If you want to alter your sense of reality, you neednt resort to hallucinogens.
You just have to enter the world of DSL, where you can embark on one of the more
disorienting trips the industry has to offer. So many flavors to sample. (And so many
acronyms.) Indeed, DSL generates so many tantalizing possibilities that you cant
help but feel confused. Unfortunately, this confusion serves to undermine DSLs very
real promise.
Just when I think Im up-to-date on DSL (Digital Subscriber Line) technology, I
hear about yet another DSL variant, and another round of claims about how this DSL is
faster, cheaper, or simpler than its counterparts. Of course, I should welcome anything
that would speed up my Internet connection, give me a dialtone simultaneously, and allow
me to remotely connect to the office at a comfortable speed. However, Im wary of
DSL. I fear it will go the way of ISDN.
ISDN was supposed to solve our bandwidth problems. A good solution for its time (the
1970s), ISDN could have served us admirably, but we never got around to taking advantage
of what it had to offer. As I pointed out in a previous column, ISDN never took off
because of its erratic deployment. And now it appears ISDN wont get another chance.
At 128 Kbps, the standard ISDN connection just doesnt impress todays
bandwidth-hungry consumers. They dont see the need to suffer ISDNs hassles,
not when they can use analog modems that provide 56 Kbps. With continued improvement in
modem technology, and with multilink technology, consumers should soon realize ISDN speeds
from analog connections.
But lets get back to DSL. Since its so much faster than ISDN, users may be
willing to endure some confusion and complexity of the sort that discouraged users from
working with ISDN.
DSLS ROOTS
DSL originally referred to technology developed by Bellcore. Ten years ago, if you
mentioned DSL modems, youd actually be talking about BRI (Basic Rate ISDN) modems,
which were able to transmit full-duplex data at 160 Kbps. Today, however, DSL is used
interchangeably with xDSL, which refers to the newer DSL technologies, such as ADSL.
While no one talks a lot about traditional DSL any more, xDSL shares DSLs
original purpose: squeezing every bit of bandwidth from the copper line (specifically, the
prevalent Unshielded Twisted-Pair, or UTP, wire). While getting more bandwidth out of
copper is difficult, the alternative is even less attractive. Say we were to assume copper
had, for all practical purposes, a data bandwidth ceiling. The only other way to realize
higher speeds would be to replace copper, and to install in its place fiber or coaxial
cables.
Such a migration would be costly and disruptive. The phone companies, faced with
increasing bandwidth demand, would consider any solution that could eliminate the need for
supplanting the omnipresent UTP. Fortunately, with the advent of DSPs (Digital Signal
Processors) and intelligent compression algorithms, bandwidth barriers that seemed
insurmountable only a few years ago are falling every day. The fact is that transmitting
voice only requires about 4 KHz of bandwidth, a small fraction of the total 2.2 MHz a
copper wire system can support. DSL uses this extra bandwidth to transmit data at speeds
of up to about 9 Mbps.
DSL-INDUCED ECSTASY, THEN REALITY
I first encountered DSL a couple of years ago at a trade show. A few high tech companies
were showcasing ADSL modems, claiming downstream speeds of 1.5 Mbps, as well as
simultaneous support for inbound or outbound calls on the attached telephone set. When I
saw these modems, I was ecstatic.
That was then. Today, I regard DSL more soberly. I realize that I have yet to
experience a real functioning DSL connection in my home state. Moreover, I am dismayed at
the confusion created by the multiple DSL technologies and implementations. All this
confusion may ultimately stifle a great technology. There is no substitute for simplicity.
You may have a wonderful technology, but unless you make it plain and simple to install,
use, and maintain, it wont find a home with end users. Oh yes, you also need to make
it affordable. These basic requirements arent met by DSL. It is anything but simple.
It is rife with multiple standards and complex implementation techniques. Also, prices are
still high. DSL cannot gain market acceptance if the current problems remain unresolved.
Will DSL overcome the challenges of complexity and expense? Id say DSLs
outlook is good on the pricing front. Competition and market demand will eventually lower
prices. However, DSL remains complicated. To assess the prospects for simplifying DSL,
well have to look at a few key issues. These include signal degradation, flavor
proliferation, modulation schemes, and CO (Central Office) equipment options.
SIGNAL DEGRADATION
DSL is sensitive to an intrinsic problem with copper signal degradation. After a
few thousand feet, a signal over copper wire becomes so weak that DSL cannot function
properly. And length isnt the only problem. Other line attributes can introduce
signal disturbances that hamper DSLs operation. These attributes include load coils
(used to amplify signals), bridged taps (used to tap the line to serve other locations),
and line gauge inconsistencies.
To sum up: Most DSL implementations require that no more than 18,000 feet separate the
CP (Customer Premise) and the CO (Central Office) and that the UTP connection be straight
and clear, free of load coils, bridged taps, and line gauge inconsistencies. By some
accounts, 85% of CP locations served by telephone companies are within the 18,000 feet of
the CO. Thats the good news. The bad news, however, is that many of these locations
may still have lines unsuitable for DSL. The proportion of clean lines to dirty lines is
unclear. Many potential users wont find out whether they have a suitable line until
after they install DSL equipment.
FLAVOR PROLIFERATION
By now you should know that DSL isnt just DSL. It has turned into a hodgepodge of
technologies and implementations collectively known as xDSL. (The x part of
the xDSL label is variable, and serves to distinguish each kind of DSL implementation.
Basically, any given DSL technology has a letter of the alphabet in place of the
x. Hence, we have ADSL, HDSL, VDSL, and so on...)
What accounts for all these variants? Ask the creators of the xDSL technologies, and
you will hear that the variants address market requirements. But ask the market, and you
will discover that perplexity is the order of the day. It would appear that users are too
innocent to realize how sophisticated they are. While the need for so many DSL variants
may escape users, at least one market reality has a clear connection to DSL options. The
reality is that most target applications for DSL are asymmetric, meaning that the user
receives (downloads) much more data than he or she sends (uploads). For example, in a
typical Web-surfing session, the surfer sends very little information, but downloads lots
of information from Web pages to his or her browser.
Users who fit this profile will want to know the difference between symmetric and
asymmetric DSL implementations. In the symmetric configuration, the upstream (CP to CO)
and downstream (CO to CP) data speeds are identical. In the asymmetric case, the
downstream speeds are usually much higher than upstream speeds.
Now that were clear on that point, lets muddy the waters a bit and review
the current DSL technologies. But please keep in mind that by the time you read this, a
couple of new ones might have splashed onto the scene.
- ADSL (Asymmetric Digital Subscriber Line) Perhaps the most recognized
type of DSL technology, ADSL is positioned to be the DSL protocol of choice between the CO
and the CP. The implementation of ADSL would include an ADSL modem at the CP with a
corresponding ADSL modem at the CO. The UTP copper line running between the CO and the CP
will include three channels, a high-speed downstream channel, a medium-speed duplex
channel, and a POTS (Plain Old Telephone Service) channel. Each modem will be fronted with
a device called the POTS splitter, which is used to split off the POTS channel used to
connect regular POTS telephone sets. The front splitting of the POTS channel guarantees
continuous phone service should ADSL fail. ADSL operates at downstream speeds of 1.544 to
8.448 Mbps and upstream speeds of 16 to 640 Kbps with a maximum range of 18,000 feet over
UTP.
- RADSL (Rate Adaptive ADSL) A variation of ADSL, RADSL can achieve
downstream speeds of up to 7 Mbps and upstream speeds of 1 Mbps. An advantage of RADSL is
that it can dynamically adapt to the condition of the line, maintaining a continuous
connection.
- HDSL (High data-rate Digital Subscriber Line) HDSL is an alternative way
of transmitting T1 (or E1) over two pairs of UTP without requiring special repeaters. HDSL
is the most mature form of DSL and has been in use by Telcos and ISPs and in PBX
environments for some time. HDSL is a symmetric protocol with an operating speed of 1.544
and maximum range of 12,000 feet over UTP. HDSL-2, the next generation of HDSL, promises
lower power dissipation, simpler system implementation, and lower costs.
- SDSL (Symmetric or Singleline Digital Subscriber Line) A symmetric
technology similar to HDSL, SDSL uses one pair of UTP for transmission, achieving a speed
of 768 Kbps with a maximum range of 12,000 feet. Generally, two SDSL interfaces can be
combined to achieve full T1 (or HDSL) speed.
- VDSL (Very high data rate Digital Subscriber Line) Formerly known as
VADSL and BDSL, VDSL is an asymmetric technology that resembles ADSL but provides much
faster speeds. Obtaining these speeds can be difficult, however. VDSL requires fiber
medium beyond 4,500 feet. VDSL operates at downstream speeds of 13 to 52 Mbps and upstream
speeds of 1.5 to 2.3 Mbps for ranges between 4,500 down to 1,000 feet over UTP.
- IDSL (ISDN Digital Subscriber Line) IDSL uses the same industrystandard
line coding technique as ISDN and is compatible with existing ISDN access equipment. A
symmetric technology, IDSL has an operating speed of 128 Kbps and maximum range of 18,000
feet over UTP.
- CDSL (Consumer Digital Subscriber Line) The newest entry into the DSL
arena, CDSL is similar to ADSL, but it promises to eliminate the need for POTS splitters
and significantly simplify DSL implementation. CDSL is proposed to have a downstream speed
of 1 Mbps, an upstream speed of 128 Kbps, and an operating range of 18,000 feet over UTP.
MODULATION SCHEMES
As if the different implementations of DSL didnt introduce enough complexity, there
are at least two types of line coding or modulation systems that are used to implement
many of the different DSL flavors (such as ADSL and VDSL). These line coding systems
include CAP (Carrierless Amplitude/Phase modulation) and DMT (Discrete MultiTone). CAP
uses the entire frequency spectrum as a single channel and separates upstream and
downstream signals using special algorithms. DMT, on the other hand, divides the spectrum
into multiple sub-channels, sending the data at an optimized rate over these subchannels.
There are yet even more types of line codes, but CAP and DMT are the most recognized. The
different line coding techniques cause incompatibility between the different DSL equipment
on the market even if they handle the same type of DSL implementation.
CO EQUIPMENT OPTIONS
The CO itself has become a hotbed of debate among the equipment vendors. At issue is the
best way to handle the many DSL connections that will be terminating at the CO and
switching the data connection to the backbone network. One proposal is to deploy a device
called the Digital Subscriber Line Access Multiplexer (DSLAM). This device is tasked with
switching the aggregate data from DSL modems to the backbone, eliminating the need for
switches and routers.
Another question is how the backbone should be upgraded to handle the extra traffic
generated by DSL. There are different opinions as to what technology to use to allow for
the higher capacity. The divided camps include those pushing for ATM, frame relay, and
straight-through IP. We have yet to see who will prevail in the backboneswitching battle.
If the past is any indication, it will take a long time for this dust to settle.
CONCLUSION
When it comes to xDSL, one thing is certain: uncertainty. Who can say which (if any) xDSL
flavor will emerge as the favorite? Will CAP or DMT line coding become standard? And what
about divergent CO and CP implementations? With all this uncertainty, I cannot see how we
will realize the DSL dream anytime soon. Sure, there are test-markets in pockets here and
there, but I have yet to see anything substantial to hang my hopes on.
Uncertainty inspires caution and inaction. Consider the plight of a small company
trying to capitalize on a certain xDSL technology. Any such company would have to live
with the risk that another (new and improved) xDSL flavor would enter the market, stealing
attention from the older technology.
This fear also inhibits action on a larger scale. Look at how much the Telcos invested
in ISDN, only to preside over confusion and weak market reception. You can bet that they
will be extra-cautious this time around. We can only hope that through standardization and
corporate cooperation some form of DSL will see the light of day. Otherwise, I suspect
another technology will come along, and that well go through this mess all over
again. Finally, I would like to add that I have reserved the acronym YDSL for my own
future use. You never know.
|