Greed For Speed
Cable modems and DSL promise high bandwidth - but can IT
really capitalize on them?
For years, IT managers have lived with a big gap in the types of
data-networking services available to smaller remote offices and
tele-workers. There have always been either unjustifiable costs or
inadequate bandwidth at every turn.
But that's about to change.
Two relatively new technologies, digital subscriber line (DSL) and
cable modems, are now offering network managers last-mile links
from sub-T1 speeds and above for prices as low as $40 per month.
Hooking up these services is generally as simple as connecting a
conventional modem. And, at least during their initial rollouts, early
adopters have had few complaints about service reliability.
But as with any new technologies that apparently offer something for
next-to-nothing, buyers need to keep in mind their fair share of
caveats before jumping on the latest bandwidth wagon.
Coming Soon To A Network Service Provider Near You ...
Hopefully
One of the main considerations is availability. Though cable TV
service is available throughout most of the United States, less than
20% of the installed plant is set up for the bidirectional
communications necessary for data networking. Market researchers
estimate that there are about a quarter-million active cable modem
users today, but most of those are concentrated in a few pilot
territories. Retooling cable facilities is not an inexpensive
proposition, and it remains to be seen just how much of an
investment cable operators will be willing to make to pursue a
business in which they have no real experience. The AT&T-TCI deal
dramatized how important the deep pockets and technical expertise
of the telcos are when it comes to the rollout of next-generation
local-loop services. So network managers may have legitimate
long-term concerns about the viability of cable operators as datacom
business partners.
Greed For Speed
That's not to say that cable Internet access isn't a good deal. Avid
Web mavens, high-tech telecommuters, and a variety of small
businesses are using cable modems to get from 200 Kbps to 2
Mbps of bandwidth for a few dollars a day - and are gaining
substantial advantages as a result. Hall Color Projects, a West Long
Branch, N.J.-based digital imaging service bureau, is getting
500-Kbps connectivity from Philadelphia-based cable services
provider Comcast Corp. "File transfers that used to take hours now
get done in minutes," says company president Joe Hall. Not only
does that speed allow Hall to handle more transmission volume, it
also enables him to extend his company's geographic reach since
he can now become more aggressive in offering services
electronically.
Because cable companies already have relationships with a large
percentage of households in their local regions, cable is also
well-positioned for the single-user telecommuter market. Employees
and/or contractors who need to hook up to corporate resources don't
need a new line or NSP account - although extending a coaxial
cable to a home desktop is a bit more of a hassle than running a
regular copper phone line.
There are also plenty of vendor entrants in the field, including
LANcity Corp. (which is being acquired by networking powerhouse
Bay Networks/Nortel) and Motorola, whose CyberSURFR is the
industry's most popular model.
These factors, combined with the demand for increased bandwidth
for consumer Internet services, should fuel substantial increase in
cable modem deployment over the next few years. In fact, the
market analysts at Dataquest project that 1.77 million cable modem
users will be sold annually in the United States by the year 2000 -
up from only 25,000 in 1996.
Greed For Speed
Is Cable Capable?
While the cost/bandwidth ratio cable access offers seems appealing
at first sight, there are some problems with deploying it as a
corporate networking solution. First of all, cable bandwidth is shared
over all connections, from the head end to the desktop. So while
current customers are happy enough with performance today, that
performance is likely to deteriorate as more users come online.
Shared connectivity is also a security issue. Cable providers insist
they are monitoring their lines to ensure that no one is doing any
eavesdropping or generating any kind of disruptive traffic, but
experience has shown that hackers are usually a step ahead of
service providers. Encryption schemes can provide adequate
protection against theft of information, but the real exposure lies with
infiltration of cable-connected desktops by other users on the same
segment.
Perhaps the biggest obstacle to the deployment of cable modems
for remote connectivity to the enterprise is the difficulty using the
technology in shared LAN environments. Contention issues in this
type of configuration are highly problematic, making cable modems
impractical for the very type of small remote-office environments
where the need for inexpensive multi-megabyte bandwidth is the
greatest.
Cable companies are also at a disadvantage in marketing to
corporate customers because equipment is generally being sold
directly by the cable providers themselves, instead of through the
VAR/integrator channel that now accounts for a large percentage of
traditional networking sales. Without this channel support, cable
technology may get cut out of many network planning discussions
and proposals. The lack of a reseller channel also means that cable
providers will have to shoulder installation and support tasks
themselves - a daunting prospect for an industry that has had
problems taking care of its existing consumer customer base.
Greed For Speed
DSL Fuel
A stronger alternative for corporate networkers over the long term
may be DSL. DSL comes in a wide variety of flavors, each of which
offers a different amount of bandwidth. Many DSL variants are
"asymmetrical" - that is, they provide more bandwidth downstream
than they do upstream. This is obviously appropriate for Web
browsing, since users only send keystrokes and mouse clicks
upstream, but want to receive highly graphical Web pages and/or
full-fledged file transfers back downstream. Asymmetry may be less
appropriate for certain business applications - for example, where a
telecommuting developer needs to transmit big chunks of code back
up to a corporate IT department.
Speeds of DSL service vary greatly depending on parameters such
as transmission frequency and distance from the closest telco
Central Office. On the low end are consumer-oriented offerings that
run at around 384 Kbps in one or both directions for under $100 per
month. At the high end are multi-megabyte services that can cost
upwards of $300 for unlimited monthly usage.
DSL is even more limited in its current geographic availability than
cable modems right now, although most incumbent Local Exchange
Carriers (LECs) do have trials underway. In some ways, these
limited rollouts and big variations in cost from region to region are
reminiscent of the problems that plagued the introduction of ISDN.
But make no mistake: DSL is not ISDN.
One difference is that DSL is being introduced to a market that has
clear bandwidth hunger. When ISDN was introduced, it was unclear
exactly who was going to use it and why. In today's
communications-intensive marketplace, even school children have an
understanding of what a fast Internet link means.
Greed For Speed
Provisioning DSL infrastructure is also very different from ISDN or
cable modems. Rather than requiring large-scale retooling
investments up front, DSL modems can be installed incrementally at
central offices based on current demand. That's a much more
attractive business model for telcos than the megabucks crapshoot
associated with other new technology deployments.
And, unlike cable, DSL integrates well with both the SOHO office
environment on the user end and the ATM backbone environment on
the carriers' uplink side.
DSL is also catching on with resellers, who are starting to see it as
useful in helping them capitalize on the next-generation remote
access requirements of their corporate clients. "Corporate resellers
would much rather partner with a DSL vendor than a cable
company," notes Greg Cline, principal analyst for Internet
architecture research at the Newton, Mass.-based In-Stat Group.
"Cable companies don't have relationships with the network products
distribution channel and don't seem likely to start now." Cline's
research also indicates that while less than 20% of NSPs have plans
to support cable modem users, almost 50% say they will deploy
DSL within the next two years.
The biggest question mark with DSL is the copper wiring plant that
sits between potential customers and companies. Because of the
frequencies at which it operates, DSL is very sensitive to the
distances and quality of the wires on which it runs. Research is
being done to overcome these problems, and trade-offs of speed
enable the technology to work one way or the other in most major
metropolitan areas. But the copper in place today is full of splices
and connections that, while adequate for conventional voice services,
just don't cut it for the unanticipated physical requirements of DSL.
Greed For Speed
Pilot Lights
In addition to understanding the peculiarities of cable modems and
DSL, it's important to keep in mind just how they will be used in a
corporate context. Most enterprise networks are abandoning the
concept of direct-dial into central modem banks for remote access.
Instead, they're using the Internet to provide users and small offices
with local dial-in points and to aggregate remote access traffic into a
single pipe at headquarters. They're also starting to use
Internet-based virtual private networks (VPNs) to create wide-area
links at a cost far below traditional frame relay services.
That means that network managers who want to build a coherent
remote network access strategy for individuals and small offices will
have to work closely with their NSPs to determine exactly who can
be given what type of access, and how soon that access will be
available. The last thing most IT departments want to do is to have a
fragmented approach where different sets of users employ different
access services from different providers. The support and
management issues associated with that kind of arrangement are
logistically and economically unacceptable to most organizations.
So a good approach at this stage of the game is probably to:
1.begin determining exactly where high-speed Internet/VPN
access would do the company the most good;
2.enter into discussions with NSPs about their roll-out plans
(taking their stated timetables with a large grain of salt); and
3.pilot next-generation technology in a test-bed setting
Greed For Speed
It's also a good idea to start considering what additional bandwidth at
the edge of the network will do to the enterprise backbone.
The digital business environment is now poised at a similar stage as
the hard-goods economy was when the interstate highway system
was put in place after WWII. Both consumers and businesses are
about to gain unprecedented access to speed and capacity. Those
who don't actively prepare for this change are likely to find
themselves at a significant competitive disadvantage. |
