 Okay, welcome back everybody, this is SiliconANGLE's coverage of Splunk Conference, this is our exclusive two days of live coverage, we are in day two, this is theCUBE, it's our flagship program, where we go out to the events extracted from the noise, I'm John Furrier, the founder of SiliconANGLE, I'm joined by my co-host. I'm Dave Vellante of Wikibon.org, Carl May is here, he's the co-founder and CEO of Velo Systems, a networking specialist. Carl, welcome to theCUBE. Thank you very much. So, software to find everything. Software to find systems, right. Software to find world, what does it mean to Velo Systems? So, let's start with what you guys do and then we'll get into it. Well, what Velo Systems does is provide the software platform to abstract all of the hardware underlay, if you will. So, we actually view the network as a system bus, we built an operating system based on Linux that treats an open standard network as a system bus, we then use that as the plug-in for compute and storage elements that go directly into that network. And we build applications on top of it, one of what we're demonstrating here at CONF is how we're leveraging big data, how we're actually gleaning information from the application environment, from the network elements, from the compute environment, from the applications, using that information to flex the underlying system and therefore respond to the application. So, it sounds a little bit like VMware with an application component. I will promise not to bring that back to Gelsen. So, actually virtualization has been hot, network virtualization, which became SDN, the Sierra acquisition, but you're seeing that the tide's turned. So, are you getting data from the application and the network at the same time? Is that what you were doing? And the other elements. So, what we do, so let's talk a little bit about SDN. The core tenant, really, of SDN is about separating the control plane from the forwarding plane. And there are many different applications. There are many different ways you can leverage that particular separation of function. Number one, what that does, let's talk a little bit about that does, what that does is it allows you not to look at the network as a collection of tens, hundreds, thousands, or whatever, hundreds of thousands of devices, but now it looks like one big system. So, that's number one. Number two is, if you look at what people have done, or other companies have done with this, like NYSERA, is they've really focused on virtualizing that. So, they don't really care about the underlying fabric. What's that transport look like? They've built these overlays. And most of the other companies that are currently active in the so-called SDN space, software-defined networking space, are really focused on building these overlays. What's different about us is we've really focused on the underlay. So, we were one of the first to not only support white box or so-called merchant hardware platforms, but we've also extended the notion of the networking device beyond just packet switches to optical cross-connects, DWM devices, LTE radios, and so forth. So, the idea is that while NYSERA is focusing on building that overlay for VMs, let's say, we're really building the underlying infrastructure to support, for instance, the likes of network virtualization solutions. Dave and I have been really intrigued since over a year ago now, since the NYSERA acquisition, because that kind of was like the big bang theory. I woke everyone up, but we would look at network virtualization, and then you see the rise of OpenStack, right? So, when we talk to CIOs, they're like, hey, you know what, I don't necessarily want Amazon, but I really got to get my data center act together first. Absolutely, yeah. And so, let's get our on-premise facility nailed down. There's all kinds of issues, power and cooling, capacity planning, all those things, but now the cloud is definitely there. So, and then the DevOps world kind of spun at the same time. So, you've got the confluence of software define or virtualization, data center kind of rebooting or refabricating that. Absolutely, yeah. And then you got OpenStack driving kind of like the hype around possibly a public-private hybrid cloud, and then the DevOps piece, which this show really speaks to that DevOps mindset, where you got guys who are in IT where they have one role. They're developing, I don't really want to get into the network side. So, what do you guys do in the DevOps area? Could you talk about how that trend relates to the show here and your company? That's actually a great question because as it turns out, the DevOps teams tend to be the customers that we go and sell to. If we look at the mid-market and even the high-end, the Fortune 200, 300 customers that we have, and we have some very large ones like State Street Bank, it's really the DevOps type of groups within those organizations that have bought from us. What they're looking for is they're really trying to assemble solutions based on a variety of, we'll call it for lack of a better term, open standard components, that they can then use to customize their environments or their data center, that data center reboot you talked about, for their applications. And I think that for us, as it turns out, and perhaps it's more coincidence than anything else, but we've actually found that almost all of our customers have really been the DevOps teams within those companies. And so the idea is that if you look at what we provide, remember we're selling essentially a Linux distribution with open APIs, with all sorts of tools that you can use to now build applications on top of that. So consequently, you now have the ability to give an organization, a customer, the ability to now construct their own applications. We have customers, we have a customer that's built his own exchange platform directly on our OS. And they're using complete merchant hardware to deploy it. And they've got, they've realized some significant CapEx and OpEx savings, but that's sort of secondary, what was primary to them is they were able to implement this in exactly the fashion in which they needed it. So there's converge infrastructure changes, really, is that the wind at your back for you guys? You know, with the flash, you got converge systems now integrating in. That's an excellent point. Actually, as you see in our mid-market customer space, we actually are right now working with two very large partners who have taken our software and are using it as the platform, the software abstraction for their converge infrastructure. So now they combine the compute, the storage and the networking all in one big box. And that then is customized. We have a variety of templates that we support. The templates then are coupled with with a variety of applications. It could be exchange or SharePoint or whatever. And those combinations are really what is, we actually do see that trend as fitting right into this DevOps. DevOps, turn me as well. It's all got started with the hyperscale guys, you know, probably a decade ago, the Amazon's and Google's and more recently, you know, Twitter's and Facebook. And so my question relates to that playbook that some of the Fortune 500, let's say, have begun to adopt. DevOps fits in there, extracting value out of what was locked inside of a controller and putting it into software, layering it across the infrastructure. You mentioned one large customer in financial services. Are these large customers able to replicate that hyperscale like capability of say, you know, an Amazon or a Google? Or was it more, you mentioned converged infrastructure, they're looking for you guys to sort of fill that gap because they don't have the technical expertise, not that they're not super technical they are, but they don't have the bench depth, the PhDs, the engineering talent. So are they looking to buy more solutions over the long term? Or do you see them more getting the big guys looking like the hyperscale players? Well, I would probably have to split the market in a couple of ways. You are going to find very, very large customers who actually will tell you, no, we do have the expertise that the Googles and the Facebooks have and we are going to try to do as much as we can on our own. And they are in making those investments. And they are in fact, in some cases, even outsourcing development of custom hardware. I mean, it's not just about the software. They're defining a hardware platform, just saying, you know, we want our service to look like this, our switches to look like this. Like OCP, like Facebook would do, yeah. If you look at the open compute project, that's an excellent template that I think is being very aggressively embraced by a large number of customers. So yes, there's a signal of the market you address that is very willing to use its personnel. Frank would rather use its people to invest in a high value solutions that are tailored for its purposes. The next group, and this may not be necessarily tied to size, maybe just the particular culture of the company. The next group then is the companies that are saying, you know what, we're sort of in the middle. We don't have all the expertise, but we'd like you to work together with a systems integrator or somebody else to help us tie those together. And that probably is arguably, that's the bigger market. And that's your sweet spot, right? And that, and the next one is the mid-market. The real surprise, and this goes back to, I think, the question that you asked about the converged systems where we're seeing substantial demand right now is in the converged systems area. I'm tired of having to buy this from this guy and that from that guy and this from that guy. I want, you know, a la IBM Pure or VBlock or whatever. I want a nice system that is all tied up in a bow that I can deploy my applications on, but I don't want to pay the price of all the branded elements, right? And frankly, I don't want to be tied. We have a customer right now we're working with who has said, you know, look, we're using both VMware and OpenStack. How can you help us manage placement of VMs such that we're not, you know, we have a generalized toolset? So they're really looking for that converged system to, based on commodity hardware to bring our capital. Talk about the story of the company. Obviously, you guys, you've been the news recently, put a test bed for SDN in Equinix out there in the past summer, I think it was July. Tell us about the story, how it started, who funded it, was it self-funded, is there investors, some employees? Give us the quick story. So we're a little bit unusual in that respect. We actually founded, I'm one of the founders of the company, we started in June of 2009. We funded ourselves actually for the first couple of years. We had a vision and the vision really was that there was another way that you could, that we could bring together the core elements in IT. You know, we were looking for the perspective of application architectures. You had these silos that existed. What if there was a way that we could turn Linux into the enterprise OS? And- For the data center. For the data center. And networks and all the follow up here. Yeah, we're not going after the carrier market, so. And so, for the data center. Now it's, what we did from there is we, so we built our own operating system, VELOS. What was different is along the way we stumbled upon, and it's the way we tell the story, we did stumble upon. OpenFlow. OpenFlow. I went to- I can finish when you said this for me. I appreciate that, you're better than my wife. I'm not going to go there. I appreciate that. We stumbled upon OpenFlow at a clean slate event. This was actually before OpenFlow was really the common term. And we started saying, you know, why couldn't we expand the horizons of OpenFlow beyond just packet switches? What happens if you could start looking at, for all this evolution we were seeing of east-west traffic? What if I could start using an optical cross-connect to connect servers? Because I don't, I really care about latency and being able to move a large block of data very, very quickly. Or I want to move different types of traffic up through my packet switch network. Well, how do I do that unless I've abstracted the intelligence, the network intelligence, so that it's not directly tied to both, to that forwarding element? And so we built the, we built, sure, sure. Can I answer your second? So we had a crowd chat this morning. David Floyd, a Wikibon analyst, made a comment, I want to get your perspective on this. He was about data economy, the future of data. He says, let's start with the obvious. Data is more valuable when it is close to the other related data. The further apart, the less valuable the data network. Do you agree with that statement? That's a great statement. Our whole point is, that's how we've evolved. That's the latency issue. That's how we've evolved also in our business, which is that if you can now, where everybody else talks about latency, for instance, and even the networking space, they talk about latency, but it's at the switch level, by moving the control plane, our knowledge of latency is now through the entire application stack. So now we know exactly what the latency is from the application to application, not just one hardware box. So you have an optical cross-connect, so finish the story. So it's not ours. So what we've done is, we've now developed deviceware. The deviceware gets embedded in packet switches. We support currently three different packet switch chips, Broadcom, Centek, and Intel. We're adding one other, I think, later on this year. We also support optical cross-connect. Again, our deviceware sits in that. That's our component. And then also DWM. So we've actually embedded that into a DWM device. What's different about our model is that what's different about what we're doing is it's not just our architecture, which we believe is very compelling. It's the fact that we've also embraced a very disruptive business model because one of the challenges you have right now is that as system vendors have gotten bigger, if you look at the amount of margin stacking that has been added in, you're no longer getting really the fair value for that system that you're paying. You're paying for a lot of somebody else's piece that I glue into a PC board and put it in a box. Can you sell the OS? We sell the OS. That's your moneymaker. And then the hardware actually goes, you certify your reseller, you certify our partners, our YKMC, and resellers. They then take the white box hardware and their own, combine that together with our software and take it to market. And it has had dramatic impact on the economics of the solutions. Because you've been- And software. Absolutely, that's right. I mean, we love talking about this stuff. Dave and I always joke on theCUBE that under the hood is the engine of the future and that's really what's going on underneath these applications, the fabrics, whatever layer you want to call it. But, you know, conversion of the structure with flash and the network are being redesigned. I mean, networking has been the bottleneck, you know, East, West, North, South. I mean, that's where the action is. And it's hard problems to solve. So, you know, virtualization helps that. I guess we're getting caught on time. I want to get one final comment here for you to end on is, talk about this Splunk demo that you guys did here. We'll get that plug in for Splunk and get some- Yeah, actually that makes a lot of sense because it goes back to that vision that if you can learn from your application environment, you can make better decisions about how the infrastructure actually responds to the application. So we built an application. We started evaluating this about a year ago. We developed an application called Farsight. What Farsight does in its first release is it goes out and it analyzes data from the network, from the compute sources, from the application. Where we've taken it a step further and what we demonstrated this morning here for the first time is that we also now can automatically remediate problems. So now if I'm running an application, for instance, a critical e-commerce app, which is what we showed, and all of a sudden I start some other part of the organization starts a database backup. As it turns out, we had a customer who came to us yesterday and had exactly that problem. That then sends a spike in traffic in the network. All of a sudden, the response times to the e-commerce website go way up. People get very unhappy. What we demonstrated is how we can automatically remediate that, reallocate bandwidth back to the e-commerce application, throttle back what's given to the backup, but then when the demand goes down, we can automatically give it back to those applications that are sort of on there. You've automated that process with no contribution. Automated completely, absolutely. This is a power of Splunk, enabling all kinds of innovation, allowing guys who are building the infrastructure and the OS's of the future. Carl May, thanks for joining us on theCUBE. Great conversation. A little deep dive on some infrastructure stuff, which is a kind of, we haven't had the deep dive on this stuff. We needed a little fresh dose of infrastructure talk here. Velo systems, congratulations. Co-founder, self-funded, love that story. As I always tell Dave, it's looking valley. If you can self-fund and not take outside capital, you're the true tech athlete. So congratulations. We'll be right back. This is theCUBE, live in Las Vegas for Splunk Conference. We'll be right back after this next guest, after this short break. Great.