 It's theCUBE. Here is your host, Jeff Frick. Hi, Jeff Frick here. We are on the ground in the heart of Silicon Valley. The Open Daylight Summit event, it's their second annual event, about 800 people talking, open networking and open daylight software to find networking. It's been a long day. Morgan Kiyosi is joining me. She's a distinguished engineer from AT&T. She's having fun on theCUBE. Welcome. Thank you. No, you've got me going. I got you going. So AT&T, what are you doing here at this open networking event? Well, AT&T's decided for 2020, we have a direction that we're calling Domain2.0. And the goal is to virtualize, I call it SDIN Control Enable Cloud, our network, our targeted network, 75% by 2020. And we're actually on track of virtualizing and putting it on to a SDIN Control Network. 5% by the end of this year. 75% by 2020 and you're 5% by the end of this year. So we're definitely going to meet that. The reason why we got into this technology was, our biggest hurdle AT&T has was creating new services quickly. So I know a lot of folks focus on maybe costs in this space, but we really focus on cycle time. So instead of taking 18 to two years, 18 months to two years, our goal is to do it within six months. In fact, or very quickly, as quickly as possible, of instantiating new services, of creating new services, of even deploying infrastructures. In fact, our network on demand services, which is actually based on this technology, basically went from concept to delivery in like, well delivery in beta for customers in about six months. And it was actually due to a lot of this technology. So it's definitely strategic to us. And this conference focus on open daylight, which is more focused in a controller space, is critical for us. We actually segmented the controller world into two spaces. We have a global controller, which is more this end-to-end global, end-to-end controller that sets up networks end-to-end. And then we have concepts like local controllers, which are more specific to very specific network elements that might be unique, and they might have different nuances that a global controller might not have the performance that's necessary. And so we see the combination of a global controller and a local controller sort of fitting into our design. And we see ODL fitting in both places. So AT&T has already rolled out a SDN global controller on open daylight, which is why we're here and why we're participating. In fact, we just became a silver member. I think there's only like two users, or maybe there's more users now that are members of ODL. Because like I said, we're using this in our network. We're working with it big time. We're feeding back into the organization of what we not want, what we don't want, and what's working, what's not working. So we decide we really need to join and we just need to basically step up our presence. We also see ODL also in our local controller area. So we're looking at things like in the optical world, this concept called an open Rotom, and we're actually going to be introducing an ODL controller in that space. So open source and AT&T. What role does open source play? Are you guys active contributors? Are you just trying to take advantage of participating? Is it really more the standards? Is it the speed of innovation? You know, what's just being involved in an open source project like ODL mean to AT&T? Well standards will never go away and there's definitely a role for standards. The difference might be now is we don't see standards defining everything. We need to get into implementation faster. And so code is actually, we're finding helping us better. Once you start migrating to a software, and now it's just software, talking to software, you start getting to APIs. And you know, bringing code to the table to flush out those interfaces, our view is a lot faster way of doing it. So we're not like building hardware. So when you build hardware, you might need more standards because you can't just go experiment and innovate faster because it's hardware. So you really have to have the discussion and work everything out before you build. But in software, it's actually going to be more agile. So you can actually then create these APIs and interfaces more quickly through code. So that's why we view open source as important, code as important. We are getting, definitely getting more and more engaged in open source. We have decided, we've pivoted our company big time where we're going to be doing development, where we are doing development. We're actually building more and more of our platform ourselves versus buying. And so because we're getting more and more into the development ourselves, then we're also now creating a base that can of course then be more effective at working in the open source community. So some of it we're working with vendors. Some of it we're doing ourselves. Like we actually submitted into Open Daylight a Yang model design studio because we believe in Yang as a modeling language for this space and we as users, we have like about a hundred users in our company designing services through Yang. And a lot of them aren't really programmers. And when you try to define things through Yang, there's all these brackets and all this indentation which is very common in programming languages. And we didn't want these folks who were like pivoting to this to sort of get hung up on the whole formatting but to really focus on the design of the model. And so we created the design studio so we could make it easier for them to create things. So we actually contributed that to the Open Daylight. In fact, we only showed it to them like a couple of weeks ago and the users of ODL that were on that call immediately wanted to try it out because they felt it was actually one of the best tools they saw on the market even compared to what you can buy. So we are definitely contributing back and hopefully by this contribution people will help us design studio better and so forth. Talk about the business impact of being able to deliver services and roll out new services. You said typically before it would take there's a two to three years and now you're 18 months to 24 months. 18 months to 24 months and now you're down to six months. Is that, how much of that's being, how much of that's a stick and how much of that's a carrot from the marketplace? How much is it coming internally for really trying to roll stuff up? Or is it, oh my gosh, we have to roll stuff out faster because everybody else is? I would say it's a little of both. There's more of our internal structure to get things to go faster. And of course, when you deliver things faster then you don't tie up your resources so much, right? You can now, let's say it's every three months. So just think about it, every three months you basically free up those resources to do something else every three months. But it's not as easy to say let's just do open source, let's just do software and just all magically happens. It's the, you have to have the right building blocks to make it programmable so you can instantiate things faster. And so there's a specific type of design that has to be thought through that we've been working on within our company so that we can create these building blocks so that you can instantiate faster. So that's one thing. One is to create the service faster, freeze up resources so we can create other services. But the other is we can experiment more quickly. So as we, so with the never gone demand we experiment, we had a vision of what we wanted. We started prototyping. We started getting, started doing a beta with a customer in Texas. And we, based on their feedback we started iterating very, very quickly. So also from creating new services we find we can create services faster that fit to the needs of the market. And so you're creating more services faster and you're getting them out. So then what's next? What does this going to enable you to do in terms of new businesses, new opportunities that you couldn't do before? Well, so once we, once you start getting to this whole cloudification that we call it, we're not looking at just AT&T COs. We're looking at every place AT&T has a presence that includes even the customer prem. So you're talking about hundreds to thousands to 10,000 to maybe a million sites. And can you imagine, once you have this infrastructure in places which can enable all these applications then it comes down to the imagination services. And I believe that there will be new services that'll be very latent sensitive, which will cause the need for wide distribution of that application and who else but like an AT&T would have this points of presence of a million sites. So our vision is we'll have a platform that can create new services we can't even imagine. Okay. So then what's your next hurdle to get from, what's between five and 75? Is it just implementation methodology and just getting it out? Is there any big technical hurdle that you have to get over, regulatory hurdle? What's, if I come back and talk to you next year, what will you have done that's really a priority in the next six months, nine months, a year? So I would say when you put this whole platform together, now that you take the slide where you try to do this vision and you try to use open source and I would say the biggest focus we're going to have the next 12 months, not longer is actually getting all these different pieces of open source working together, which is why we created this OP NFE, open platform for NFE. Because in the end, just having open stack by itself is not sufficient. We need open stack, I mean, for example, we need open stack with ODL or maybe onus with maybe DBDK, OBS, it goes on and on and on and the whole thing has to work together. And no one's putting it together. Everyone's working in their own project, optimizing in that world. But when we as end users needing this platform put it together, it doesn't work as efficiently as we'd like. And so our real focus is to try to get things to work more efficiently together for this whole thing. But that's always a challenge, right? If you do it by yourself, you've got total control, you can optimize to whatever you want to optimize, you throw it out to a larger group. You know, you're kind of trading your choice of optimization based on what kind of the group decides where they want to go or their prioritization doesn't really fall. But I assume the benefits still outweigh the cost. Well, OP NFE is really was created by a set of carriers. We're trying to get enterprise folks to also get engaged in it. But from an end user point of view of basically creating a platform we're all going to use. In the end, what does AT&T's goal? Our goal is to make money, provide services for our own customers. The customers don't care whether you're on a virtualized platform or nesting a controller. They really don't care. And then when- It's got to work, right? Small bell, it's got to work. Yeah, and the next thing is you have these virtualized applications. They also want the platform to work. They just want to focus on their app. They don't want to focus on the platform. But what happened was when this whole industry got churned in this area, it dragged us all down into worrying about the platform. My, as an end end user, trying to just provide services, network functions, just trying to focus on applications now get dragged in. And all the platform people, they're all trying to get it to work together, which shows it's very hard. If it was easy, the platform people would have done it. The network functions would have just focused on that. And I would have just been able to, you know, create my services. But instead we're all getting dragged into this because it isn't working and it's not that easy. So, OPNF's goal, hopefully, is to get the industry together from a networking point of view, users, suppliers, vendors, you know, so forth, to just, I'd say fix it so that it works so we can all get on with life and get back to our core competencies. All right, well, Margaret, thanks for stopping by. All right, I'm Jeff Frick. You're watching the Cube Word Open Daylight Summit. Thanks for watching.