Tom Stanton admits that it’s a bit early to implement SDN in the access network. But, he says, Adtran wants to have a product set in place that can handle that transition.
Accordingly, Adtran is guiding its service provider customers on their virtualization path. The vendor recently introduced Mosaic, a software-defined Access (SD-Access) architecture that combines Web-scale technology with open source platforms to accommodate multi-vendor environments. Adtran is also advancing G.fast and NG-PON2. Earlier this summer, Adtran was chosen as one of two vendors to take part in a NG-PON2 lab trial with Verizon.
Sean Buckley, senior editor of FierceTelecom, recently caught up with Stanton during the Adtran Connect event at their headquarters in Huntsville, Alabama, to discuss the evolving broadband access market.
FierceTelecom: A big initiative for Adtran is the Mosaic SDN platform. Can you talk about how your Mosaic SDN vision fits into the access network and how it gives service providers a migration path for service providers moving to SDN?
Tom Stanton: In hindsight, we’re probably pushing SDN too quickly. With the G.fast piece, which was our first SDN-controlled access product line, we initially came to the market with these software pieces after the fact. It turns out they’re not hard to do, but they have to be done. We’re incredibly early into it. Anything we’re bidding today has an SDN element to it because we’re trying to use as many of the modules as we have, but I would say most carriers today don’t have their network in a state to utilize the functionality that we have. In most cases we’re providing a step path for them to be able to do SDN in the future.
FierceTelecom: As part of the service provider SDN vision, AT&T and others are advocating central office redeployed as a data center (CORD). How do you see CORD playing out and how is Adtran helping customers take advantage of it?
Stanton: Right now, we have the only CORD operating access architecture available today. Some of that is related to us making sure we’re ahead of the curve. The dream of a data center-centric CO is what we’re chasing. There are companies that are afraid of it. Some companies are bringing scale of the same function. If you think of routing, you make bigger routers, but they’re routers. Really, the things we think we’re bringing in being moving a central office to a data center is fundamentally different. We know a lot about the TDM network moving over to IP, and we knew fundamentally how to make that change from ATM to Ethernet. There’s an awful lot of know how that has to be packaged in order to convert a CO into a data center. We think people coming from the data center are going to face a difficult learning curve.
FierceTelecom: Are you seeing more of your customers moving to test and implement CORD?
Stanton: They talk about it. There are two things: One is collapsing COs by getting rid of Class 5 switches, but all you’re really doing is aggregating those in soft switches and duplicating the same thing. Then you really see other carriers that are trying to transform operationally how they do something. When you go from hardware to a software Class 5 switch, you don’t operationally change anything because you’re just moving that functionality. When you’re talking about be able to scale for major increases then you have to fundamentally change something.
FierceTelecom: Okay, so what are the benefits for CORD and virtualizing functions in the CO?
Stanton: Scale. If you take a look at how long it takes to provision an end-to-end circuit today, there will be dramatic changes to that. There will be substantially less vendor specific software elements that you have to talk to. The network will look a lot flatter and will have a single plane that’s vendor agnostic. In being able to do that you’re pushing a lot more intelligence into the network. Every time you push intelligence into the network, you’re taking away a click; you’re taking away some provisioning you used to have to do manually. It’s all about speed and the amount of man hours it takes to deploy services.
FierceTelecom: Besides CORD, Adtran is moving fast with NG-PON2 and has been invited by Verizon to take part in a lab trial. How important will the transition to NG-PON2 be for Verizon and other carriers?
Stanton: When you think about what NG-PON can do when you have a pair of wavelengths for potentially different uses. Think of where that would set you up in the wireless world where you’re trying to deliver a gigabit to a 5G node. Think about how many 5G nodes there is going to be. It’s a completely different world. All of a sudden you’re trying to deliver a gig to different 5G nodes, and potentially different customers on each one of those wavelengths, it makes sense. If you were doing it just for residential service, there’s probably cheaper ways to do it. When you’re talking about a differentiated carrier service by wavelength it makes a lot of sense. For Verizon it’s a good changeover point. If you look at GPON today, we have more activity around 10G PON. It’s either fixed 10 Gig or NG-PON2.
FierceTelecom: Fiber is clearly the end game for the last mile, but new technologies like G.fast are becoming game changers for the copper network. Where do you see G.fast playing a role in the last mile network?
Stanton: I would say 10G PON is right now not being deployed in any meaningful sense right now. You’re really looking at 10G PON as something being deployed in coincident with significant 5G roll outs. I think the timeframe for that makes sense. You’ll see 10 Gig services to businesses.
Then you have the problem of, what do I do for my fixed network for residential customers? That’s where all of a sudden the copper pieces make sense. One of the big differentiators we have is with NG-PON2 (and) G.fast. While vectoring is passé we’re the only vendor that has a major deployment in the U.S. with VDSL2. That’s now just starting to roll out and you’re just starting to see people adopt 100-150 Mbps services for those customers. We have those bases covered and we bring those all under the SDN-capable umbrella.
If you’re looking at trying to sell to a carrier, you can go and say: I am the G.fast vendor or I am the best NG-PON2 vendor, but what they really need is an access vendor. They need someone that brings all of these pieces together and allows them to move that network forward.
FierceTelecom: A number of telcos including AT&T, CenturyLink and Frontier see G.fast being applied in multi-dwelling units (MDUs). Do you think that’s the best place to use G.fast?
Stanton: Whether G.fast makes sense for single family units is dependent on your particular usage case. With MDUs there’s no reason not to use G.fast. If you look at the G.fast solution we’re developing, it is coax-fed or copper-fed. You can go into any MDU and if there’s coax or copper there you can plug into that, but there’s no reason to not use it, especially when you can deliver gigabit speeds. Why would you go and rip walls down?
The real change we’re going see is what happens when the MSOs offer DOCSIS 3.1. Everyone can talk about fiber today, but if you’re in a hurry and you have a competitive threat in more than one city, you’re going to be looking at the fastest and cheapest way to get a gigabit competitive service out there. That’s where I think you’ll see the usage case for G.fast change.
FierceTelecom: Comcast and Cox are moving fast on rolling out DOCSIS 3.1 over existing HFC, but are you hearing them ask for fiber-based solutions?
Stanton: There’s not a single major or even minor MSO that isn’t looking at deploying PON technology typically in the 10 Gig vein. They’re already doing it today for business and even in residential. In any Greenfield market, you’re going to see them deploy PON today. If you think about the deep fiber initiative with DOCSIS 3.1, you’re pushing the node out to get that higher speed and buying yourself some time to recoup your capital. Before you do that last leg, is exactly what G.fast is. It’s just the telco version of a distributed node architecture.
FierceTelecom: Let’s switch gears to talk about the competitive landscape, which now has a few key players in Adtran, Calix, Huawei, Nokia, and ZTE. How can Adtran continue to stand out from the pack?
Stanton: Well, fewer competitors are always better. It’s getting hot again so it will be interesting to see how some of these other vendors decide. Ericsson has been in and out of it and Cisco has been in and out of it many times. We like the space right now and like the competitive environment right now.
We’re not the only ones seeing the shift towards the access space in the service providers’ capital budgets. If you look at a 5G network, it is going to be controlled by the same controller and architecture they are using on the fixed space going forward. You’re going to see a media-agnostic architecture by these carriers and 5G remote node will be just one of the ways they can deploy it.
FierceTelecom: A number of your service provider customers are moving forward with leveraging funds from the second phase of the FCC’s Connect America Fund (CAF-II). What’s the status on that for Adtran?
Stanton: Some are using this as a way to deliver higher speeds. Some carriers are saying, 'I can deliver 10 Mbps and deliver a service that’s viable, or I can deliver 25-50 Mbps and offer differentiated services like IPTV.' The cost differential between those two in a lot of cases is immaterial. They are being opportunistic about it.
Most of the big players have started rolling out CAF-II already. In every single one of these cases these service providers are rolling out CAF-II, but that’s not the only thing they are doing. In the overall scheme of things that’s one element. If you think about three or four years ago, they were doing nothing on the access side and at best were happy with 10 Mbps.
The whole mindset in those companies has changed. A lot of them said I will go and offer data center services or go after business services. Others said, 'why don’t we take our customer base and offer them something they’ll pay for?'