Voice as a special case of data

By: Scott Bradner

The current Internet is only a distraction in the world of voice. In spite of all of the publicity that the Internet has been receiving it is a very minor player in the business of telecommunications. For example, the domestic voice business amounts to about 200 billion dollars per year while the annual Internet service provider (ISP) business totals only 3 or 4 billion dollars. Perhaps this is why there has been so little apparent inclination on the part of the US regional telephone companies to get involved in the Internet business.

While the revenue streams are vastly different, the traffic levels are not anywhere near as far apart. In fact, a number of observers believe that there is already more data traffic than voice traffic in parts of the US, for example in Silicon Valley area near San Francisco. Richard Notebaert the chairman and CEO of Ameritech, a large US regional telephone company, has just written that the crossover will happen this year for his company.

Today most telephone companies are running data over their voice networks so that currently, data is a special case of voice. This will not last, within a few years the reverse will be true and the telecommunications networks will be primarily data networks, and voice will become a special case of data. There will be a convergence of the voice, video and data networks into a single data-based network. The timing characteristics required for interactive telephony will ensure that voice remains a special case but the percentage of the data network that will have to deal with these special requirements could be quite small. John Sidgmore, the President and CEO of UUNET, has predicted that, at some point, voice will become a niche market.

But voice is not a niche market yet. The vast majority of revenue for all telecommunications functions comes from voice and will for many years into the future. This is a case where the minority of the traffic will produce the majority of the revenue. The very large scale of the voice revenue is attractive and we are starting to see the first tentative steps along the path towards convergence with interactive voice as the initial driver.

Part of the lure is likely to be transitory. A significant part of a current phone bill is the result of taxes and fees imposed by regulatory authorities. The taxes and fees, along with the higher costs normally associated with services offered by monopoly providers, can be avoided by running the same service over the Internet. But it is likely that regulators will impose the same fees on Internet-based services if they see the governmental revenue start to decrease as the applications migrate. There is a lot of money to be made until this happens. A number of people in the Internet business think that delivering voice over IP may be more efficient than delivering the same voice traffic over a traditional switched-circuit infrastructure, so there may be additional savings in the long run. It is possible that some of the existing regulations and fees will be dropped rather than extending them to the future Internet-based service model since in many cases it is technically impossible to differentiate voice-based applications from non-voice ones. But this is far from a sure bet, governments do not often reduce taxes and fees.

But for whatever reason there is a lot of activity in the voice over IP arena. Hardly a week passes without some company announcing some new voice over IP product. The latest batch of products were announced at the Networld + Interop show in Las Vegas in May. In this case some of the companies making announcements are big players in traditional telephony networks including Lucent Technologies, Ericsson and Siemens. One of the commentators pointed out that these and other long term telephony companies could not afford to ignore what is going on in this area.

There are two quite different types of products being developed. The first type assumes that the phone calls will be placed and received on traditional telephones and carried for the beginning and end of the communications path on traditional telephone networks. An IP network is used to carry the voice traffic between the two telephone networks. The other type of product bypasses one or both end telephone networks by carrying the voice traffic over IP to the user's PC. This type can use IP for end to end communications or to just carry the voice to a gateway to the traditional telephone network.

There is a great deal of standards work that has been done in the second area. The IETF's Real Time Protocol (RTP) is used to transport the voice or video data over the network and the ITU's H.323 standard is used as the most common signaling protocol. There are many implementations of these standards including the NetMeeting application included in Microsoft Windows.

There is a significant problem standing in the way of realizing the benefits of this move to convergence -- the underlying IP data network is not yet ready to offer the higher quality of service that interactive voice and video applications require. But this problem is being worked on in the IETF Differentiated Services (diffserv) working group where standards track documents should be ready later this summer.

One thing that seems to be particularly hard for the traditional telephone people to understand about the Internet is that, in general, Internet-based services are peer to peer without much intelligence in the network itself. This confusion is clear when listening to a traditional telephone company talk about the PBX of the future. They think that the PBX will be a large server somewhere in the network that all calls will go through. Someone who understands Internet technology will realize that all local PC to PC calls will be direct and bypass any central server.

There is much potential in this area but it will be quite a shock to the traditional telecommunications industry to move from a world where the huge revenues from the voice business will be under pressure from the small revenues from the data business. Not that it will be fun to watch those companies try and figure out how to deal with the challenges.

Scott Bradner