Article Info

The Cablecos are Waking Up to Reality

The WSJ today has an exclusive look (free version of article here) at a report to be released by CableLabs that outlines the potential need for cablecos to undertake a massive infrastructure upgrade in order to stay competitive. Unfortunately, the report is not yet available to non CableLabs members, though we would sure like to get our hands on one (HINT HINT HINT) and CableLabs has informed me it never will be.

The report, which has been reviewed by The Wall Street Journal, warns that at present growth rates cable operators’ existing technology may not be able to compete efficiently with Verizon on Internet services. “At some point, optimization of the (cable) network becomes more expensive than simply deploying” fiber directly to homes, the report warns.

CableLabs is an organization funded by equipment providers and cable companies to ensure interoperability of equipment and set standards. CableLabs maintains the DOCSIS standard for cable modems, and just recently released the 3.0 version.

Cable companies have lampooned the deployment of FTTH by the telcos as an expensive endeavor, and highlighted the fact that their own infrastructure can meet any challenge.

But the report, dated July 31, raises the specter that cable operators may have to sharply boost spending on wiring in the future. Companies for several years have been trying to assure investors that big outlays for big network upgrades are over, leaving cash flow for share repurchases and other purposes.

The report is inflaming some cable executives who insist existing networks can meet future broadband demands. “I wholly disagree with the conclusions,” says Mike LaJoie, chief technology officer of Time Warner Inc.’s (TWX) cable division. Its assumptions “are not reflective of what our reality is.” Says Dave Fellows, chief technology officer at Comcast Corp. (CMCSA) : “This report does not reflect our view.”

Whoa. Talk about biting the hand that feeds you.

The History

The cablecos have managed to squeeze admirable performance from their existing coax plant by transitioning to a Hybrid Fiber-Coax architecture (HFC). They have made excellent engineering tradeoffs and deliver an adequate data service.

These are the tradeoffs:

  • Cable modems rely on shared bandwidth. DOCSIS 1.1 forced a set of 100 users to share 38Mb/s of bandwidth, and DOCSIS 3.0 raises this bandwidth to 160Mb/s by bonding multiple 6Mhz slices of spectrum. While the absolute speeds are sufficient, the shared nature of the medium creates a potential problem.
  • There is limited frequency spectrum on the cable due to the need to offer analog channels and it isn’t clear if this spectrum will ever be recovered. This results in an absolute limit of several hundred megabits to be shared among several hundred users. Reclaiming this spectrum will require a switched video architecture and/or the installation of set-top boxes at every TV.
  • The amount and the efficiency of the spectrum allocated for the upstream direction results in a relative scarcity bandwidth when compared with the downstream direction. This bandwidth is also shared among hundreds of users.

The Problem

Applications are breaking the assumptions inherent in these tradeoffs. People are using their connections for longer durations due to file sharing, video and music downloading, and online file storage. This trend is accelerating, and new applications that could stress the shared nature of cable are bound to appear.

To borrow an ancient telephony term, Erlangs are going through the roof. Data from Japanese subscribers that migrate from DSL to FTTH clearly show (see ‘Tragedy of the Commons‘) that higher speeds lead to much higher utilization, creating a geometric explosion in raw bits transported.

This, in turn, creates problems with the latency on cable modem connections. Latency is a big, big problem since it impacts the quality of applications like VoIP and gaming. High utilization leads to traffic jams, which leads to high latency, which leads to a poor user experience.

This is what we suspect is at the heart of the CableLabs report.


We recognize the FTTH technology now employed by Verizon (VZ), NTT and others is truly superior to any alternatives. I have the FiOS service at my home and couldn’t be more pleased. (see ‘My FiOS FTTH Install‘) We also feel that the FTTN architecture being deployed by AT&T (T ) is a half measure (see ‘The Jedi Mind Trick‘). But great technology is meaningless without a killer app, and none exists that radically differentiates FTTH service from cable service to the end user at this time.

We expect such an application to arrive for FTTH (see “The Industrial Accident“) though we know not the time and nature of the form it will take. If you do, please let us know!

The problem is that new internet applications (Napster, YouTube, iTunes, Skype) are unpredictable overnight sensations. Cable got the jump on DSL when Napster hit and suddenly everyone wanted broadband. DSL deployments from the telcos are finally starting to cath up, but deploying the infrastrucutre took years.

Infrastructure is overkill until the day it isn’t. And no one really knows when that day is. The millions of FTTH subscribers in Japan are incubating the next killer app right now.


Comments are disallowed for this post.

  1. I fully agree with your Industrial Accident theory quite happily subscribe to the thesis that I will eventually have Terrabytes of storage, Gigabytes of memory and Gigabits of bandwidth (what for, I dont know, but I am sure something will crop up).

    BUT, and this is a big BUT, I cannot imagine why profit seeking corporations would want to deploy a retail fiber infrastructure across entire countries when they can achieve the same by using *existing* copper.

    The critical flaw in your argument resides in your assumption that copper is limited. It is not. In college, I had a 2.4 kb/s modem. By the time I got to 56 kb/s, I reliably read everywhere that limits had been reached. ISDN too was thought to be a last gasp. We are now at VDSL with >20mb/s. The Industrial Accident theory applies here too, just as it does with Microprocessors or Storage. I would be astonished if copper just stayed still.

    Fiber too will get better/faster/cheaper, and I would wager that the performance ratio between copper and fiber is likely to stay fairly static, and that the same could be said of pricing.

    Basically what I am describing is a situation where the cost, performance and installed base characteristics of two technologies (fiber + copper) will largely decide how the are deployed. As long as copper remains cheaper, not-as-fast-but-fast-enough and generally more highly penetrated than fiber, then it will also likely remain the last mile technology of choice. Fiber will be more prevalent in the backbone.

    Think of it another way: the account you wrote of your FIOS screamed of one thing: COST. I counted two (trained) guys (?), a truck, some fancy gear, all mobilised for at least a couple of hours on the spot + some travel time (1 hour?). So I am thinking Verizon has at least a couple of hundred $ of CPE at your place (that will depreciate) and has spent well north of $1000 to install your gear. Now do the same back of the envelope calculation with DSL, and I bet it will be 1) much cheaper and 2) achiever the same as your FIOS install (e.g.

    All this to say that of all the FTTx variants, FTTH has to be the most ridiculous and least likely to succeed. Mixing faster and more expensive fiber with cheaper and slower copper is likely to remain the most likely combo for some time to come.

    Posted by chris seilern | August 18, 2006, 5:21 AM
  2. You guys need to check out the Utah State’s initiative of subscirber owned fiber optic network. There is an article about it in Salt Lake City Tribune

    Posted by Devang Shah | August 18, 2006, 4:38 PM
  3. The ongoing telecom infrastructure improvement trend is not so much about short term capital gains as it is survival. The days of twisted pair are over for Momma Bell. If you think DSL and voice are going to cut it in the 2000s you are very much wrong! Cable has finally energized as a huge competitor and the hand writing is on the wall, modernize your outside plant or loss your customers.

    I worked for cable during the huge build out of the 1970-1980. The Verizon FTTP build out reminds me of the good old days. The cost on installation will go down as the equipment improves and the techs get better trained.

    The real saving is going to be in plant maintenance, especially when compared to coax plant. The fiber is noise free. Cable system are pledged by noise problems and their system problem calls are months behind. The problems have long cleared by the time the techs look at them, but by then customer has had to endear poor intermittent service.

    This is about survival of the fittest just like in the jungle. The customer demands the very trouble free high bandwidth service for his hard earned dollars.

    Posted by Jim | August 21, 2006, 12:15 PM
  4. I find this discussion interesting.
    Wouldn’t you think that just like telcos figured out how to use twisted pair for internet access (e.g. DSL.. ADSL.. VDSL2+…) cable will figure out a solution to use existing plant in a much more efficient way to deply newer services?

    If so, its really a matter of equipment and network design (in addition to services design) and not neccessarily replacing the HFC plant. Which by-the-way will also be a huge endeavour but not of the same scale that would involve replacing HFC plant…

    Any opinin ojn which equipment companies are better positioned to leverage this situation?

    Posted by Nilam Ruparelia | August 23, 2006, 1:30 PM
  5. Trackbacks / Pingbacks

  6. GigaOM » Blog Archive » The Great Cable Bandwidth Debate | August 17, 2006, 4:51 PM