Comcast Proposes Its God Box

DENVER -- Cable Next-Gen Broadband Strategies 2010 -- With an expected flood of unicast services on the horizon, Comcast Corp. (Nasdaq: CMCSA, CMCSK) is developing a super-dense, do-it-all network device that significantly cuts headend space and power consumption requirements and trims down overall costs per bit.

Jorge Salinger, Comcast's VP of access architecture and today's opening keynoter here, said the MSO started developing that product -- the Converged Multiservice Access Platform (CMAP) -- after determining that it would soon have to support more and more narrowcast services, including those fed by Docsis (for traditional high-speed data and VoIP), as well as video on demand (VoD) and switched digital video (SDV).

Comcast typically uses about 10 narrowcast QAMs per service group (six QAMs for VoD and four for Docsis). With the expected addition of SDV and an anticipated surging demand for other unicast services like VoD, Salinger sees that QAM requirement growing rapidly to 30 or more.

The challenge, Salinger said, is how to squeeze all those new CMTSs and edge QAMs into already-limited headend space without undergoing expensive expansions or sending power demands through the roof.

Comcast hopes the CMAP will address that challenge by combining many discrete functions -- including the cable modem termination system (CMTS) and all the edge QAMs -- into one device. Salinger said the CMAP is the "heart" of the MSO's broader Next Generation Access Architecture (NGAA) project.

The CMAP will also include PON for business services, plant sweeps for plant health monitoring, and video encryptors. Two things it won't do (at least not initially): process video streams or perform ad splicing.

Because so much is being packed into the CMAP, the services that run off it could be more exposed to potential hardware failures. Comcast hopes to mitigate that risk by requiring redundancy for every line card that graces the CMAP.

Specs nearing completion
Salinger said Comcast expects to complete the CMAP hardware specification within the next month. A second specification for modular implementations will call for the development of an Access Shelf (for the downstream and upstream PHY and MAC layer) and a component called the Packet Shelf (for packet processing). That spec will be completed within the next few months, Salinger said. A third specification focused on configuration and management should be completed by mid-2010. Together, these represent Comcast's view for next-gen CMTS and edge QAM products.

Comcast is not forcing vendors to implement the Access Shelf and Packet Shelf together. One vendor, for example, could develop the Access Shelf, while another creates the Packet Shelf. However, those vendors would still need to integrate the pieces using a common interface. Other vendors may develop both access and packet functions into a single chassis.

Salinger said Comcast is talking to a wide range of vendors, but as examples, he cited Alcatel-Lucent (NYSE: ALU) and Juniper Networks Inc. (NYSE: JNPR) as candidates for the Packet Shelf, while companies such as RGB Networks Inc. and Harmonic Inc. (Nasdaq: HLIT) could develop the Access Shelf. Other suppliers might work on both shelves, or include both functions in a single device.

Chipmakers Broadcom Corp. (Nasdaq: BRCM) and BroadLogic Network Technologies Inc. are also involved in the technology development, Salinger said.

"Our intent is to define a product spec that allows for different vendors to implement [the CMAP] in different ways," he said. "It's a very open process. This is not a secret project in any way."

No matter the implementation, the expected goal is similar. Comcast hopes the CMAP architecture will provide a 200 percent increase in QAM capacity in one half of the shelf space, leaving room for 32 more QAMs. Comcast also believes the CMAP approach will produce headend power savings of 50 percent or more.

Salinger said CableLabs and Cable Europe Labs are participating in the project, since CMAP will reference many CableLabs specs. However, the spec itself will remain with Comcast because it describes a product, rather than an interface.

Moreover, the CMAP specs will not replace or modify any of the CableLabs specs, but instead offer an equipment architecture option for implementing interfaces such as Docsis 3.0, PacketCable, and other CableLabs specs.

Comcast certainly needs to get vendors behind CMAP, but the MSO is also working with MSOs and the National Cable Television Cooperative Inc. (NCTC) , a consortium that helps Tier 2 and Tier 3 MSOs get volume pricing on programming and hardware, to incorporate their requirements to achieve unified product specifications for CMAP, Salinger stressed.

— Jeff Baumgartner, Site Editor, Light Reading Cable
Jeff Baumgartner 12/5/2012 | 4:42:00 PM
re: Comcast Proposes Its God Box

Based on some email that's been coming in in response to this story, there seems to be some misunderstanding (or at least some questions) regarding  whether existing CMTS vendors (ie. Arris, Cisco, Moto, and perhaps Casa systems), will be involved in CMAP developments.

Although vendors mentioned during the keynote were in reference to modular implementations, Jorge Salinger reitereated to us that it is safe to assume that CMTS vendors **WILL BE INVOLVED**  in the development of CMAP technology, including but probably not limited to the integrated implementations that Jorge spoke to that combine CMTS and QAM functions.

We've been in touch with all the CMTS players on this question and we'll have a story soon that describes (in as much detail as they're willing to give at this point) how they are involved, or intend to be involved, in the CMAP project.

So, bottom line... expect CMTS vendors to be involved in the creation of CMAP products...we'll just have to see if they shoot for modular or integrated implementations, or perhaps both. Stay tuned.


Sign In