you at in your funding? Have you had your B and C rounds? Where are you guys at? Just give us a quick update. We're on round G, no, just kidding. No, just, that's a joke. We just recently raised a round. We typically don't talk about it from a PR perspective, but we've got enough money to last two, three years. Second round, third round? It's a B round, yes, yes. Series B, we have a great set of investors. They believe in what we're doing. We've got a lot of customer validation points, so they're obviously out talking to our existing and prospective customers to understand what the possible traction looks like, and they've put money on the table to give us, you know, less weather any short-term storm that Cisco can throw at us and come out on the backside as the winners. What has been the biggest surprise for you, my final question, over this past year, in terms of adoption out with the clients? Any aha moments for you? Any kind of revelations around where your value proposition is in the product market? Yeah, there has been. I expected a lot of customers to kind of not get it. And, you know, not everybody we talk to becomes a customer immediately, but every single one of them we talk to absolutely gets it. And they say this is exactly aligned with where I want my business and my IT department to go. It may take me a while, but this is where I want to go. And, you know, we see this happening now. I mean, we launched the company in June and we went GA with our software at the same time. So we're seeing customers now that are calling us back that we talked to in July or August and said, hey, it wasn't the right time. And they're calling us back now and saying, you know what, now it's the right time, let's go. What do you think's changed? They're organizations. They, you know, a lot of times they look at something like this and they say, this makes a ton of sense. I need to think about it for a while. I need to kind of let it settle and set. I need, you know, a lot of things have to align right. And they start looking at where they want to go with their infrastructure. You know, the fact that we have this, you know, the VMware support, for instance, a lot of them are picking VMware NSX as their network virtualization solution. They say, hey, great, this ties in really well to that. You know, let's go this way. I want to try it, you know, on this path. You know, and a lot of it, another part of it is just the organization. They realize it's not scary for a networking guy. For a networking guy it's just horribly empowering because now you can spend less time, you know, dicking around with simple things. You can automate those parts. It's been most of your time around architecting your data flows, increasing capacity, dealing with application level problems. Adding value. JR Rivers here, the CEO of Cumulus Networks. Obviously he was on theCUBE in VMware. A lot of great updates since. Big deal with Dell, Open Compute Summit. This is the homebrew computer club of what it would look like if it was in today's world. Open source at the backbone of it. Open environment, great open hardware. Great traction in its fifth event. Facebook really anchoring this out. JR, great traction on your end. Again, speaks to that same religion. Openness and choice and freedom. This is awesome about what's happening here. Great story, we'll be right back with our next guest after this short break. This is theCUBE, I'm John with Dave, we'll be right back. Thanks guys.