Voice And Data Togeth er: Not So Fast

By Patricia Schnaidt Ten years ago when I started in networking, the rage was the coming integration of voice and data. ISDN was going to spark a revolution in which users could speak and transmit data simultaneously over the telephone. Sonet and ATM were in-the pipeline and promised integrated voice, video and data transmission rates so high one could barely imagine a practical use for them in an end-user enterprise network. At the same time, the first miracle of local networking occurred: 10-Mbps Ethernet over unshielded twisted-pair telephone wires.

Ten years later, while it's Gigabit Ethernet transmissions over unshielded twisted-pair, voice, video and data still aren't commonly integrated over a single network--either for the enterprise or the carrier. The lack of integration isn't because of the vendors' lack of fervor. Today's integration technology solution is packetized voice over the data network--a network that will be imbued with quality of service that guarantees the delivery of voice and other time-sensitive payloads such that voice conversations don't sound like Donald Duck with a cough and SNA sessions don't drop midstream. The transition of voice over the data network, with the exception of international connections, will be slow in coming.

The widespread integration of voice over data networks won't happen because of predominate user need, cost-effective product availability and reliability. Let's assume the vendors can get the voice quality where it needs to be. Bandwidth isn't the issue, for voice takes little bandwidth. Video is a different story, of course, but IP multicast shows promise here.

Running voice traffic over a campus data network requires that the data network have the same reliability as the phone network. PBX networks are measured in 99.999 percent reliability, and a dip to 98 perc ent can be career-ending for a telecom manager. Network managers would be delighted to get 98 percent reliability on servers and data networks, and 99 percent can be achieved only with many precautions and extra hardware. There has to be a great incentive to risk your voice traffic to a less reliable infrastructure.

Data networks also have higher operating costs. Many an IS manager has justified to upper management the higher cost in salaries and equipment required to operate a data network rather than a PBX network. Data networks need to gain the self-healing, self-maintaining capabilities of the voice networks before widespread integration is practical. When's the last time any of your network equipment fixed itself?

Why take the highly tuned, cost-efficient voice network and run it over the expensive data network? You'd integrate the networks if the application warrants, of course. Applications like call centers, distance learning, desktop videoconferencing, and integrated e-mail and voicemail boxes call for an integrated voice and data transport, and will result in the limited deployment of integrated networks. As far as IP telephony in a corporate environment, fax over the Internet may well end up as the most common integrated application, given the high volume of fax traffic in the United States and the high cost of international phone calls.

For the enterprise, the integration will happen on the campus backbone before it does on the desktop, but we still need the technology to support the integrated application. For years, ATM was the knee-jerk answer for both the carrier and the enterprise. On the carrier side, IP over Sonet comes up more frequently as an alternative or an adjunct to ATM. On the campus side, ATM deployment is stalled by the promises of Gigabit Ethernet and a handful of Internet standards, including IP multicast, RSVP and RTP. Yet this mixed-bag of standards is unfinished and unproven. Forget about ATM to the desktop.

Watch carriers closely for signs of voice, video and data inte gration for they have the biggest cost incentive to integrate and the biggest risk if they do. Today, carriers have separate networks for voice and data, and while they may make a wholesale upgrade of equipment in the next 10 years, they are now grooming off the data traffic from the voice network and putting the data traffic on a separate network.

This is Patricia Schnaidt's last Networkologist column and December 15 will be her last issue with Network Computing. After more than three years with this magazine, she will be joining Windows Magazine, also a CMP Media publication, as associate publisher, editorial. We'll miss her.

On The Edge
By Art Wittmann
FreeWire
By Bill Frezza
Corporate View
By Robert Moskowitz
Net Results
By Dave Molta

Updated November 10, 1997