[UFO Chicago] Comcast Cuts Off Bandwidth Hogs

Neil Ormos H ormos at ripco.com
Mon Apr 9 14:41:28 PDT 2007


Christopher D. Heer wrote:
> Neil R. Ormos wrote:

>> Although Comcast is the only broadband provider
>> that has been widely criticized for this kind
>> of behavior, I wouldn't be surprised when other
>> carriers adopt this approach in the future.  I
>> suspect that all large broadband carriers have
>> provisions in their contract that allow them to
>> terminate or limit service if, in the carriers'
>> sole discretion, a customer's usage impacts the
>> network or other subscribers.

> I'm sure, but I think where you'll see this
> happen more frequently is in cable-esque setups
> where neighborhood bandwidth is shared.  After
> all, on your DSL connection it's all you between
> your house and the DSLAM, more or less.

I wouldn't be so sure that this issue is unique to
cable.

Any telecom service offered at residential rates
(whether phone, internet, whatever[1]) is based on
the prediction that aggregate usage by each
individual subscriber will be a tiny fraction of
saturation.  Networks are "engineered" (a telecom
term-of-art) consistent with that prediction,
which is based on historical traffic measurements.
Demand discontinuities (e.g., use of telephone
modems to access internet service providers,
peer-to-peer file sharing, and the like) upset the
apple-cart.

Although DSL may be less sensitive than cable to
saturation of the the access network, the access
network--from your home to the DSLAM or its
equivalent--is only part of the total network.
Obviously, the traffic an individual subscriber
generates must be aggregated with that of other
subscribers as it passes through the carrier's
proprietary network on its way to interchange with
other carriers.  So even if saturating the local
loop has no direct impact on other customers'
local loops, it does have effects upstream.  It
wouldn't take too many fully-saturated subscribers
to use up the available bandwidth in a router or
link.

Moreover, consider that carriers pay for
bandwidth, either directly, or through their
investment in owned facilities.  Subscribers who
use significantly more bandwidth than predicted
upset the assumptions on which residential
all-you-can-eat rates are set.

None of this is intended to excuse Comcast's
outrageous conduct in the situations mentioned in
the article.

>> I want *real* broadband: 100 Mbps fiber-to-
>> the-home, like the services available in Korea and
>> Japan.  Unfortunately, SBC has adopted a much less
>> capable fiber-to-the-pedestal technology for their
>> Lightspeed network, and even Verizon's FiOS FTTH
>> product is limited compared to the Asian offerings.

> Well, it helps that (A) Japan and Korea have
> vastly different population densities than we
> do, and (B) they think Federal subsidies for
> broadband is OK.

I've seen this population density argument before,
but I'm not convinced it's valid.  I don't see
where population density significantly changes the
economics of providing high-speed access,
especially where the carriers have already
resigned themselves to investing in running fiber
into, or near the home, at suburban density.  The
cost of running the actual fiber drop from the
pedestal or pole to the home should be comparable
to the cost of running a conventional cable drop;
I've read the latter is in the range of $48 to
$65.  At some point you need high-speed backhaul.
If you have eight 100-mbps users running off a
pedestal, you need some backhaul. If you have a
100-unit apartment building, you need *more*
backhaul.  There might be a small marginal
difference in the cost of backhaul for large
aggregations, but it's unlikely to be more than
say 40%, especially where the cost of the fiber is
already sunk.  So the next stumbling block is the
cost of CPE, which would become manageable if
purchased in quantity.  DSL and cable "modems"
used to be hundreds of bucks in quantity; now
they're under $50 at retail.

I suspect the real problem is that the carriers
don't want to provide "real" 100 mbps broadband,
because that would let the subscriber buy content
from anyone they want, commoditizing the service
the carriers provide.  The carriers desperately
want to be in the loop where they supply, and
charge for, content, because that's where the
money is.

> AT&T is starting out with fiber-to-the-pedestal
> for quick deployment (Verizon is moving much
> slower),

This article from USA Today seems to indicate
Verizon has made more progress than SBC/AT&T, as
measured by subscribers.

   <http://www.usatoday.com/money/industries/telecom/2007-03-01-verizon-usat_x.htm>

> but I predict they'll go back and roll
> FTTH once the cash is rolling in for U-Verse.

Seems unlikely they would junk their investment,
especially if the backbone network is built-out to
support the lower-performance access loops.

--Neil Ormos


More information about the ufo mailing list