New Orleans -- SCTE/ISBE Cable-Tec Expo -- Charter Communications and a handful of suppliers joined hands here this week to detail the progress being made on a fiber-node initiative that will standardize the node housings and the interfaces for various service modules that can be plugged-in like Lego bricks.
From an operational perspective, this evolving SCTE/ISBE standard -- called the Generic Access Platform (GAP) -- could make a ton of sense for cable operators. Rather than having to deploy or rip down discrete, proprietary nodes from individual vendors for FTTP, DOCSIS 3.1, WiFi, 5G, wireless backhaul and so on, operators could instead go with one standardized node housing that can snap in or discard those modules as needed. The broader idea is to extend key functions of the MSO's HFC network toward the edges while also cultivating a vibrant supplier ecosystem that could pave the way for new and incumbent suppliers to make the housings and the individual modules and compute functions.
The GAP standards effort was announced last year, with some initial details about the initiative and its aims presented at the 2018 Cable-Tec Expo in Atlanta. This year, Charter Communications, a service provider champion of the initiative, and a batch of suppliers updated the industry on recent progress, including a functional proof-of-concept (PoC) that was on display at various booths.
While the aim is to have a large number of suppliers involved, the GAP PoC shown here was assembled by a group of five suppliers, with each supplying a piece of the puzzle: ATX Networks (the GAP housing with DOCSIS 4.0 capabilities for a 1.8GHz cable network); Cisco Systems (software for a virtual converged cable access platform); Intel (compute); Applied Optoelectronics (the remote PHY device); and Silicom Connectivity Solutions (an Intel Xeon-D-based main board and midplane for FPGA modules).
A year ago, there were concerns, particularly among vendors, about GAP and how the effort could further commoditize the node, shave product margins and reduce the ability for vendors to differentiate. There were also concerns among suppliers about whether enough cable operators would back GAP at a scale to justify the investment. Although GAP could help some suppliers gain market share as the standardized nodes are adopted and deployed, some incumbent node makers, naturally, view GAP as a possible encroachment on their territory.
Power and heat are among the other barriers facing GAP, with concerns about whether a standardized housing could properly dissipate the heat generated when a GAP node is packed with compute and power-hungry modules.
While there's no doubt that some of those concerns remain, this year's Cable-Tec Expo was a clear indicator that the GAP standards initiative has made significant progress since last year's show in Atlanta.
The initial GAP standard isn't expected to be completed until Q1 2020. But execs with Charter and the GAP node prototype vendor partners held a session here this week to update the industry on the effort.
As for the GAP node itself, the current design effort envisions having both the lid and base sides remain modular. Under the current plan, the lid half will contain the "intelligence" while the node base will contain the powering components and the interfaces for the various modules. Additionally, the compute element will now be a module rather than a piece that is woven into the GAP node's high-speed backplane, Ed Dylag, the market development manager at Intel's Networks Platform Group, explained.
Matt Petersen, VP of access architecture at Charter, said GAP is attractive to cable operators in part because it standardizes a node that can be deployed once, but can then be continually expanded and enhanced with the snap-in modules. That, in turn, will also reduce development cycles for vendors and accelerate deployment times, he reasoned.
The initial top use cases for GAP, at least from Charter's perspective, will likely center on the deployment of remote MAC/PHY modules that support new distributed access architectures (DAAs), Petersen said. But the same node could also snap in modules for other services offered today or on the roadmap, such as WiFi, a remote OLT and even CBRS radios, he added.
And while the module interfaces are standardized, the modules themselves are not, allowing room for suppliers to innovate and differentiate, Jay Lee, chief technology and strategy officer for ATX's Broadband Access unit, said.
That modular approach will also open up a node ecosystem that has largely been proprietary. "That's what's beautiful about it," Elad Blatt, chief strategy and business development officer at Silicom, said. Standardization will also shorten the time-to-market for new GAP modules significantly, he predicted.
The opportunity for suppliers is "immense," enabling them to "pick their spot" and differentiate on those elements, Sean Welch, VP and GM of Cisco's Cable Access unit, said.
Still, space, power and heat issues will need to be further resolved for a fully packed GAP node. Last year, some engineers on the sidelines wondered half-jokingly how the GAP nodes would support a multitude of modules without melting.
"You have to fit a lot of stuff into this box," George Goncalves, business development director at Applied Optoelectronics, allowed. "It's going to force innovation... to design smaller and to design with less power."
Dylag said the orientation of the modules with the lid size will help in that area, as modules that plug into the backplane will also come into contact with the actual housing to help dissipate the heat.
Even as a standard, GAP will likely require a degree of integration work, possibly opening up an opportunity for a "new entrant" that can facilitate those efforts, Petersen suggested.
From an MSO perspective, Charter is the only cable operator to put its full weight behind GAP, and it's doubtful that Charter will be able to carry the market alone. More operators (and suppliers) will need to get involved and commit for the GAP ecosystem to truly scale.
Intel's Dylag said multiple operators are interested in GAP, but acknowledged that some are still in the "wait and see mode." Cox Communications and Shaw Communications are among the other major MSOs that are active participants in the GAP working group.
In all, the GAP standard working group totals about two dozen companies and organizations. "I would invite them to become more active," Dylag said, hopeful that showing that a working prototype is already achievable might spur more active participation from other suppliers.
Petersen was also questioned about what market dynamics will justify the replacement of legacy nodes with standardized GAP nodes. He noted that DAA efforts will require a new platform that support these new remote PHY devices. "It's more of a cap-and-grow style [deployment] as we drive DAA out there," he said.
Jeff Heynen, research director, broadband access and home networking, at Dell'Oro, said GAP "on paper, sounds like a great idea." But he believes more operators will need to get behind outside plant upgrades that would push the spectrum ceiling to 1.8GHz, which will be supported in the developing CableLabs DOCSIS 4.0 specs.
GAP "is interesting but will need a consensus from a larger group of operators that this is the direction we want to move in," Heynen said.
ATX believes that more operators will take advantage as GAP continues to evolve and develop. "It's going to happen; I think that's the reality," Lee said in a pre-show interview. Managing five to six proprietary nodes on the network is "pretty challenging" for an operator, he added.
GAP has become a "centerpiece" initiative for ATX as it has engaged more closely with Charter and other operators interested in the project, ATX CEO Charlie Vogt said in an earlier, separate interview.
ATX recently hired several Cisco engineering developers in Lawrenceville, Ga., to accelerate its product development capabilities specifically aligned with the company's evolving DAA cable access networking portfolio.
Lee said the GAP housing unit shown at this week's show in New Orleans is close to what the finalized standard will call for, noting that "refinements" for the enclosure are being ironed out.
Dylag said the working group is close to publishing a CAD (computer-assisted design) model of the GAP housing. "We have some work to do to hit that end of the year [target], but I think it's doable," he said.
The initial GAP standards effort is largely focused on strand-mounted nodes that are common in North America. There's also interest in developing a variation of GAP that would support housings that are more common in other markets, such as Europe, that can accommodate the individual GAP-compliant compute, power and service modules.
- SCTE/ISBE Standards Group Creates Generic Access Platform Project
- Using a Lego Approach to Cable Node Upgrades
- HFC may never die
- DOCSIS 4.0 ends cable network feud
— Jeff Baumgartner, Senior Editor, Light Reading