Communications

Viewpoint: Broadband ISDN is happening--except it's spelled IP

W. David Sincoskie

A funny thing happened on the way to BISDN, the broadband integrated-services digital network. For many years the Holy Grail of the telecommunications industry, it was to integrate voice, data, and video services into a grand unified high-speed network. Many technologies were invented with it in view, including asynchronous transfer mode (ATM), fiber to the curb or home, and synchronous optical network (Sonet).

But last year as the BISDN started to deploy, it became the Internet Protocol (IP). Looked at another way, the Internet is the "killer app" for BISDN.

Why did this happen? For one thing, the Internet--specifically, Bob Kahn and Vint Cerf in their classic 1974 IEEE Transactions on Communications paper--got the services architecture right. Where network services are simply applications running host computers, a tremendous industry in third-party applications springs up. When a best-selling "killer app" like the World Wide Web browser is invented, the development and deployment speed is breathtaking. When the killer app changes, as it will, the Internet can react far faster than any other network. E-mail, file transfer, and remote terminal access are the past killer apps, the WWW is the present, and ahead lie real-time voice and video.

At the same time, the Internet must run on top of something. Being a network of networks means that underlying subnetworks must proliferate as the Internet grows. And proliferate they have. Any network that will support the Internet well is selling like hotcakes. The growth of local-area networks (LANs) has been going on for years, but with the interest in public Internet services widespread, second phone lines, ISDN, frame relay, and ATM are selling, in many cases, beyond the most optimistic forecasts. The huge modem traffic on the phone network has phone companies scrambling to reengineer their offices. InternetMCI, Atlanta, Ga., the largest provider of backbone Internet services, installed a nationwide ATM core network to interconnect its routers. In fact, it's fair to say that most long-haul U.S. Internet traffic now runs over ATM.

How did we arrive at this state of affairs? Think about some older network architectures. Plain old telephone service (dubbed POTS) is really one service with one network custom-designed to deliver it. The 3-kHz analog end-to-end service has endured through several generations of switching and transmission technologies. To be sure, today's public switched telephone network barely resembles the manually switched, analog transmission system of the first half of the century. Yet the service is still the same.

By the 1970s, through, people wanted data services, as well. So the concept of the ISDN evolved. ISDN does two things, one a big plus, the other a nearly fatal minus. The plus is that ISDN digitizes the loop. However, the challenge ISDN faced for so many years--finding a killer app--stemmed from its architecture. It tried to integrate several services into a single line. Yet its channel structure made assumptions about the services it would deliver: two 64-kb/s voice channels and a 16-kb/s data channel.

ISDN struggled for years because POTS voice and modem services, as well as LAN services in the centrex market, competed very successfully. Also, a nationwide ISDN had to be constructed at great expense, since, after all, the 64-kb/s channel service is end-to-end.

Only recently, with the growing demand for Internet services, combined with terminal adapters that can put data over several 64-kb/s channels, ISDN sales have been taking off. In retrospect, ISDN took us from the one-service, one-network POTS world to a multiple-service, one-network world. But it nearly failed because at first it got the services wrong.

By the mid-1980s, it was clear that even the 144-kb/s ISDN lines would not sate the bandwidth desire of computer and video customers. Sonet was born only as an extension to the DS1 to DS3 transmission hierarchy, which multiplexes voice channels at speeds of 1.5 to 45 Mb/s, taking transmission to speeds in the gigabit-per-second range and beyond. ATM was invented to bring bandwidth flexibility to the switching infrastructure, so that many services of differing and variable bit-rates might be carried in the same network. Thus, the BISDN, built on top of ATM and Sonet, would allow the integration of any imaginable service into a single network.

Then why is the BISDN being spelled IP in the mid-'90s? It's the networks, stupid. What the original vision of BISDN missed was that the proliferation of LAN technologies in the '80s would in effect make it impossible ever to integrate the entire world into a single network. The Internet, with its network-of-networks architecture, can do this. ATM and Sonet, far from being useless, are rapidly becoming the preferred switching and transmission technologies for the wide-area broadband parts of the Internet. Since ATM was designed to carry any service well, it is perfectly happy carrying IP--even when that IP, in turn, is carrying data, voice, or video.

What will happen next? Something I'm calling IP Dialtone service. It is really quite simple to understand. IP is becoming the end-to-end service of choice for telecommunications consumers, and any service a telephone company deploys must run on top of IP. Plus, any network it deploys must support IP. When a telco has done this completely, it is offering ubiquitous IP Dialtone service.

The impact of this on telcos is breathtaking. It's evolve-or-die time. With the telecommunications monopolies being opened to competition, now a global trend, new players in that business will be looking for cheap, fast ways to offer telephony service to new customers. One of the most promising ways to do this is over the Internet, which already carries fax and voice messaging traffic. Real-time voice, experimented with over the MBONE (the Internet's mulitcast backbone) for several years now, is starting to be commercialized in the form of Internet telephony software products. Why pay $1 per minute or more to call overseas when you can talk via the Internet for just $9.95 per month? Or, even more compelling, why send a fax at $1 a minute when you can send e-mail instead? Of course, the current Internet pricing structures may change, and there are service quality issues. But Internet telephony is sure a good way to attract business.

What does IP Dialtone do for companies in the networks and services businesses? Reduce risk and enable new services by decoupling services from networks. A telco can deploy a new network, like frame relay or ADSL (asymmetric digital subscriber line), without worrying what service is going to drive demand. Any service that takes off on the Internet will run over the new network, driving demand.

Concurrently, IP Dialtone service also lets new applications developers deploy new services, like WWW browsers or Internet telephony, without waiting for a new national network to deploy.

What's ahead? It's hard to predict the next killer app, since things are evolving so quickly. However, I think often about the integration of data, voice, and video into a single, seamless application. Imagine watching a TV show through a WWW browser. A commercial comes on showing something you'd like to buy, so you point your remote-control air mouse at the picture and click. The WWW page for the advertised product comes on, and after supplying your credit card information, you click again and buy the product. The phone rings, and you click the telephone icon on the screen and have a chat with your mother. You click one more time, on the "back" button, and your show resumes where it left off. Oh, and by the way, you're doing this in a car, hopefully while someone else is driving.

Sound interesting? Give me a phone call at 192.4.18.74, and we'll chat.

W. David Sincoskie (F) is vice president of the Internet Architecture Research Laboratory of Bellcore in Morristown, N.J., responsible for leading Bellcore's research in Internetworking and broadband network architecture. A member of the Internet Society's Internet Architecture Board from 1993­1995, he now serves on IEEE Spectrum's editorial advisory board.

(c) Copyright 1997, The Institute of Electrical and Electronics Engineers, Inc.