Cable modems offer a path to genuine two-way communications, but they're not ready yet.
By Glenn K. Schulke
This month I'd like to take a somewhat different approach than usual. Among the new communications technologies emerging for home use are some that, at first glance, don't seem to connect directly to the system administrator's job. Yet as telecommuting increases, even ancillary technologies may take up some of the bandwidth inside your head as well as on your network. So it may be worth thinking about one of them, which is interesting for its own sake also. After all, sysadmins have homes, too--don't they?
Take cable modems, for example. They are targeted primarily at home users, but if cable modems actually catch on, some of those users will be cybersurfing to your office computers. Although these products are the responsibility of the cable operator that actually is providing the cable modem and service, because they are using the cable modem to cybersurf to work, the users may assume that any problem they encounter is with the connection at the office.
Of course, it really isn't your job to support cable modems for telecommuters, but they will be calling you anyway if they hit a snag. You will need to be able to determine whether the problem is at your end or theirs. As a result, you should be familiar with cable modems and able at least to troubleshoot the connection at the telecommuter's home, if only to be sure that you don't have trouble at your end.
It's possible that this market could go in either of two opposing directions. Cable modems could open a high-speed on-ramp to the Internet from the home. Or the hundreds of thousands of cable modems already on order by the cable companies could end up in the garbage because the communications tree of cable TV becomes root-bound. There is plenty of excess bandwidth on the cable networks, but they were designed to carry broadcast signals "downstream" into subscribers' homes, not the other way around. Like a salmon swimming upstream against the current of a river, the data being sent out of a cable modem-equipped computer has to swim against the wall of data running downstream, and that is only part of it.
Cable operators are doing their part to make this a reality by rapidly upgrading their distribution infrastructure. To further increase the available bandwidth of their systems and reduce or eliminate the upstream bandwidth problem, they are creating a hybrid distribution system by combining new fiber-optic cabling with existing coaxial cable. As these hybrid fiber-coax (HFC) networks are beginning to take shape, literally hundreds of cable modem tests are being conducted worldwide, from cities in the U.S. and Canada to the Netherlands, Denmark and Chile. Promising though this may sound, in the U.S., for example, fewer than 25 percent of all cable systems have the HFC networks required for high-bandwidth, two-way communications. This number is expected to rise to 50 percent in 1997 and to about 80 percent by somewhere around 2000, according to Merrill Lynch.
Cable systems equipped with cable modems for Internet access have advantages. They are fast, easy to make and cheap (costing $300 to $500). Because a large number of companies are or will be producing them, they should get even cheaper. The companies producing cable modems demonstrate the convergence of two industries, with companies from the computer side such as DEC, Hewlett-Packard, IBM, Intel and Motorola in the mix. Not to be outdone, though, the big cable-box manufacturer, General Instrument of Chicago, has its own cable modem, called the Surfboard. Also in the running to grab a share of this market is Terayon, a company partially owned by the networking giant, Cisco Systems of San Jose, CA.
Still, there are challenges facing the cable operators, both technical and nontechnical. The intrinsically interactive nature of the World Wide Web, when combined with bandwidth hogs such as programming on demand, realtime full-motion video and data conferencing, and expansion of telecommuting, are a few of them. Cable operators--notorious for unreliable service, frequent service interruptions and poor-quality signals--may find it difficult to win the trust of the consumer when it comes to communicating information more critical than prime-time pabulum. When America Online became America Offline last Aug. 7, not only did it inconvenience their six million or so subscribers (including me), but the outage had a cascade effect on anyone else attempting to send e-mail or files or conduct business with companies doing business on the Internet through AOL. Moreover, the high turnover (about 50 percent per year) of AOL subscribers indicates the fickle nature of the home-based cybersurfer.
Chief among the technical issues is the layout of cable systems, which resembles a tree with trunks, branches and limbs, with the leaves being each subscriber's home. As mentioned before, this is good for broadcasting a signal in only one direction. Everyone gets the same basic signal broadcast over the cable networks' trunks and branches; it is up to the intelligence built into your receiver or cable box to decipher and break down the signal into the discrete channels.
What's more, the cable operators will need additional equipment in their systems for switching, routing and noise filtering. For cable TV, noise is merely a nuisance, leaving you with poor-quality video or audio signals, but lost or corrupted data poses a larger problem. Compounding the problem is the tree-and-branch layout of the cable system, because "upstream" data not only has to fight for bandwidth but also fight the noise on the entire cable system. Fortunately, the cable modems themselves are better than they used to be at identifying and filtering noise, and have new mechanisms to verify the integrity of the data sent upstream.
Another problem with the tree-and-branch layout for data communications of the cable systems will surface as the cable modem user base grows. The bandwidth of the main trunk lines will get spread thinner and thinner along its branches. Cable companies hope that by the time this occurs, they will have enough subscribers generating enough revenue to justify adding additional trunk lines to the cable network. The cable companies use a series of hubs to distribute the trunk signal to the subscribers in a given area. These hubs are shared by around 300 or so subscribers, and can deliver data at rates between 10 and 20 megabits per second (mbps). The problem is that this data-rate bandwidth must be shared by all users on the hub.
Some of the other hurdles facing the cable companies--while less technical--also pose real challenges. These include learning how to manage interactive networks, being able to bill customers based on usage and (as mentioned previously) service reliability. To this end, a Palo Alto, CA-based consortium called @Home was formed by the three largest cable companies, including TCI, to set up a national backbone network to connect the regional operators to the Internet, as well as provide network management and billing functions. Some other cable companies are forging alliances with the regional telecommunications companies to help solve these problems.
The cable modem trials currently being conducted are yielding mixed results. In one such test, being conducted by Adelphia of Coudersport, PA, using LANcity's cable modem, are providing 10mbps data rates for Internet access for both downstream and upstream data. However, the test only involves 30 homes. Time Warner's tests in Elmira, NY, using Zenith's products are yielding substantially less performance. In their tests, data rates of 500 Kbps are being achieved in both directions. Time Warner's tests are being run in around 200 homes.
While these data rates are certainly better than a 14.4 Kbps or 28.8 Kbps modem, the sizes of the test groups are very small; if deployed en masse, the results would likely be substantially lower data rates. Also, these tests are being conducted primarily for Internet access applications. When you overlay the bandwidth required for the emerging technologies discussed earlier such as videoconferencing, interactive TV and programming on demand, it is doubtful that cable modems will be the large-bore needle for our daily cyberfix. In the end, cable modems will be only one of many high-speed methods of Internet access.
Glenn K. Schulke is president of Open Technologies, Inc., a systems integrator specializing in software integration services, located in Tempe, AZ. He can be reached at firstname.lastname@example.org.