D3 Upstream: What’s the Hurry?
Upstream channel bonding is a year away for most major MSOs.
Not too long ago, the industry was planning to expand upstream capacity using DOCSIS 3.0 channel bonding right about now. Isn’t happening. Consumer demand for faster upstream just isn’t developing. Meanwhile, most major cable operators are still in the midst of deploying faster downstream connections.
But even though faster upstream speeds are on the backburner for now, there are still signs that if upstream demand grows as fast as it could, DOCSIS 3.0 upstream channel bonding might not be the endgame for the upstream.
DOCSIS 3.0 allows multiple downstream and upstream channels to be used together at the same time by a single subscriber through channel bonding. DOCSIS 3.0 can achieve downstream speeds of up to 160 Mbps by bonding 6 MHz – or in the case of Europe and some parts of Asia and Latin America, 8 MHz – channels together, while DOCSIS upstream channel bonding can provide up to 120 Mbps of shared throughput for cable operators.
But for various reasons, upstream channel bonding isn’t as hot this year as it was last year, even with Verizon and Qwest recently announcing that they’ve increased their upstream speeds.
“The interest level has actually diminished from over a year ago,” said John Mattson, Cisco’s senior director of CMTS products. “I think a year ago, there was a lot of interest in it. Almost everybody thought it would be a big deal this year, and this was even before most of them were deploying downstream channel bonding in a big way. I think most of them are kind of wrapped up in deploying 3.0 downstream channel bonding, and upstream channel bonding has become a little less urgent for them.”
Dermot O’Carroll, senior vice president of network engineering and operations for Rogers Cable, said one of the obstacles in getting upstream channel bonding to the starting blocks was the lack of DOCSIS 3.0 CMTS hardware and software, although that has started to change.
“We’re expecting to see [CMTS gear] in the fourth quarter of this year,” O’Carroll said. “Then we’ll start testing it and trying it out in the network in Q1 and Q2 of next year in terms of trials.”
Cisco’s Mattson said his company would have a new software release for its existing, deployed line cards – as well as newer line cards down the road – in September or October, and that customers have already started beta trials.
“At this point, we have equipment in the labs that is capable of doing upstream channel bonding,” said Jeff Finkelstein, Cox Communications’ director of network architecture. “We’ve just received it, so at this point it’s really more of a lab exercise, and we’re not in a position where we could do field trials. I don’t honestly know when we’ll get to field trials, but I would venture sometime next year.”
While there is currently just one CMTS vendor with full CableLabs certification for upstream channel bonding, which is Casa Systems, there are other reasons why upstream channel bonding has yet to take flight.
“It’s not a question of not having the software because we’ve been working with Cisco on getting all of the pieces in the lab,” said Ivan Lamoureux, Shaw Communications’ director of IP network engineering. “It’s really just getting the resources freed up from downstream channel bonding to work on upstream channel bonding. It’s really going to be next calendar year before we have enough resources freed up to get it tested in the lab. Quite frankly, without the business drivers in place to deploy upstream channel bonding, we’re comfortable letting others do the work on the leading edge issues and troubleshooting.”
UPSTREAM CHANNEL BONDING NOT IN VOGUE WITH SUBS
Which brings up another point: Despite increased upstream speeds by the likes of Verizon and Qwest, there isn’t a huge demand from residential subscribers for faster upstream speeds just yet, which also means cable operators wouldn’t be able to find an immediate return on their investments, although there are some efficiencies through statistical multiplexing on bonded upstream channels.
“Since a 3.0 modem theoretically connects to all of the channels simultaneously, it could pick the channel that has the most bandwidth available, and it could send the data over that channel so you get the better balancing with the stat muxing and better performance for time-sensitive applications like gaming and video,” Lamoureux said. “So we do see that benefit, but that’s tough to sell to customers, especially when there’s a premium on 3.0 modems and we basically give the customers the modems for free when we do installations. Unless there’s an opportunity for increased revenue, then we’ll shy away from giving away the more expensive 3.0 modems if we can still use the 2.0 modems.
“I agree there are some subtle efficiencies with 3.0, but it’s tough, both internally to make the business case with the bean counters and externally for the customers if we need to charge them more for the 3.0 modems,” Lamoureux continued. “We need to position it properly so they understand the benefits. It’s a subtle benefit because your upstream isn’t going to be X percent higher; it’s that you’ll see statistically better performance over time.”
BAG OF UPSTREAM TRICKS
Cable operators have other tools in the toolbox before they get to upstream channel bonding, including upgrading to 64 QAM and using SCDMA below the 18 MHz to 20 MHz range.
“SCDMA alleviates the interference from impulse noise,” said Floyd Wagoner, Motorola’s director of global product marketing. “Our view is that cable operators don’t need to go right to 3.0 upstream channel bonding overnight. They can use SCDMA with DOCSIS 2.0 and gain 50 percent more capacity with incremental costs.”
O’Carroll said Rogers currently isn’t using SCDMA today but plans to use it. As for 64 QAM, Rogers has it in 10 percent of its systems, while Lamoureux said Shaw has it deployed in 85 percent of its upstreams.
“There’s two different reasons for getting more throughput on your upstream,” O’Carroll said. “One is giving higher speeds to customers, if they want it, and the other is capacity management, or segmentation. When you go from 16 QAM to 64 QAM, you get significantly more capacity on that upstream channel, which can be used for segmentation.
“We do 64 QAM today for that purpose. It’s really an economic question of if you have to segment in the upstream but not in the downstream, do you put in more fiber, do you turn up another upstream channel – and there’s not that many of them around – or do you increase your modulation? As you get to 64 QAM and you have higher bandwidth on the channel, you get better statistical multiplexing, so you could offer higher speeds.”
Wagoner cited one study that said upstream traffic increased 17 percent on average from 2007 to 2008, while peak rates increased 24 percent. Shaw’s Lamoureux said his company is seeing a 30 percent to 40 percent annual growth rate in upstream traffic.
“I think the consumer demand on the upstream is increasing,” O’Carroll said. “I think the need for a higher speed upstream is certainly evolving, but I think you can provide a lot of that with DOCSIS 2.0. If you get 64 QAM on a 6.4 MHz channel, you can get a better upstream experience. When you start getting into channel bonding and wider upstreams, then I think you’re more into commercial areas today, but I don’t see that need on the consumer side today.”
Figure 1: Example upstream and downstream
channel bonding configuration. Source: Cisco.
DRIVERS FOR UPSTREAM CHANNEL BONDING
Business services is one area where it makes sense to offer upstream channel bonding, since business, in theory, would be willing to pay a premium for the increased speeds.
“That would be a great win because business customers tend to clog the individual DOCSIS 2.0 channels, so if we could get them on 3.0 and balance their traffic and spread it around the channel, that would lessen the impact on everyone who is on the system,” Lamoureux said.
Robert Burke, Netgear’s product marketing manager, said cable operators both here and abroad have interest in his company’s DOCSIS 3.0 modems and EMTAs. DOCSIS 3.0 gives cable operators the flexibility to bond as many of the upstream channels that they need (see Figure 1), which he said gives them a strategic advantage in serving new customers or going into new areas.
“For existing customers, upstream channel bonding is going to get them very high-quality video teleconferencing,” he said. “I think video conferencing for both residential and business customers is going to be a huge driver for upstream channel bonding. In a recession, people turn airplane tickets into video conferences wherever possible. I would expect [cable operators] to get additional levels of pricing for higher levels of upstream bonding.”
With upstream channel bonding, Burke said cable operators could snag business contracts away from telcos by serving small- and medium-size businesses with DOCSIS 3.0 cable modems, instead of the more expensive services they currently use.
Lamoureux said upstream bandwidth in Canada is being taxed by peer-to-peer file sharing since Canada doesn’t have the same copyright laws in effect as the United States does.
“A lot of people turn on their upstream channel sharing and just let it run 24/7 without worrying about the consequences,” Lamoureux said. “We worry about the bandwidth. This might be one of the instances where Canada is ahead of the U.S. Maybe we’re a few years ahead in terms of upstream bandwidth usage, but we’re going to hit a wall in three or four years when it comes to upstream bandwidth. Maybe even in two or three years if something like video conferencing really takes off.”
While Shaw hasn’t deployed DOCSIS 3.0 upstream channel bonding, Lamoureux is already worried that it won’t be enough. He said 64 QAM, channel bonding and SCDMA aren’t silver bullets when it comes to provisioning the upstream over the long run, and that the industry needs to be looking ahead at other technologies.
“The short-term answer is going to fiber, but if we can leverage our existing coax plant a little longer, that’s a huge win,” he said. “The only other options are expensive node splits and upgrading all of our plant, which is also expensive. We need to start figuring out what to do next. I think it hasn’t hit the mainstream yet because the Comcasts and Time Warners aren’t seeing this wall with the upstream. We’ve had some discussions with the folks at CableLabs, but maybe we just need to bite the bullet and go with fiber or upgrade our MIB split to 88 megs on the return path.
“We see the end, and we don’t feel comfortable that we have a good plan to address it. We’re comfortable on the downstream that we have a transition plan, but not on the upstream.”