SPEED CHANGES EVERYTHING. (Column) Tom Yager.

Full Text: COPYRIGHT 1999 Miller Freeman, Inc.

Remember when you got your first T1? How you stayed up all those nights downloading stuff, hitting graphics-laden Web sites and listening to streaming audio? The Internet seemed like a different place through that big, fat pipe. Many organizations still rely on T1, fractional T1, and Frame Relay for their in-house Internet link. One and a half Mbps was plenty fast; supporting a bunch of simultaneous connections, assuming everybody was climbing on with modems.

That's no longer a valid assumption. The day is coming when those browsing your site are likely to have a faster connection to the Internet than your company does. In several areas, broadband has arrived, bringing with it access speeds up to 10Mbps. Within the next year, the number of broadband users will skyrocket, and the upper speed limit will reach 55Mbps. Pretty soon, you'll need to adjust the formulas you use for Internet services capacity planning. You may also want to rethink your telecommuting policies.

This month's column on broad-band isn't aimed at telecommunications wizards (in whose class I do not fit), but rather at administrators, managers, planners, and developers who put services on the Internet and manage staff-connectivity policies. I'll limit this discussion to what technology I've experienced myself or know to be on its way. There are many whispers of things to come. We'll do well if we can put a plan together that covers the next six months.

OLD TECH MEETS NEW

Broadband technologies take advantage of transmission media which already exist (telephone copper wires, cable TV coaxial, and the airwaves) and adapt them to the carriage of high-speed data signals. Trench warfare has broken out between cable and telephone carriers, each claiming superiority over the other. If wireless seems to be missing from the battle-field, that's because it hasn't quite come into its own.

[Graphic omitted] If it hasn't done so already, your local phone company will soon adapt your neighborhood's telephone circuitry to carry Digital Subscriber Line (DSL) signals. DSL has been around for several years, but only in the past two years or so has it made the transition from limited regional trials to widespread deployment. I use the term "widespread" loosely, as I'll explain.

There are presently three popular flavors of DSL: ADSL, IDSL, and SDSL. The A in ADSL stands for asymmetric. This class of service favors Web surfers by splitting transfer rates in favor of super-fast downloads. Upload speeds are much slower, in part to make more room for faster downloads, and in part to prevent DSL from cutting into more-profitable business-oriented offerings.

A DSL modem carves your phone line's signal capacity into frequency bands. Voice traffic occupies a tiny range of the available frequencies, leaving plenty of room for data signals. Data is injected into your phone line at frequencies well above the audible range. The DSL data carrier is always there, even when you're making a phone call. You just can't hear it. The modem peels off those inaudible frequencies, decodes them, and encodes outbound routed traffic.

In my area, Southwestern Bell offers two DSL packages to its subscribers. For $39 per month, a user can get 1.5Mbps downloads and 128Kbps uploads. For $129, the download speed is goosed to a maximum of 6Mbps, while the upload speed rises to only 384Kbps. At both levels, the phone company sets a minimum and maximum download rate. The 1.5Mbps service has a minimum rate of 384Kbps. Speed freaks will appreciate the 6Mbps package's guaranteed 1.5Mbps minimum download rate.

Users of Frame Relay service will be familiar with the approach taken by DSL providers. If you're the only user transferring data, you'll get the maximum bandwidth. As more users climb on, all users' bandwidth will be ratcheted down toward their minimum level. During periods of maximum load, you'll receive data at the minimum rate, with occasional bursts as holes open up.

ADSL service is a bad deal for people running servers, because of the upload caps. It's a great deal for home surfers, and those who make out best are telecommuters. Work-at-home types are more likely to have the bandwidth to themselves as the majority of fellow subscribers toil in the workaday world, away from their home PC. A 6Mbps ADSL line is so close in performance to a LAN connection that it makes telecommuting sensible for entire classes of users it couldn't appeal to before. The upload cap does hurt somewhat, but pulling data from company servers over ADSL is a pleasant experience compared to dial-up or ISDN.

Entrepreneurs who want to run low-volume Internet businesses are better served by symmetric DSL. SDSL matches upload and download speeds. The resulting connect rates aren't as impressive as ADSL (around here, SDSL tops out at 1.5Mbps) but there is little to distinguish a 1.5Mbps SDSL Internet link from a more expensive Frame Relay connection. Much depends on the ISP you choose and on the quantity of traffic on your shared line.

My attempt to get wired with DSL met with an odd demise--the telephone wiring in my area is too new to support high-speed data. For all the talk of DSL's advanced technology, it turns out to be awfully sensitive. There are several things that can keep DSL from making it to your building. Fiber-optic cabling is one. If any portion of the path between your local phone company's central office (the CO, also called the "wire center" because that is where all the phone cables in your area converge) is fiber, DSL fails. It requires old-fashioned copper wire from end to end, but you can't have too much of it between you and your CO. Southwestern Bell's DSL supports a maximum wire distance of 17,500 feet between you and your CO. SDSL stretches the limits a bit, but for all DSL services, the further you are from the CO, the lower your top speed will be. Your phone-service provider will determine the variety of service classes and range of speeds available.

Some of the tricks used to lower wiring costs in high-density dwellings can keep DSL from getting through. Your apartment's or office's telephone-wire-distribution system could stop DSL in its tracks, or render it unreliable. The phone company may have you connected to a multiplexer which splits one phone line into several--an arrangement that nixes DSL. The customer most likely to qualify for DSL is the owner of a single-family home in an established neighborhood within a few minutes of a major city.

If you tried to get DSL through your phone company and were turned down because of cable distance or lack of availability, don't give up. U.S. law requires that phone companies share their infrastructure with Competitive Local Exchange Carriers (CLECs). If your Baby Bell (or GTE, or whomever) turns you down, you may find a CLEC operating in your area who will hook you up. The CLEC buys the right to push data over your phone line and co-locates its equipment at the primary carrier's CO. CLEC access costs more, but if it's your only option, you may not mind paying.

A great many people can't get DSL, either for the reasons I described or because their CO isn't wired for it. It helps to request it, and to get your neighbors to do the same. Demand tends to drive deployment schedules. For those in deployed areas who can't get DSL because of wiring or distance issues, there may be other alternatives. One is IDSL, an ISDN-class 144Kbps service that can handle distance and mixed fiber/copper media. IDSL has the advantage of being cheaper for ISPs to serve than ISDN since it doesn't use dial-up equipment. Another option will be VDSL, a very high-speed link (up to 55Mbps download and 2.3Mbps upload, according to specifications) that relies on the existence of fiber transmission lines near the customers' home or office. Both technologies are in trials as of print time.

DSL is perfect for home and small-business use. It lives on your existing phone lines and does not interfere with regular phone service. Like all broadband methods, it is always on. A dedicated IP address is typical. ADSL home-class service usually provides one IP; the more expensive SDSL will sell you blocks of addresses based on need. DSL gives you a direct digital link between you and your ISP. You have a choice of ISPs when you subscribe to DSL, but those ISPs must contract with the DSL provider to be linked to their network. You might find your favorite ISP is not on your carrier's DSL partner list.

Early reports of DSL reliability and performance have been extremely favorable. Keep in mind that as of today, only a small percentage of telephone customers will ever qualify for ADSL or SDSL. As technology advances to get more customers connected, users may find congestion eating into their bandwidth.

THE CABLE GUY

Thwarted in my efforts to get DSL, I joined the "revolution" (the company's marketing word) by signing up for the @Home cable modem Internet service. This symmetrical broadband service promises download speeds of up to 3Mbps into your home through your cable-TV outlets. Like DSL, cable modems require near-ideal conditions on the wire between you and the cable company's equipment. In my area, we had to wait until the cable provider upgraded its distribution system to a mixture of fiber and standard coaxial cable (the cable wire that runs into your house).

@Home and its regional alternatives are strictly residential services. There are reports that @Home is about to reach into business data services over cable, but no schedule has been mentioned. Cable modems have stirred controversy from the beginning, not least because of cable companies' reputations: doesn't it seem that everyone complains about lousy picture quality, repairmen that never show, and service that winks out every year during the Super Bowl? What is it that makes cable operators think we'll trust them with our data?

Not surprisingly, the quality of cable data services vary to the same degree as that of cable television. Those served by well-funded, technologically up-to-date cable operators should make out well. If you're served by an operator that can't keep a decent picture on your TV, you should do some hard thinking before relying on it for data.

Television coaxial is capable of carrying lots of information, far more than copper telephone cable. To get 100 or so channels to your home, the operator uses equipment that assigns video signals to frequency bands. Each band becomes a channel that your cable box or TV/VCR tuner can capture. Cable data is also delivered over channels that the operator assigns. The modem, which you connect directly to your cable TV tap, has a tuner inside it that scans the channels looking for a data carrier. There is one channel for downstream data and another for upstream. Once the modern identifies these, it links to data equipment at the cable company, which activates the modem as an IP router. Once your cable modem is connected, it stays permanently connected unless something goes wrong.

There are tens of thousands of @Home users in the United States. Considering that, there are surprisingly few complainers showing up in Usenet newsgroups. They are vocal enough to make it look like cable is laying an egg, but they are a small group nonetheless. @Home has created controversy by imposing an upload cap on what used to be a symmetrical service. Across the United States, @Home is setting its equipment to throttle uploads to 128Kbps. This is being done to solve a big problem--uncontrolled growth. At $30-$40 per month, the speed of cable proved irresistible to a great many users. Trouble is, many of those users decided to set up servers. Web servers and FTP servers created problems of their own, but most troublesome were game servers. They require no special hardware, software, or knowledge, only a game CD and a connection to the Internet. Any teenager with a copy of Quake II and a cable modem can eat an entire neighborhood's bandwidth with a worldwide frag fest.

Nobody likes to be the victim of a bait-and-switch, and @Home should be more forthcoming about the existence of the cap. A leaked @Home memo instructs sales staff to downplay the cap, or to pitch it as a feature. It was an understandable business decision, so why not just be honest about it? The company's announcement that it intended to analyze customer Internet usage wasn't smart either. Like Santa Claus, @Home said it knows if you're being naughty or nice. It knows which newsgroups and Web sites you're looking at, whether you spend most of your time downloading software or JPEG graphics, and whether you're running a server on your network. Okay, every ISP has this ability, but they know not to make a press release out of it. Internet users are among the most paranoid, privacy-minded people in the country.

As with DSL, cable data services can be a powerfully good thing if you can live with the upload cap, be served by a reliable and responsive operator, and are not too concerned that your every move can be logged and analyzed. I decided the speed and value were irresistible, so I signed up. As yet, I am not cruising at full speed--my operator upgraded its technology and hasn't quite caught up with it--but the future looks promising.

Cable shouldn't be your only data service. It is subject to outages, congestion, and growing pains--more so than DSL. But cable should remain cheaper than DSL, and download speeds are usually comparable. Some operators cannot offer two-way service because they can't swing the required hybrid fiber/coaxial upgrades. In those areas, speeds drop to as low as 300Kbps and the upload side of the link is dial-up. Still, 300Kbps is a lot faster than a modem and it could be the fastest data service far-flung people can get.

Cable does not yet provide users a choice of ISPs. Cable companies own their copper and fiber, and cable is not yet subject to the infrastructure sharing requirements imposed on telephone carriers. This has been challenged in court, and a handful of municipalities have won the right to demand that cable operators make their media available to competitors. It's hard to say how this will affect cable data services. Residential service is a low-margin business. If operators have to shoulder the entire cost of a neighborhood upgrade only to see its competition share that media, they might elect not to make data services available. I believe this accounts for the slow deployment of new telephone data services. As soon as a telephone company upgrades its facilities, by law it must open them to competitors.

If @Home does introduce a professional version of its service, expect to see upload caps rise, server restrictions lifted, and fewer big-brother policies, in exchange for a much higher price. Current cable modems are capable of carrying 3-10Mbps of data. Perhaps the best reason to go cable is to be there as your operator raises the service to new levels with new technology. Cable will never replace Frame Relay, but may soon work in a small business setting for non-critical connections. If the cable business model chanes, companies could use cable to move employee browsing off the main Internet connection, saving the bandwidth for paying customers and business traffic. At $30-$40 per month, cable is also the cheapest company-subsidized high-speed telecommuting solution available.

IT FLIES THROUGH THE AIR

Wireless offers a glimpse into the most likely future of high-speed data. One local provider sells 10Mbps links for about $85 per month, and the price includes several channels of wireless digital television.

A wireless provider transmits a broadband data signal via an omni-directional microwave transmitter. A commercial or residential subscriber must be within the transmitter's effective radius and line of sight. A rooftop antenna receives the signal and sends it to a decoder. Only the download signal is sent through the air. The upload link is dial-up analog or ISDN. This type of wireless does not offer subscribers a choice of ISPs. One ISP must cooperate with the wireless provider to split each user's traffic between dial-up and broadcast routes.

Wireless has suffered terribly at the hands of cable, DSL, and direct broadcast satellite (DBS) TV. Wireless companies started out offering "wireless cable TV," but DBS trumped them with more channels and no line-of-sight or broadcast-radius limitations. Wireless companies rolled out Internet offerings, but business models counted on the income from the lower-overhead TV business. As a result, my local wireless provider has only managed to sign up about 200 subscribers since its inception in 1997. It has ceased to offer commercial services and has killed pricing plans that permit subscribers to opt out of the TV channels to save money.

Time will tell whether these early high-speed wireless players will be able to stay in the game. More-advanced solutions are rolling out in several areas and taking advantage of the 5GHz unlicensed frequency band set aside by the FCC for wireless data transmission. This technology is being used in limited trials and to enhance the data infrastructure of widely dispersed facilities like universities and corporate campuses. Expect to see a lot of two-way wireless deployments in 2000, at which time they will become a viable competitor to cable and DSL.

We're finally seeing some significant advancements in inexpensive high-speed data services. Once the retail market heats up, you'll see wireless, cable, and DSL technologies packaged in special offerings for business users. The time may come when you can ditch your $1,500 T1 in favor of an aggregated cluster of business-class DSL lines, vastly increasing your speed and saving money to boot. In the meantime, it might be worth advising your telecommuting employees to check into these options. Set up a few trials, collect users' feedback, and then decide how reliable these services in your area are. Broadband could play an important role in your shop's Internet future.

Tom Yager (Integration, p. 19) works at HealthWeb Systems Ltd. and is a freelance writer with a private research lab. E-mail him at tynger@maxx.net.