Comcast Preps Docsis 3.0 Trials

Comcast Corp. (Nasdaq: CMCSA, CMCSK) is looking to begin Docsis 3.0 trials this year, with deployments possibly following in 2008, according to Tony Werner, the MSO's executive vice president and chief technology officer.

Werner discussed those plans and other elements of Comcast's technology strategy Tuesday during the MSO's analyst and investor day.

Werner did not say where Comcast would trial or deploy the new platform, but did note that those decisions would be based on certain business requirements, and pointed out that 3.0 can be targeted to specific portions the market.

He didn't say as much, but it's expected that Comcast and other MSOs, will apply 3.0 first in markets where telcos, especially Verizon Communications Inc. (NYSE: VZ), have deployed fiber-to-the-home networks.

Docsis 3.0 is an emerging CableLabs specification that will use channel bonding to produce broadband speeds greater than 100 Mbit/s. It's also designed to support IPv6, IP multicast, and other advanced IP features.

The faster speed isn't the only benefit of Docsis 3.0. According to Werner, 100-Mbit/s cable modem service tiers will provide similar cable modem termination system (CMTS) economics to today's 6-Mbit/s tiers.

Werner spent the bulk of his time Tuesday discussing cable capacity requirements for future services and the "levers" Comcast can pull to ensure it has plenty of bandwidth to throw at them. Some levers he cited included node splits, digital optimization, and switched digital video (SDV).

Node splits, he said, can be used to "surgically" add capacity to service areas, a move usually driven by downstream applications such as high-speed Internet and video-on-demand (VOD).

This year, about 65 percent of Comcast's node splits are of the cheapest "logical" variety, which cost about $3.35 per home passed and enable a tripling of capacity. About 25 percent are "modular" splits ($8 per home passed), and only 10 percent are "physical" node splits, which, at about $26 per home passed, are the most expensive of the lot.

Comcast is also exploring the bandwidth efficiencies afforded by variable bit rate VOD, a technology being championed by companies like Imagine Communications . (See Imagine Connects With SeaChange .)

Another lever in Comcast's arsenal is SDV, a technology the MSO will first test in parts of Denver and New Jersey. (See Comcast Reveals SDV Test Beds.)

Citing Comcast's own research, Werner said that, out of the 200 least-popular cable channels, a maximum of 40 are being watched at any given time. If those viewing patterns hold up, Comcast "can start stacking additional channels almost to infinity" without increasing capacity, Werner said.

By applying "conservative assumptions," Werner estimated that Comcast could free up 78 MHz of spectrum using a mix of SDV, analog-to-digital migration, reclaiming bandwidth from legacy apps, and advanced compression (MPEG-4).

— Jeff Baumgartner, Site Editor, Cable Digital News

ISEngineer 12/5/2012 | 3:08:53 PM
re: Comcast Preps Docsis 3.0 Trials I wasn't able to attend The Cable Show - which vendor(s) were used for the CMTS and cable modem in this demonstration? Docsis 3.0 submissions don't seem to help narrow this down...
Jeff Baumgartner 12/5/2012 | 3:08:52 PM
re: Comcast Preps Docsis 3.0 Trials The demo used modems and CMTS from Arris
alchemy 12/5/2012 | 3:08:50 PM
re: Comcast Preps Docsis 3.0 Trials The charade here is that what is really being shown is what was formerly called "DOCSIS 2.0B" and is now named "DOCSIS 3.0 Bronze". They're demonstrating limited downstream channel bonding and limited IPv6 where IPv6 is only used by the NOC to manage the cable modem. Everything is using existing DOCSIS 2.0 silicon.

Personally, I wouldn't want a 3.0 cable modem driven by a fully conformant DOCSIS 3.0 CMTS until 2010. It's new silicon, new hardware designs, and a lot of new software. CableLabs and the MSOs seem to live in some alternate universe where the vendors have infinite resources and this all emerges from the lab bug-free. The MSOs then want the vendors to sell all of this at near-cost. I don't understand the huge hurry when big chunks of the network are still running on ancient Cisco UBR/VXR gear that is still DOCSIS 1.1. I'd be surprised if the MSOs are truly in any hurry to junk all of that gear. I predict a huge amount of noise for several years and most networks will end up being, at best, DOCSIS 2.0B for a long time. The vendors will spend millions of dollars on NRE that will mostly sit there unused.
Jeff Baumgartner 12/5/2012 | 3:08:49 PM
re: Comcast Preps Docsis 3.0 Trials The "unofficial" Docsis 2.0b project had 3 downstreams/1 upstream at the cpe. Looks like that implementation will only be offered in STBs. (Broadcom and SA both have chips for it)

True, however, that "bronze" cmts will do downstream only. But there's a March 2009 sunset on bronze and silver cmts testing. all 3.0 modems must test against the minimum specs of 4up/4 down from the beginning.

alchemy 12/5/2012 | 3:08:44 PM
re: Comcast Preps Docsis 3.0 Trials Jeff Baumgartner writes:
True, however, that "bronze" cmts will do downstream only. But there's a March 2009 sunset on bronze and silver cmts testing. all 3.0 modems must test against the minimum specs of 4up/4 down from the beginning.

Right. That's what the MSOs told CableLabs to tell the vendors to urge them on two months ago. In a perfect universe where the Broadcom CMTS silicon rolls out with no showstopper erata and the TI Puma 5 rolls out with no showstopper erata (didn't it just slip 3 months?), those dates might be hit. Most of the vendors are still saying "show me the money" since DOCSIS 3.0 is a big hit on their cost of goods sold & a significant chunk of NRE. I've never seen project schedules but I'll bet this is a $50 million project for each CMTS vendor and none of them particularly want to enter that strange and wonderful world of CableLabs conformace testing twice in two years for Bronze and then Full 3.0 testing. I'll bet most vendors drag it out so all the NRE it doesn't hit their bottom line so much.

Passing CableLabs testing in no way means your product works. The Syndeo soft switch passed the first wave of PacketCable testing way back when. It couldn't even make a basic SS#7 off-net call and has now vanished off the face of the earth after briefly having a pulse for a few months at GenBand. For something as complex as a CMTS, passing CableLabs tests clearly doesn't mean your product works and vendors typically spend considerable time stabilizing their products afterwards without retesting. You can't seriously think the code image running in the field for a 10K, 64K, C4, or Cuda in any way resembles what was tested at CableLabs.

As I said previously, I don't think I'd want a DOCSIS 3.0 cable modem in my house until 2010. The MSOs are also rather busy spending money on the video side of the house so I wouldn't look for huge demand for the gear when it's actually available. Like usual, they'll insist that all the vendors give them upgrades for free for any non-conformant gear sold in 2008 and 2009. In my next life, I want to come back as an operator rather than a vendor.
Sign In