Even as deployments of Converged Cable Access Platform (CCAP) technology start rolling out, there are already signs that CCAP -- as it was originally conceived -- may not provide an adequate response to the evolving demands on cable network infrastructure. Simply put, a completely converged architecture may no longer fit the bill as cable operators make the transition to IP video delivery. (See Life After CCAP.)

In an exclusive interview with Light Reading, Jeff White, chief strategy officer and a co-founder of stealth technology startup Gainspeed , shared his company's views on the near-term progression of cable network infrastructures. And White (pictured below), whose firm is promoting a new "virtual CCAP" approach, makes the case that cable operators may need to re-think how CCAP is deployed. The following transcript was edited for length.

Figure 1:

LR: Very broadly, how do cable infrastructures need to evolve to support the continuing transition to IP video delivery?

White: When you're looking at this, you have to put it in context of what's happened over the past number of years. The cable operators have been wildly successful. Both on the video side and the data side, they've just had huge, huge success. Just to give you an example, on the data side, cable terminates over 83% of all [US] connections that are greater than 6 Mbit/s. In the US, they have coverage of over 93% of the homes. So they have vast coverage and they have huge pipes.

The challenge for cable guys with all this success is that their world has kind of been turned on its head over the past five years. And if you look at what's happened, we all know that the driving factor for Internet capacity growth has been the over-the-top guys.

In the past, a measured evolution has worked for the MSOs, but it can't work going forward. When you talk about this transition to IP video delivery, we don't believe that it can be standard operating procedure. [Cable operators] have to take a completely new look at how they're doing this because they need dramatically more data capacity. We just don't believe that today's solutions can scale -- even CCAP -- we just don't believe they can scale technically or financially.

LR: How do you think CCAP deployments will play out in 2014?

White: I think that operators will deploy CCAP where they really have density and capacity problems and they just can't wait. But we're really in the very early stages of these deployments, and I don't think we're going to see widespread deployment.

We've spoken to pretty much every major operator in North America, and many in Europe and South America and Asia, and most people we're talking to today are talking about distributed architectures. We think there's been a very clear mind shift around the concept of the architecture for the access network, and we think that they're working through what the right strategies are today.

In Gainspeed's vision, [the right strategy is] a virtual CCAP. It's leveraging distributed architectures, leveraging SDN [software-defined networking] and network function virtualization [NFV] type technologies to virtualize CCAP.

LR: What gets distributed in a distributed network?

White: There are a couple of options for what you would actually distribute. Remote PHY is just distributing the PHY to the edge of the network. [But] Remote PHY unnecessarily separates the DOCSIS MAC and the DOCSIS PHY, and that creates potential issues for timing synchronization. Just as importantly, it leaves the CMTS [cable modem termination system] as a physical box in the headend, which is probably one of the most costly and cable-specific devices in the headend itself.

[On the other hand,] in the virtual CCAP solution, in addition to pushing the PHY to the node, we push all of the DOCSIS to the node. And by doing that, we put the DOCSIS where the coax is, which is, in our view, the logical place for DOCSIS. By doing that, you eliminate the physical CMTS completely. It turns the link between the headend or the hub in the node into a fully standards-based Ethernet. Then you can do a lot of unification of the fiber transport for residential services and business services.

The benefit of all is this is that you can eliminate all the cable-specific equipment from the headend, and the headend fundamentally becomes fully digital and Ethernet-oriented.

LR: What do you see as the path forward towards this type of architecture?

White: Obviously [operators are] not going to go ripping out equipment from headends across the country, and they weren't going to do that for CCAP either. I think in either case it's going to be a cap-and-grow type approach initially, and then a migration approach.

LR: What happens in 2014? What are your predictions?

White: I think in 2014 you're going to see broad-based acceptance of the concepts of distributed architectures [as] an enhancement to the concepts of CCAP. I think you're going to see a huge amount of communication about IP video and companies moving in the direction of IP video. I think you're going to see much higher data service offerings. I think you're going to see more of those across the board from the cable guys.

LR: What should we expect from Gainspeed in 2014?

White: We're not ready to talk yet about our specific products and solutions, but we're a huge proponent of virtualizing CCAP and what we call virtual CCAP. I think you're going to start hearing this term more and more as we drive forward. [We're going to] continue working with partners around an ecosystem around virtualizing CCAP -- guys like Juniper Networks Inc. (NYSE: JNPR). We work closely with Netgear Inc. (Nasdaq: NTGR), JDSU (Nasdaq: JDSU; Toronto: JDU). There are a number of different partners that you'll see us working with.

LR: Will you be launching actual products this year?

White: You'll have to wait and see.

— Mari Silbey, special to Light Reading