 from Seattle, Washington. Extracting the signal from the noise. It's the Cube on the ground at LinuxCon North America 2015. Now, here's your host, Jeff Frick. Hi, Jeff Frick here. We are on the ground in Seattle, Washington at the LinuxCon North America 2015 conference. It's kind of a small, intimate affair, but it's got a lot of smart folks, very geeky, very techy, and way overdressed for this conference. That is for sure. I don't know the right switch. I could need a hoodie. Someone send me a hoodie. So we're really excited to be joined by our next guest, Shrajeet Mukherjee from VP Engineering Cumulus Networks. Welcome. Nice to be here. It's a really exciting conference to come to, and this is clearly one of the highly monumental and highly important conferences for cumulus networks, because as you know, we are very, very focused and centered around the Linux networking stack and Linux as sort of the cornerstone of the modern data center. So before we jump into that, we've had JR on a number of times. Why don't you give us kind of an update on cumulus for the people that aren't familiar with the company, kind of what's kind of your guys' core mission and kind of an update on the company since we probably talked to JR, maybe AWS re-invent, I think, almost a year ago then. Almost a year ago, yeah. So as then, so now we are in the business of promoting the Linux networking model, not just as a consumable API or as an interface, but also as the way networking should be done. And fundamentally integrating all the components in the data center using Linux as the language. So we have been trudging along on our mission. We have since added probably another 15 new platforms. We've added 40 gig support in the meantime. We've added support for something called multi-chassis lag. There's a bunch of our routing performance has improved leaps and bounds in the last year or so. So the fundamental point that we are trying to address still is making Linux networking super easy to consume, just install it into the core of your data center and go. Right, so networking was kind of the last of the three tiers of the modern compute structure, right, to kind of get into the whole virtualization space behind compute and behind storage and kind of this whole software is eating everything, but you guys are taking it even another level, as you said, really integrating open source and making Linux the kernel of that process. So talk a little bit about why you chose that path and what are some of the results of choosing that path that have taken you to where you are now? So as I said, right, our world view has always been the problem is not about storage, networking and compute. The world view has always been that it's about the data center. The data center, the language of the data center is very unambiguously Linux today, right? There is no real doubt about that. So if you were to go down that road, networking, as you said, was the last stand out and storage in some sense is sort of abstracted away because you're talking to it to a very well understood set of interfaces. So networking, managing, management of networking, which was the biggest bottleneck in the sort of the modern data center sprawl had to be addressed. Now, multiple companies have tried to address it by saying, we'll build you a better abstraction, but the abstraction can hide only so much. Whereas if the underlying infrastructure was actually speaking the same language as every other component in the system, i.e. Linux, wouldn't your system just get simpler? It doesn't remove the abstraction, but it makes abstraction much more valuable and much more valid. So that's why we picked that path. Having picked that path, the most obvious choice was to make sure there was a wide platform base of usable solutions that customers could actually deploy that ran Linux on the front end, became the user's interface and interaction model and could run it on industry standard silicon at wire speed the way people are expecting to use it. But then you get the added benefits. So not only is it Linux because that's consistent with the language that's spoken by the other components inside the data center, but now you get the open source piece. And the open source has proven to be the model for innovation. So how has that really impacted your guys' ability to deliver on your vision? Right, so that's exactly right. So from a technology perspective, we have been very active in a bunch of open source communities, even in the Linux networking kernel. We've made major contributions in the last year or so. And the idea, therefore, exactly like you said, since open source becomes a platform for innovation, this innovation that could be developed for compute, could translate into networking, could be developed for networking, could translate into virtual networking on the host and so on and so forth. User interaction tools, automation tools, everything that people have developed at large scale, i.e. chase the compute dragon, if you will, has direct translation into networking. Right, right. So then where do you guys, so you got the vision, where are you kind of delineating between your contributions back to open source versus your guys' engineering contributions back to your proprietary? Our general goal is if Linux is our ecosystem, if Linux, the ecosystem doesn't grow, we don't grow. There is really not a very hard delineation of that along those lines. The general idea is that the common use of Linux as a networking platform has to be proliferated. And we are doing everything we can to make Linux an enterprise-grade, mostly on the feature side, clearly on the quality side, Linux has had a lot of people working on it before we showed up even. But mostly on the feature side, we are trying to make it such that the Linux networking platform becomes enterprise-grade, becomes something that's easily consumable, automation tools can work with it in a very simplified way, and it fits into the global vision of a data center. If all of those statements come true, we benefit from it directly. We don't have to worry about some hidden, closed-source technology that's only ours, and therefore can be used eventually as a cudgel against other people. So what are the next big hills to climb? What are some of the immediate challenges, short-term, mid-term that you guys are working on? So clearly very long-term goals we can't talk about at this point, we are a startup. In the short term, the general goal is to win the entire enterprise data center space because the general point is whether it's us or somebody else, the meta point is that Linux will be the dominant networking platform going forward, and we obviously want to have a very, very large role to play in that. And when you're approaching a customer, is there places for them to use you as kind of a land and expand strategy where they can, or do they have to rip and replace their entire kind of infrastructure? I assume that that's not the case, or you'd have a hard time selling as a startup if it's a rip and replace. So how are people kind of entering into this realm? So I mean, there are many, many different answers to that obviously, but the most general trend we've seen is that people have gone towards scalable data centers. They want elastic scale. Elastic scale generally lends itself very well to layer three networks, and people going into layer three, like doing that and approaching, as you said, land and replace, or land and expand would go very well with that L3 inclusion into the network strategy. So last question for I let you go. Give for the people that aren't here in Seattle, your impressions of this conference, have you been here before? What's the vibe? What's kind of the hallway chatter? What are you taking away as you leave here after day three? So I would say I'm a pretty consistent Linux plumber's attendee, and I have seen over the years it going from what was a truly intimate affair to now what is called an intimate affair, but it's not really given the actual strength in numbers that shows up. It's great. I think it is actually a, Linux in some sense is kind of the shining light of the software development movements, right? There's a lot of open source movements that have had real trouble trying to get a community going, getting the community actually moving forward, getting a bunch of responsible people driving it and actually doing a good job of steering it. Linux is the shining light and is the gold standard today for open source development and Linux plumbers is definitely sort of the core group and the core set of people that are making it happen. So I think it's great and I hope it continues on its current trajectory and on top of that, Seattle is a great city to host it in. It's close to us and it's still a fun city. Excellent. Well, Shajit, thanks for stopping by for a few minutes from Cumulus Networks. I'm Jeff Frick, we're at LinuxCon North America Conference 2015 in Seattle, Washington. You're watching theCUBE. Catch you next time.