Cable Modems [Telecom]
Neal McLain
[In a thread on Model 32 and 33 Teletype machines,] hanc...@bbs.cpcn.com wrote:
> In my opinion, the real driver of the communication revolution
> was the huge decline in the price of central data servers
> (computers) and communication lines. Cheap servers made it
> possible for people to afford to offer useful information
> on-line, and, to do so in a very user-friendly format. Cheap
> communications made it possible to provide full scale
> interconnections between servers and the users, and again, to
> do it in a user-friendly format.
As a former cable guy, Lisa's comment "... huge decline in the
price of ... communication lines" prompts me to write about
something that's been on my mind for a decade: the cable TV
industry's early efforts to offer internet service.
The initial effort was led by @Home Network, a company founded
and funded by Kleiner Perkins Caufield & Byers, the same Silicon
Valley VC that had originally funded Netscape. @Home's initial
efforts were successful, and the major cable TV retailers all
signed up.
Numerous pundits had a field day: That brain-dead cable TV
industry is actually planning to offer internet service? The
pundits fell into two camps:
- "They'll never make it work." This camp claimed that the
industry would never be successful, pointing out, among other
things, that the 5-40 MHz return band was too vulnerable to RF
interference. Stewart Alsop, in a famous 1997 Fortune editorial,
noted that "Cable Modems are a Fantasy" (written, as it happens,
about the same time that Fortune's sister company Time Warner
Cable was gearing up to offer internet service).
http://tinyurl.com/d2h5ep
- "They'll never be able to handle the load." This camp claimed
that the industry's network could not handle the anticipated
number of subscribers. This camp also cited the 5-40 MHz return
band, claiming that the industry would never be able to solve the
interference problem. Others claimed that the cost of connecting
to the internet would kill the idea; one editorial even claimed
that every cable company would need an OC12 just to handle the
load.
It other words, one camp claimed the industry wouldn't get any
business and the other claimed that it would get too much
business. One of my former associates, referring to the latter
camp, noted: "well, I sure hope we have that problem!"
A few years later, @Home went bankrupt, and cable TV companies
introduced their own versions of internet service. @Home was a
noble effort, but IMO it failed for two reasons:
- It tried to become a "portal" like AOL and CompuServe. At one
point, it even bought an electronic greeting card company.
- It was too successful in building a workable product. After a
few years of using @Home, cable companies began to think, "this
stuff isn't that difficult, so why are we paying @Home to do
something we can do ourselves?"
Now, a decade later, equipment specs have been standardized and
numerous manufacturers make "cable modem termination system"
(CMTS) equipment for headend installation. Just about every
cable TV retailer in the country now offers some sort of internet
service. As long as the RF network is properly maintained
(correct signal levels and stringent control of ingress/egress),
the CMTS runs with little attention.
Modem manufacturers have proliferated too, and the modems
themselves have gotten easier to use. Many cable TV retailers now
offer install-it-yourself modem kits. Except in cases of signal
failure, most service problems can be resolved by rebooting the
modem.
Of course, the industry still faces problems today. In my
experience, the biggest problems are slow response in some
geographic locations, and lack of service in rural areas.
In most cases, the slow-response problem is caused by congestion
upstream of the cable TV headend. Maybe there's not enough
capacity in the connection between the headend and the internet
(usually a T1 or a T3), or maybe it's farther upstream.
Lack of internet service in rural areas parallels a similar
problem the industry faced in the 1980s: lack of video services
in rural areas. The rise of DBS (DirecTV and Dish) has largely
resolved the video problem, but the lack of internet service
continues to be a public-relations headache. Perhaps the
recently-passed stimulus package will provide the REA with funds
and authority to assist cable TV retailers extend their internet
services.
As for Alsop's famous editorial, perhaps he should have checked
his sources. He notes, "The show operators said that the local
phone companies could not provide enough bandwidth for all the
cable-modem demos. Uh-huh. Right." Well, that was the problem.
The exhibits on the convention floor were connected to a LAN that
was supposed to be connected to the internet, probably by a T1 or
a T3. Whatever it was, it wasn't there: the local phone company
didn't get it installed in time.
Neal McLain
Neal McLain
> In most cases, the slow-response problem is caused by
> congestion upstream of the cable TV headend. Maybe
> there's not enough capacity in the connection between the
> headend and the internet (usually a T1 or a T3), or maybe
> it's farther upstream.
Before Robert B. jumps on me, I guess I better elaborate on
that statement.
Slow response can indeed be attributed to congestion within
the cable network itself. Furthermore, it can be caused by
ingress/egress; e.g., "signal leakage." If there's a break
in the physical shield (wind-induced crack; hungry
squirrel; drunk driver; errant backhoe; vandalism;
whatever), ambient RF from the outside world can enter the
cable and interfere with cable signals. Such interference
can severely degrade digital signals.
Pinpointing the source of a slow-response problem lies in
an analysis of the physical locations of the complaints. If
all complaints come from one node, then that node, or one
of the cables or fibers connected to it, is clearly the
problem. If a node is simply overloaded (too many
subscribers hooked up), the node can be split into two or
more smaller nodes (equipment sales guys call this
"scalability").
But if complaints originate across the entire cable
network, then the problem can usually be attributed to the
headend equipment, or whatever's upstream. Unless it's an
extremely small cable system, with only one node, this is a
useful tool for isolating slow-response problems.
Computer problems can cause response problems too. A
computer owner who switches from a dialup connection to a
cable modem (or DSL) expects faster response. But a
sluggish computer connected to dialup is still a sluggish
computer when connected to a high-speed connection.
Neal McLain