 The Cube at OpenStack Summit Atlanta 2014 is brought to you by Brocade. Say goodbye to the status quo and hello to Brocade. And Red Hat. Here are your hosts, John Furrier and Stu Miniman. And we're back with The Cube, SiliconANGLE TV's live coverage of OpenStack Summit Atlanta 2014. I'm Stu Miniman with wikibond.org. Lots of different projects and pieces of the infrastructure that we're digging into here. Joining me for this segment, we're going to be talking about networking with Neela Jakes, Executive Director of the Open Daylight Project. Neela, first time on The Cube, thanks for coming on. Thank you, I'm honored. So, you know, we've been covering, you know, cloud pretty heavily on The Cube this year, and as well as all of the open projects. So we were with the Open Compute Summit at the Open Networking Summit. Earlier this year, we had a lot of the folks that contributed to the Open Daylight Initiative, Red Hat Show, and now OpenStack Summit. For people that aren't familiar with Open Daylight or people that know, we would call it ODL for short usually. Can you explain what is Open Daylight and, you know, what's the foundation's role in it? Sure. So Open Daylight is an open source project. It's, what we're doing is that we're building a platform for software-defined networking. But really the problem that we're going after is while everybody in the world agrees that SDN is the future, that we need to centralize some level of intelligence, the problem is not that nobody is building a product around this, it's that everybody's building a product around it. And so you've got 15, 20, 30 implementations of SDN out there. And so what customers right now have, what end users have is a situation where they have to pick a horse in a 15-horse race. And rather than going in and picking one off and they're sitting on the sidelines saying, let's see where things shake out. What Open Daylight is doing is building one code base where the entire industry is going to either pick up and put within their products or plug into. And we're already seeing that momentum. Okay, and can you speak to a little bit of who makes up the Open Daylight project, who contributes code? And there was some news at the beginning of the week, HP is now a Platinum member. So give us the update. What's the lay of the land of Open Daylight? We pretty much have all of the major players in the space that are participating in Open Daylight. You know, from Cisco and IBM to Brocade, Ericsson, Citrix, VMware, HP, you mentioned. Obviously, and I'm sure one of my board members will beat me up. There are 10 of them. So, you know, from large companies like that, we also have small companies out there, small company out of Canada called InnoCybe. We also have, you know, some smaller startup divisions of larger companies like Nuage. So pretty much if you're a player in the networking space, you want to participate in Open Daylight and they are. All right, so in the key notes here at OpenStack, they made a point of saying that they're reaching out to other Open Source projects and Open Daylight was one of the ones. Can you reconcile for us, you know, what's the overlap or the intersection between Open Daylight and OpenStack? You know, we've got Neutron for, you know, networking. So how do they fit together? Sure. So, I mean, let's start at that level. What is OpenStack doing? OpenStack, obviously, is a cloud management platform. It's an open source project to create a cloud management platform. Now, if you think of a cloud management platform, what you want to do is say, let's say I'm creating a new tenant, or let's say I want to make a change and expand the resources for a given application. What I'm going to do is go out and say, I need to stand up a set of VMs. I need to connect that to some storage and I also need to connect that to some network. The challenge that we have in networking unlike the other two is that they haven't fully been abstracted yet. And so on the network side, very quickly what you're going to do is call your network guy and he's going to create a whole bunch of VLANs and then you're going to test those VLANs and connect them into your VM. You want to expand or change the way things are routed. You call your network guy, he makes a set of changes. So, Neutron's already an improvement in that in that with Neutron you're able to send some level of symbol commands down to the underlying networking hardware. But what Neutron is missing is a controller. Neutron in itself doesn't hold any intelligence. And so, SDN controllers, Open Daylight being one of them, certainly, bring that level of network programmability that is sorely needed to truly have a fully automated cloud or software-defined data center. We connect into OpenStack via the Neutron plugins. All right, think about since the Open Networking Summit, one of the bigger news items that I tracked was Cisco's OpFlex solution is now moving into Open Daylight. You can explain what that means. I think OpenStack has lots of different projects and there's different pieces of it. Is Open Daylight a monolithic software stack or are the components, how does that work? Yeah, I mentioned that in SDN, one of the key things in SDN is bringing intelligence up out of the underlying hardware. And different people have different views of how much to do that. What Cisco has said is really the most important thing for them or what they believe in the world is that you need to be able to capture application policy. You've heard people like Mike Dworkin who I'm sure has been on at least once on theCUBE talk about his view of an imperative model versus a declarative model. In his view, the imperative model just telling the switch what to do with the flow is the wrong way. The right way is to capture the needs of the application and to pass that down to the hardware. We in Open Daylight don't take a view. Both models could be right. We wanna support both models. Specifically the OpFlex group, and it started with Cisco, but there are other folks working on it, working closely with the group policy. You've got IBM, you've got Mitakura, you've also got Plexi, who are all working on this project, is an attempt to create a framework for capturing that application policy in a policy repository, and then communicating that down the stack, down the southbound side so that it can be implemented by the hardware. I think it's very exciting. Yeah, so people on the outside obviously know Cisco has been dominated in networking for so long. When Open Daylight came out, there's a bunch of pieces of the code that have come from Cisco. We had Lou Tucker on theCUBE talking about how Cisco's moving into open source. Can you give us your viewpoint as to, is Cisco dominating there? How are they fitting in with this whole open source movement? Yeah, it is absolutely true that SDN is fundamentally disruptive to everybody in the space, and obviously the largest member of the space is Cisco. And so it's not surprising that Cisco isn't sitting on the sideline and is participating in the biggest project, what is becoming the biggest platform in software defined networking. They also have the most resources, and so they've contributed, they've been a really great partner and member for us. They contributed some of their controller code, some of their teams are in there, but it's a really diverse community. What you see is that on many projects, you've got a set of people who are paid for, who receive a paycheck from Cisco, working really closely with people from IBM, from Red Hat, from Brocade, from others. But I think for me, the most interesting part is that what happens when they get in is they have a badge, they turn it over. And you listen on the call, you have no idea who's actually paying that person's paycheck. They work for ODL. And in fact, we've seen a number of cases like I'd say Madhu Venugopal is a great example, developer who is at Cisco, ended up moving to Red Hat, still working on Open Daylight, working on many of the same things. Nothing's really changed, he's still Madhu. He lives in the same house, he just happens to receive a paycheck from a different company. Yeah, we had Madhu on at the Open Networking Summit, along with Brent Salisbury, so great discussion as they are. So, tell me, how has the networking discussion in general been at that show? Maybe if you comment, I've heard some people say that Neutron seems to be one of the projects that needs a bit of work to get their stability and scalability out there. How many people here are fully understand that SDN more than the buzzword that's going on? You know, networking is certainly a big area of discussion. And part of it was where Neutron came from. You probably all know that there was a company called Nasera and really Neutron and Nasera, people thought of them in one breath and they were the company behind it, they've contributed so much to the open community with OVS that they continue to invest in. But it was a little bit of a surprise to the world when they got bought by VMware. And that obviously has brought some questions. If we look at a Neutron API, a lot of the people who used Neutron in OpenStack were actually using the paid Neutron product and it worked really well. Well, with Nasera being bought by VMware and being part of the NSX is that question for the open source community. What is our solution? Are we gonna go and buy NSX and buy into that? And there's certainly good reasons to do that but I think a lot of folks have been looking for a full open source solution. And so OVS continues to get investment but I think there's an excitement around open daylight because when you ask the question, should Neutron create a 42nd controller out there? I think the answer is no. We don't have a huge community that wants to create another open daylight within OpenStack, we have open daylight. And so I find that often, it seems like you're alluding to this, people feel like we need more networking, they're really excited about the maturation of open daylight. And increasingly open daylight I'm hearing is basically viewed as the default SDN controller of the open stack world. Yeah, I was just looking through the new survey that came out for the show that was launched yesterday. In production, OpenVswitch is still the dominant network driver out there. Can you talk about OpenVswitch? How does that fit into the open daylight framework? We love OVS. I mean, OVS in general, I'm personally very bullish on it. I'm generally someone who thinks that you should keep something private only if there's a really strong reason for it to be private. And if we look, I talked about having a million controls out there, we actually have a million V switches out there, right? In the vSphere environment now, you've got the VSS, the VDS, you've got the new one, the NSX Vswitch, you've also got Cisco's IBM's HPs. It's great that the community has a Vswitch that is open to everybody that is open sourcing that people can build on top. We leverage OVS and OVS DB and we're very grateful to Justin Pettit and Ben Fath and all those great developers who work on that now employed by VMware. But just like OpenFlow, OVS and OVS DB can't be the only way that we talk southbound. We have to recognize, I think this is the mistake I see by so many people, even vendors downstairs that I visit, is people seem to think that they can dictate one architecture and they've got a brilliant way to do something in one specific way and they're gonna assume that your data center, your carrier network is gonna buy all into that. That's not the real world. The real world is you have a whole bunch of gear and you have to hold on to that gear for three years, five years, seven years as it is. And so I see people using BGP, using NetComp, using PSAP in the cable industry, they've got their own. They've got the DMTS and the cable modems talking Doxus. And so from the open daylight side, we've made a critical decision to build a service abstraction layer so that we can leverage OVS if it's in there and we love it. But it can't be the only way to talk to things. We have to talk to physical hardware with OpenFlow or with anything else that you have. Yeah, so Neil, I'd like to get your take on VMware. So obviously then I see your acquisition really, but just put SD on on the map and everybody was talking about it. Martín Casado, we love having him on the Cube, really important person in the industry. However, I was drilling him a little bit at the Open Networking Summit because he's gone from being a full supporter of everything open source to now talking open interfaces. I happened to be able to ask a question to Pat Gelsinger, the CEO of VMware last week and I said, Pat, it seems like you're moving further away from open source, what's your position on open source? And he said, we're not changing our business model. Open source is not a primary thing that we look at. So with that context, where does VMware sit in kind of an ODL and open stack world? Right. I mean, I should start by saying I spent seven and a half years working for VMware. So I know those guys and I have a tremendous love in working for VMware. I've been a Vex for the last four years and I've worked with them since 2002, so. I think a lot of what you say is accurate. VMware, unlike most other companies, was able to get a platform, in a sense in a full proprietor without building a community around it. Because they were the first with vSphere, because there was nobody else out there, a community formed around them. And so in many ways, you can look at VMware as being this decade's Microsoft, in a sense. And that's been a fantastic platform. I think they're missing something. And I'll be honest, I was part of missing it in the early days. When I was asked when I was at VMware, as part of the group, we were looking at OpenStack. And I said, that can't work. I mean, come on. And I was completely wrong. And so was everybody else there. And I personally think they're missing a major shift in the industry. And we already see it. NSX is a great piece of technology. I think there are many good things around vCloud Director, but by keeping everything closed, by assuming that 10 people in a room can get the right answer and everybody else is going to anoint them and agree, you're the platform and I'm going to work with you. I think it's naive because in networking and frankly in cloud, unlike in virtualization, they're not the only game in town. There are many players who have a whole set of really intelligent experts and who have great hardware and relationships. And really the game is now moved to a place where you have to be able to work well with a whole bunch of players. And so I would like to see them. I think within the Sierra, they've got a great DNA that knows open source. I think Martin certainly knows open source, but the old style VMware needs to have more education. Yeah, and just to VMware's credit, they do have 30 people here. I know they've got folks working on Nova. They do. They've got developers here and absolutely you can do ESX inside of OpenStack. Just something, by the time we get to VMworld, maybe their story will have matured a little bit. I think they're evolving. I think they're going in the right direction. Personally, I'd love to see it happen a little more quickly, but yeah. Yeah, so the other thing I want to get your opinion on, people are looking at OpenStack and saying, we're four years into this and how are we with the maturity and adoption? I know I've had conversations with you. ODL really has, I forget what release number it's on, but when you talk about the first release, but if you say, is that something I should be able to put production in news everywhere and scalable and everything else, can you talk about how long it takes maturity and what proper expectations should be, especially something that's an open source project with so many people working on, but both from an ODL and an OpenStack standpoint. I mean, to start with OpenStack, I think that people have put unreasonable expectations on OpenStack. It takes a long time to build a platform and to build a great platform. I've never seen great technology get invented overnight. And the more moving parts, the more it touches critical parts of your infrastructure, the more challenging is to put all the elements together. And so if I look at OpenStack today, you've got real customers using it. Now, is it surprising that it hasn't hit 100% that it's mainly early adopters, clapped by people who really have a pain as well as have capability? No, that's in a sense what I would expect. VMware didn't get to where it is overnight either. Remember, it took 10 years and it was being used in test and dev and some limited environments for a long time. So on the OpenStack side, I think we're all waiting until, suddenly you get that inflection point and a million users start showing up at these things. I think it'll come. And what's important is to make sure you're going the right direction and got solid technology and components. I think the same is true in Open Daylight. And frankly, we're a few years behind where OpenStack is just because we got started later. As you said, we had our first release. And I want to remind you that our first objective is to create a platform that the industry will pick up. The first problem we're solving is there are 40 controllers out there. We want people who are in the industry who may be already having a controller or thinking of building a controller to build around Open Daylight. Because when you do that, you get the benefits of interoperability. You get the ability to test against a wider range of things. Nobody can own all of that. On that, great success thus far. Cisco shipped a product, IBM shipped a product. People like Sienna have talked about we want Open Daylight at its heart. Other people haven't necessarily shared specifics, but the HP announcement of HP doubling down on VMware can clearly get a sense of a shift of how strategic and important it is for those companies. So on that, really doing great. From an end user, your question becomes when will end users pick up a mostly open source solution as opposed to a vendor supported solution? Well, I'll start by saying 90% of users want the vendor supported, not a full open source. But I do think it's actually very important. Today we have end users who are using Open Daylight. Their primary use case is being able to answer big strategic questions about the network. So if you're a network architect today, you need to be able to answer overlays. Are overlays the future for us? What does it mean for our network? How would we manage it? What about white box switches? What about cumulus? Should we go do that? Are we going to go something like the Cisco APEC? There are these major Uber questions. You're being asked to be able to answer that five year strategy. And doing a POC with any one vendor doesn't help you with that so much because at the end of it you're asked to write a check. So people are using Open Daylight today to be able to evaluate overlays and understand it and to look at Open Flow and test Open Flow and that's great. I wouldn't personally put Open Daylight into my production network unless I had people who were really, really savvy and very connected to our community because our first release was a developer release. And then there's a tension in terms of when will it come? There's a tension between getting more community involvement and adding more elements and focusing on performance stability of the core. And we'll see how that shakes out. The TSC, you've got Dave Meyer next. I urge you to ask him that question. So, good setup, thanks Neil on that. So when we had Dave Meyer on it, Open Networking Summit, he said for this first release code is coin of the land. So we could look at who's contributed. We can look in there and say, we know who from what company is doing what. So that shows the participation here. As we look out over the next year, how do we really help people to see who's really involved? Is it the solutions that they ship and actual revenue? Or how should we be thinking about measuring and what metrics should we look to over the next year? I think you're asking the right question. I'm so glad because I'm often used to, people often say, how many customers do you think we'll be using the pure version of Open Data? And I'm like, again, we'll see some of that, but that's not necessarily the right metric. I think community involvement is really important and you have a number of ways. You brought up some of them. One of them is looking at how many engineers. Another one is looking at the projects. I've been really, really happy to see Hewlett Packard seeing Citrix participating and submitting new projects that we have right now. It's so great to see something like Group Policy not being dominated by one company, but having four companies participating in that. So I think I would start by looking at those two. Another one, which you might not think of, but in open source, committer status is something that's really important. And so as you see people upgrading to committer status from simply contributor status. So I think that's certainly an important one. The other thing to look at is who is doing the dirty work? It's been great to see Ericsson, for example, invest in the testing infrastructure. Huawei right now has announced that they want to help us with the testing infrastructure. And so it'll be interesting to see how many resources are truly dedicated. How involved do they get? Were they looking for a press release or are they really invested in making sure that the quality of the next release is there, that the feature set is there? And so some of that involvement is gonna come out. And obviously looking at people who are, especially the big companies, looking at people who are silver today, who do make the step that HP has. Very publicly, it's hard to say to your internal teams an open source project is gonna be core to our strategy. So brocade going out and hiring some of the best talent available out there. That's undeniable. Looking at HP, going out and signing a million dollars, minimum of 10 FTEs. That's great. Let's see what they start announcing beyond that. I'm going to be watching them to see, again, was this a press release or are they for real? I think they are, but we'll all have to watch that. All right, Niela. So I want to give you the last word on this. So for our users, especially those that might not be, they're not contributors or anything, why is this an important time? Folks out there don't know their SDN from their DNSs. Why is this a critical juncture in specifically the networking space of the IT community? Think of yourself back in 2004. Someone mentioned a virtual machine and you had a choice at that moment. You could think of all the reasons why it wouldn't work. Overhead, physical machines, nobody's gonna share so on and so forth. Or you can say, this has the potential to change IT as we know it. I'm gonna invest a little bit of my time over the next few months, over the next few years to seek to understand what is a VM, how does it work and how might it impact my job? I can tell you, if you look at those people's career, they're at very different places today. In the same way, SDN is fundamentally going to change all of your careers. And Open Daylight is at the heart of most of the major players' efforts in that. And so I urge people to get to know Open Daylight whichever way they're able to. If you're technically inclined, certainly going in and looking and downloading, if you've got a lab, we've got a lot of people, like Greg Farrow, for example, who pick it up, play with it. If you've got those skills, great. That's a great way of learning. Others may simply be to ask the vendors that you have what they're doing. And in between is to start doing some POCs as part of your company on something like OpenFlow. And Open Daylight's a great way of doing that. All right, Nila, thanks so much. Really appreciate coming up. Definitely recommend everybody check out ODL. It's been rapidly changing, especially in the networking space where it typically will take decades for things to move. We expect that timeline to shift. So this is Stu Miniman with SiliconANGLE TVs, live coverage from OpenStack Summit Atlanta 2014. We'll be right back with our next guest.