CASE STUDY: HIGH-SPEED ACCESS

Author:
Updated:
Original:

Installation and Performance of Road Runner Service

The following is one of three planned reports on execution
issues central to the successful launch of high-speed data services. It was prepared by
Janco Partners analysts Ted Henderson, Tom Friedberg and Roger Metz. Janco, an Englewood,
Colo.-based investment bank, closely follows cable and telecommunications companies.

For several months now, we have been watching the early
stages of deployment of high-speed data services through both the cable and telephone
pipelines. While we are excited about the end result -- the mass penetration of high-speed
Internet services into the home--- we remain skeptical of both Wall Street's expectations
and the operator's capabilities for the rapid deployment of these services. To that end,
we have initiated a three-part grass-roots study of these services by examining first hand
the installation and performance of Road Runner high-speed cable internet service at a
beta site in Kansas City, Mo. Later this year, we will perform a similar study of @Home
and xDSL installations. We acknowledge that the scope of this study is very limited and
our methods are only crudely scientific (at best), but we hope that an up-close look will
bring to light some of the real issues (both positive and negative) surrounding the
deployment and subsequent performance of high-speed data services.

The study involved three phases. First of all, we put a
stopwatch to a variety of Internet downloads using an existing 56.6 Kbps modem connection
at various times of the day and evening. We tried to include some of the most common uses
for the Internet from the consumer perspective -- pulling up news stories, finding stock
quotes, downloading audio and video clips, utilizing search engines, checking weather maps
and shopping. Secondly, we monitored the entire Road Runner installation process, asked a
lot of questions, and noted the successes and pitfalls of the process. Lastly, we went to
the same sites we visited with the 56.6 Kbps modem connection and compared it to Road
Runner download speed on a click-by-click basis.

BACKGROUND

Our test site was in Kansas City, where Time Warner Cable
is in the beta-testing phase of its rollout and has installed Road Runner in an estimated
600 to 700 homes over the past month. Road Runner sends out teams of two installers --
an experienced cable installer and an IT professional. Road Runner is in the process of
cross-training its installation team, which currently is up to 16 people, plus an
additional six subcontracted from a computer chain. We were told that in Houston, where a
more aggressive rollout is underway, the Road Runner installation team is up to 70 people.
Those with cable backgrounds are being sent to a two-week computer certification course,
while those with computer backgrounds are going through Time Warner's cable-installation
training process. The installers are working together during the testing phase, after
which they will begin to do installations on their own. The teams are each scheduled to
perform four installations in a 10-hour workday, or two-and-a-half hours each, including
travel time. The installers on this job noted that the typical installation takes between
one and one-and-a-half hours, with a two-and-a-half-hour job being the exception.

INSTALLATION PROCESS

The installers were running ahead of schedule, and arrived
about an hour and a half early. The biggest challenge they faced was to physically run a
cable to the computer's location in an upstairs bedroom in the northwest corner of the
house; the only existing cable jacks were located near the southeast corner of the home.
After about an hour and a half and several trips up a ladder and into the attic, we had a
live cable jack in the wall near the computer. While this much cable work is atypical,
according to the installers, it should certainly be taken into account that a large
percentage of homes will not have a cable jack adjacent to the computer and will require a
half-hour to an hour of cable work prior to the technical side of the installation. The
attendant wiring problems found in older homes (our test house is well in excess of 50
years old) will certainly present additional challenges during both cable and xDSL service
installations.

Once the new cable outlet was in place, the physical
connection of the cable modem was a surprisingly simple process, and only took about 10
minutes. There are two possible ways of connecting the service; which one is used depends
on the technical features of the PC being used. In this case, the PC was equipped with
Windows 98 and a USB (universal serial bus) Ethernet port, so the connection ran from the
cable jack to the cable modem; from the cable modem to a Peracom USB Ethernet adapter; and
from the adapter to the USB port on the back of the CPU. With a slightly older machine, a
3Com network-interface card would be installed and the connection would bypass the USB
adapter, and run straight from the cable modem into the network interface card. Obviously,
the driver software for any new pieces of hardware must also be installed, but this only
takes a couple of minutes. The modem used was manufactured by Toshiba, and was (according
to the installers) one of the first DOCSIS-compliant modems to be installed.

By the time they had done a basic walkthrough of the
service, set up e-mail accounts, and answered several questions, about 45 minutes had
expired. The installer said that this segment of the installation can last from about 20
to 45 minutes, depending on the customer's need for help and number of questions. In our
case, we were told that we asked more questions than any other installation they had done,
hence, the 45 minutes. The installers were both very knowledgeable regarding the setup and
operation of Road Runner. We were also provided with a user manual and a customized
version of Internet for Dummies geared specifically toward Road Runner service.

PERFORMANCE RESULTS

Overall, we were very impressed with the speed of the
service. As we went through the same list of sites and links that we had the previous
evening with the 56.6 Kpbs modem dial-up connection, typical speeds for opening a text
homepage with a few pictures and graphics ranged from about 3 seconds to about 9 seconds.
This is compared to a range of about 15 seconds to one minute, 10 seconds with the 56.6
Kbps modem. Simple, text-based links such as stock quotes were able to fully download with
Road Runner in two to four seconds, compared to an average of about 15 to 20 seconds with
the 56.6 Kbps modem. Downloading a 65-page research report from the Janco Web site took
one minute, 16 seconds with the old connection; Road Runner reduced this to only 19
seconds.

The most noticeable difference between the two services was
in the download of bandwidth-hogging multimedia files, such as video and music clips. For
example, we downloaded several 600-650-kilobite video clips from espn.com, which took only
12 to 13 seconds to download, compared to nearly three minutes with the 56.6 connection.
Another site that we pulled up was the Billboard.com music homepage that contained about
two dozen pictures, as well as links to music clips. This took over four minutes to open
using the 56.6 Kbps connection and only about 20 seconds with Road Runner. We also
accessed some short music clips at CDNow.com, which took between nine and 24 seconds to
download with the 56.6 Kbps connection, and took merely 3-4 seconds with Road Runner.

Another aspect of cable-modem service, which also applies
to T-1 connections and xDSL, is that they are "always on," eliminating the need
to dial into an Internet-service provider to gain access to the Internet. The incumbent
dialup ISP that we tested took an average of about 1:20 to access, including the dial-up
connection and download of the homepage. During busier times (i.e. evening primetime), it
can take several attempts to dial in to an ISP with a traditional modem, as thousands of
users are battling for a finite number of connections to the server. With Road Runner, you
simply click on your browser icon and the homepage is fully downloaded in just a few
seconds. Once it is open, it can remain on indefinitely

An interesting point made by the installers is that Road
Runner has effectively choked off the system to allow data transfer speeds up to 1.5 Mbps.
The absolute capacity of the cable (as dictated by the Ethernet connection) is 10 Mbps,
but Road Runner has learned in other systems that when early subscribers get used to the
full-bore 10 Mbps maximum speeds, they are disappointed when more users begin to clog the
system (thereby negatively impacting speeds) in subsequent months. By easing back the
throttle, Road Runner feels that it should be able to maintain the 1.5 Mbps speeds for
users in the future when subscriber penetration increases and the system becomes more
heavily used. However, we feel that it is still probable that once penetration levels
reach Wall Street's expectations (conservative estimates peg it at 30 percent of computer
households passed by cable), there may be substantial degradation of speeds (below 1.5
Mbps) due to congestion of the cable pipe, which is effectively a "party line,"
over which many will have to contend for access. If this is the case, the dedicated
bandwidth guaranteed by xDSL services could be highly advantageous with respect to
last-mile, high-speed access since they provide dedicated, uncontested access between the
end-user and telco central office. In fact, if cable modems are reliably able to deliver
1.5 Mbps, the need for the dedicated line from an xDSL service would be limited to areas
where speeds in excess of 1.5 Mbps are required, or where homes or businesses are not
passed by broadband cable or a CLEC backbone. The reality is, nobody will know for certain
until these services are mass-deployed over time.

Yet another restraint on the system, which the installers
were very clear to point out, is that when a user is accessing content from outside the
Road Runner system, the download speed is at the mercy of the server at the other end of
the line. All of the sites we visited were in fact outside of the Road Runner content
universe (see discussion of content below), and we were pleased with the download speed.

One other item to note -- dial-up networking to a LAN
cannot be accomplished using either type of broadband connection (cable or xDSL). If, for
example, a subscriber wants to dial into an office to work from home, he or she would have
to utilize a standard modem connection either tying up the household's primary voice line
or over a second line at an additional monthly cost.

CONTENT

With increased competition from xDSL, which is currently
getting down to competitive prices and similar speeds, Road Runner is banking on its
highly localized content as the bait it needs to draw customers who might otherwise
subscribe to xDSL. Right now, the content at the Kansas City Road Runner home page is
crude at best, comprised primarily of links to various national news, sports,
entertainment and other links. In more developed systems such as Hawaii (which went live
with Road Runner about a year ago), Road Runner has produced customized local content
which includes things such as weather, school information, news, event coverage and bus
schedules. Currently the main hook for the service is the speed advantage and the
always-on nature of the system. As content (including local) will be widely available to
all providers, we expect performance, pricing, and customer service will ultimately be
much more critical components of success or failure than content for all high-speed
providers.

PRICING

Since the Kansas City market is still in the test phase,
Road Runner is currently providing installation at no charge. Once the service is
mass-marketed, standard installation will cost the consumer about $100. An additional $30
charge will be added to that for the network card if the computer is not equipped with
Windows 98 and USB (see discussion above). Monthly usage runs $39.95 per month, including
the lease of the cable modem. If a consumer chooses to buy his or her own cable modem
(typically $200 and up), the monthly service fee will be reduced to $29.95. The pricing is
designed to be competitive with traditional dialup services, which average about $20 per
month, plus the cost of a second phone line (about $15 to $20). The current pricing model
is more attractive than xDSL pricing, but look for pricing parity among broadband
providers to arrive quickly.

CONCLUSION

Overall, we were impressed with the speed of the service
relative to a 56.6 Kbps connection, as well as the ease of installation once the cable
line was in place. However, the difficulty encountered with the cable work brings real
concern to the table. The concept of retail modems being purchased off the shelf and
installed by the consumer at home remains a way off for many homes in our opinion. We have
expressed concerns about the execution and deployment challenges facing cable companies
for some time now. At a rate of four installations per day per installer, we are looking
at both significant staffing increases and a very long time period before any substantial
penetration rates are achieved. Furthermore, with competitively priced xDSL choices, we
must acknowledge that there are real alternatives to cable Internet access out there, and
price competition may affect these services going forward.

We will follow this report with similar reports on @Home
and xDSL installations later this year. In addition, we will continue to monitor the speed
of the service over the next several months, as more subscribers are added to the system.
We will also keep an eye on the overall reliability of the service to see if technical
problems occur, both on the Road Runner server side and from the end-user's perspective.

Related