More emerging network technologies: Digital subscriber line
What are the pros and cons?
Digital subscriber line (DSL) is the focus of this month's installment of our series on emerging network technologies. We examine the upsides and downsides of this technology as the industry moves to expand its deployment. (2,100 words)
igital subscriber line technologies (DSL or xDSL) bring new life to old wires. There are more than 60 million tons of copper wiring around the world, put in place by public telephone and telegraphs (PTTs), local exchange carriers (LECs), and competitive access providers (CAPs). Fiber is on the rise, but it will be quite some time before we can expect to see more fiber coming into our homes and offices than copper lines. In the meantime, DSL technologies offer improved electronics and copper-fiber hybridization, which ought to keep copper-based data network solutions running smoothly for another 20 years.
DSL is an evolving family of technologies that work along similar lines but are incompatible with one another. With a continually growing list of names and variations -- at present, at least one DSL line crawls along at 192 kilobits per second (Kbps), while another operates at breakneck speeds of 52 megabits per second (Mbps); Rockwell launched a new variant of DSL (called Consumer DSL) just two months ago -- DSL is increasingly confusing to consumers. To abate this confusion, let's take an in-depth look at the pros and cons of DSL as it weighs in across several markets.
A major benefit of DSL for home users is that it is able, in most places, to utilize copper lines. This allows it to make use of approximately 70 percent of the infrastructure in place in over 80 percent of homes throughout the world. The ratio of use varies, depending on the level of industrialization and the age of the infrastructure in place, but copper lines are still the number-one point of connection between home users and the outside world -- even more so than television.
DSL requires fewer pairs of copper lines than T-1 connections, making it competitive with the T-1 market. As noted, it's use of fewer pairs of lines means that new infrastructure does not always have to be implemented when the lines are put into use. This may prove useful in meeting the ever-increasing phone and data wiring needs of today's professional users, particularly in the downtown sector of cities, where the infrastructure of older buildings must be revamped to meet modern communication needs.
One of the biggest drawbacks of ISDN in this area is the overhaul needed for most telephone systems to accommodate its digital services. Some variants of DSL allow use of the same physical line for both DSL and analog voice communications. The same pairs of wires can be used, saving infrastructure and installation costs while providing a level of backward compatibility with most telephones. Another drawback of ISDN is that digital telephones require a separate power source. DSL draws power for the analog voice from existing pairs of copper line, which can be split with a passive electrical device and remain in operation during a general power failure. In emergency power failure situations, this is extremely useful.
DSL can separate analog data modem connections and voice switches. With the growth of the Internet, the number of voice lines used by modems has sky rocketed. More and more LECs complain that they are running out of voice lines as a result. Typically, voice services last only a fraction of the time used by modem users. The deployment of telephone switches and services is balanced by a fine calculation of the ratio of population to actual use. Typically, there are seven to eight users for every available port on a given voice switch. With modem users, this ratio goes way down, and the very expensive switches quite often get used up. With people using data service systems separate from voice lines, these switch ports can again be made available.
DSL can also provide much higher bandwidth than analog modem connections. This will primarily benefit users. The wide range in speeds will allow different applications of the technology, but even at the most basic level, DSL technologies are at least twice as fast as today's hybrid anadigi 56-Kbps modems. In fact, some variants offer such high bandwidth that new services such as digital television are under consideration as a result.
Strategically, DSL is moving customers towards a digital communications network while keeping one foot in analog territory. From the sales point of view, this will open up new markets and services that were not available before, or were available only at high cost. As DSL becomes more widely deployed, the marginal costs of manufacturing and distributing such digital equipment will go down.
No new technology moves forward without problems, so there are downsides to DSL technologies as well. These downsides are most relevant to service providers (SPs), but any costs incurred will inevitably be passed on to consumers.
DSL comes in many varieties, some of which are incompatible. This means that any SP wishing to deploy DSL has to choose one or two methods and stick with them. Because deployment is often on a large or very large scale (from portions of towns to whole states and countries), SPs can't afford to make such decisions lightly. And once an SP has made its decision, its users are confined to utilizing compatible systems, even if they'd prefer more variety. An additional problem lies in the fact that the industry is still so new that no serious effort has been made to standardize products. Not all DSL products made by different vendors are compatible, even those of the same type.
DSL is expensive to the SP and the consumer. Because SPs have to deploy DSL en masse, they must make sure that they can provide the service as required in all areas of their infrastructure. Preliminary inspection of existing infrastructure can cost tens of millions and the installation of new lines in place of older, degraded-quality lines is easily a billion dollar prospect; and this is in the U.S. alone. For consumers, DSL equipment is expensive compared to modems; although customer premises (CP) DSL devices are falling below $500, central office (CO) devices are relatively expensive and do not yet have high enough density ("modems" per enclosure) to make them effective.
DSL is not a switched service. This means that the copper lines have to go directly between the CP device and the CO device without any repeaters or line amplifications intervening. Because most DSL lines are limited to a maximum of 18,000 feet, SPs have to place CO equipment closer to their users and install additional haul lines to the main switch COs. This often means using more expensive fiber lines for these higher bandwidth congregations of the various individual customer connections.
The non-switched nature of DSL also means that only larger SPs can afford to provide it, and large companies are often not quick to change. In the U.S., SPs are supposed to offer a portion of their infrastructure to competitors, but they don't do it easily. The infrastructure is most often installed and owned by the LEC; the cost of installation is amortized over very long-term (several decades) leases, and the LEC generates a profit from it only after many years. ISPs who see DSL as a great new service opportunity often cannot provide it because provisions from LECs are either slow or lacking. US West, for example, recently completely closed down its LADS (local area distribution services) lines. It had sold unconditioned lines, great for DSL, direct to businesses; these lines were historically used by burglar alarm companies to wire individual houses to their monitoring systems. Several ISPs began to buy LADS lines to operate DSL networks; some complained that US West was purposely delaying installation for ISPs. Then US West completely closed off its services across its 14-state LEC region; although the ISPs petitioned against it, US West prevailed. Most view this as an attempt to forestall anyone else from coming out with a competitive DSL offering in the US West region before it could announce its own service (which it did in late October 1997).
DSL provides consumers with higher bandwidth data services, which in turn means that SPs have to upgrade their data service backbone to support additional needs. Long-haul backbones and interconnections are significantly pricey endeavors.
One last element of DSL affects SPs directly. DSL competes with the existing and very fruitful data services market in T-1s and such. In the competitive U.S. telecommunications marketplace, home user and personal services make up the majority of connections, but are actually financially supported by the revenue from services provided to businesses. To interfere with the gross profit margins made on T-1s to support personal services does not make immediate financial sense for SPs. The business services must likewise grow to maintain that balance or the major SPs will suffer seriously. This explains why most of the major DSL deployments in the U.S. today are still targeted at the business market.
In international markets, where PTTs owned by the state government decide on service policy, the chances for DSL implementation are much higher. Even the government, however, has to financially justify its services. Government policy makers aren't all that different than those of competitive business SPs, but governments usually try to plan for nationwide services which can be implemented at one time. This combined with bureaucracy leads to delays.
Voice communications only uses a certain portion of the analog signaling spectrum. An uncompressed analog voice signal is marked at four kilohertz of the spectrum to run up to distances of 18,000 to 22,000 feet. In digital, this converts to about 64 Kbps. However, using typical Category 3 telephone wire, it is possible to actually use up to one megahertz of the spectrum at 18,000 feet. So, theoretically you might be able to drive the line up to 16 Mbps. In practice, this would almost never happen due, among other things, to copper line quality and electromagnetic interference. Every single electric or electronic device, not to mention the many natural phenomena that come into play, generates this interference. There is, however, another likely prospect. If you reduce distance, the chances of signal degradation go down. Some DSL types take advantage of this; but as the line gets shorter, the CO equipment has to be closer to the consumer. This limits the possible density ratio of the supporting CO system because close proximity also means less customers served by the reach of the system.
The theory behind DSL isn't terribly complicated, and is in many ways similar to traditional T-1 and 56 Kbps digital data service technologies. Then again, that's like saying a car and a motorcycle are based on the same design. The difference lies in actual signal modulation and data framing. Signal modulation is the set of frequencies defining bits and bytes of data as they are transferred.
The different varieties of DSL use the following signal modulation techniques: 2B1Q, carrierless amplitude/phase modulation (CAP), and discrete multitone (DMT) modulation. The 2B1Q system uses the same signaling as ISDN and is used in several types of DSL. One variety of DSL is called IDSL (ISDN DSL). It's a strange little technology that uses ISDN for low-level communications and layers DSL data communications over it.
CAP and DMT are two competing standards used commonly in asymmetric DSL technologies like ADSL (asymmetric DSL), and VDSL (very high bit-rate DSL). For these asymmetrical types there are different rates for bandwidth transfer depending on which direction you take (CP to CO or vice versa). The DSL signal spectrum is partitioned into three segments: the analog voice segment, the upstream segment, and the downstream segment. Keeping these separate basically creates separate channels within the same physical lines.
We will look more thoroughly at the different DSL technologies in the next issue, defining each of the varieties and comparing their bandwidth provisions, their likely applications, and their current deployment in various markets throughout the U.S. and internationally.
About the author
Rawn Shah is chief analyst for Razor Research Group covering WAN and MAN networking technology and network-centric computing. He has expertise in a wide range of technologies including ATM, DSL, PC-to-Unix connectivity, PC network programming, Unix software development, and systems integration. He helped found NC World magazine in December 1996 and has led the charge to the deployment of network-centric computing in the corporate world. Reach Rawn at firstname.lastname@example.org.
If you have technical problems with this magazine, contact email@example.com