Vint Cerf hears VoIP calling

For someone widely regarded as the father of the Internet, Vint Cerf can be surprisingly coy about predicting the future.

Too much has happened in the last several years and too many crystal balls have proved wanting. But when it comes to forecasting the adoption of voice over IP, the veteran computer scientist has no such reluctance to qualify this as one of the next big things to affect the technology firmament.

Cerf, nowadays the senior vice president of Internet Architecture and Technology for MCI, says that traditional telecommunications carriers are finally taking VoIP seriously. Indeed, Cerf, who created the TCP/IP protocol that defines online communication now spends a good part of his time focused on VoIP, the cheaper form of telephony, expecting it to permanently alter the telephone industry.

Already, MCI's network is being transformed into one that uses the Internet Protocol to route calls, not traditional phone switches. As a result, it can already do much more than just make phone calls. And that's just with present day technology

But as MCI moves to VoIP, it faces the immediate specter of regulatory issues that's created a schism between the nation's largest phone companies trying to halt VoIP's spread into consumer telephone plans and the "Net heads" that support this cheaper way of making a phone call. Currently, the nation's largest phone companies are trying to coax the Federal Communications Commission into regulating VoIP.

Cerf spoke with CNET News.com about VoIP technology and why he expects this upcoming transformation to pick up speed throughout the rest of 2004.

Q: Did you ever think IP and phones would merge, as they have with VoIP?
A: In 1973, I didn't have any idea it would. The Internet was just an idea. By 1988, I was convinced this was going to be a big deal and I wanted to find a way to make a commercial service. By 1989 it was. Then Netscape created a huge impact on the utility of the network. I'd be lying to you if I said I predicted all that.

Is VoIP the next big jump in utility?
It's really a small step in the evolution of the Internet. The more important things are likely to be grid computing. But VoIP starts the natural progression of another modality that the Internet can support. It also changes the whole of the telephony world substantially, so (VoIP) is hard to ignore.

If consumers begin adopting VoIP, who's going to use a phone company, whether it's a Bell or a long-distance company?

I haven't been a very happy camper about the regulatory positions taken on this.

It depends a lot on what traditional services you have to offer. These things become very commodity in nature. Long distance has gone that way, but it's less so for local service because there's only a modest amount of competition. The way you eventually have to make money is by adding value to the traffic, which means adding new services. That will be how you survive in this game, by adding value. If you want to be a purely commodity business and can survive, more power to you. My reaction is I want to do something better.

Isn't VoIP's success--for example, Vonage has 50,000 customers--forcing traditional phone companies to adjust?
They won't disappear off the face of the earth. The businesses will change because it's a more rich application environment. We don't see ourselves as telephone companies anymore. MCI sees itself as a networking computer and communications company. We don't buy or sell computers; we support computer communication. That's an important part of our business.

What happens when the Bells complete their fiber-to-the-home expansion? Doesn't that cut out the local-phone provider, which will no longer own that local connection?
If they make good on these oft-repeated claims, it's better for everybody.

The Bells' fiber to the home initiative dredges up some pretty complex regulatory issues, correct?
I haven't been a very happy camper about the regulatory positions taken on this. These new networks all ought to be openly accessible to any ISP for a reasonable price. If that were the policy, then every customer, business or consumer would have a choice of ISPs over those broadband facilities. Under the current situation, there's almost no choice. So that's a far more restrictive environment than we had with dial-up. Broadband shouldn't be any different, but it is based on my current understanding of the triennial review by the Federal Communications Commission. I've been arguing we should really open up all these broadband facilities.

But won't that eat the Bells' profits up? How are they supposed to survive?
Gosh, excuse me, but my view of this is we're willing to pay for access to the service--not asking for the service for free. We just want to be able to be part of the game.

MCI seems to be taking the lead using the Internet Protocol (IP) to direct telephone traffic, rather than the regular circuit-switched telephone networks. How's that transition going?

We want to get 25 percent of our calls over an IP backbone by the end of the year. We're at 10 percent now. We want to move all of it over by 2005.

We want to get 25 percent of our calls over an IP backbone by the end of the year. We're at 10 percent now. We want to move all of it over by 2005.

Verizon is also doing a lot of IP backbone building. But is there anyone else matching these efforts?
This year marks a time when the use of VoIP has accelerated quite dramatically. It's been going on quietly in the enterprise domain, where Cisco has been very active in developing those kinds of products. This year is the first time you'll see traditional telecommunications carriers make serious moves in this direction. Maybe statistics will justify it as the year VoIP has happened.

Why would MCI and other carriers do this?
We'll be able to operate a single backbone rather than a bunch of different networks. By moving our voice services into that framework, there's a much, much richer communications network. I can do video, imagery, text--run programs--and there's room for new services not created yet like using the Internet component to initiate applications. It's also more efficient.

What do you mean by efficient?
The capacity to carry voice over the Internet is less than it is for the circuit-switched environment. When you dial up some person on today's telephone network, you are tying down a full 64 kilobit channel. When you're not talking, what's happening is half the channel is lying idle. If neither of you are talking, the whole 64 bits are wasted. In packet-switched, (which VoIP uses) the only time something consumes capacity is when there's something to send. The reason that's important is we don't dedicate capacity in the same way as we do in the circuit-switched world. When my voice stops and the packet stops flowing, that capacity can be used by someone else.

Changing something as large as a nationwide telephone network must cost a bundle. Isn't making all that possible a huge cost to MCI?
It's a lot less than expected. We've already had a huge investment in our Internet backbone, and we have a huge capacity. The question for us is about interfacing the traditional equipment with the Internet gear. You need SIP (Session Initiation Protocol) to set up and tear down calls, for instance. But we don't have to go and build a gigantic new Internet backbone to support this.

You view VoIP as just one of several next-generation services. What are some of the other services?
There are several already showing up. You can show up at a hotel and register your normal telephone number--as long as you can plug in your PC to an Internet service. What that means is your visibility in the communications world is now portable. Wherever you are, your communications are (there also). You can control where things go. If someone's trying to send a fax, you can vector that to your e-mail as an attachment or vector it to a different fax machine. There's an incredible amount of interaction over what had been completely separate services.

What's the most important of these next-generation services?
I think it's using the Internet as a control system. If you're like me, you have consumer equipment with remotes around the house. I can't figure out which one's which. And once I get the right one, the batteries are dead. Why not Internet-enable everything? Then it's possible to just have a single radio-based device, maybe 802.11-enabled, that lets you interact with all those appliances. You don't even have to be home. Obviously there are some security issues involved. You need strong authentication to make sure some 15-year-old next door won't reprogram your house. People will soon say there was a time when you couldn't communicate with your household appliances.

There seems to be a mixed opinion about whether businesses are using VoIP. Some say they are, some say it's a myth. What's your take?
It's in the early stages. You've seen a huge investment in infrastructure leading up to Y2K. But the big leap is when you can have inventory control system placing orders automatically. By creating a heavy-duty industrial strength IP environment, it will enable huge enterprise applications. See those things coming from Microsoft, with its .Net platform, or IBM's WebSphere? Many of our customers have VoIP, and they are buying Internet service from us. They don't have three dedicated circuits for voice, data and something else; now it's just one. I can't tell percentages, though, because it all looks like Internet packets to us. In talking to the customer base, they tell us they want it.

Why are they doing it?
They have several reasons. Instead of buying local, long-distance and Internet, they get all three through one pipe, so it saves money on the access side. In terms of utility, they can move from Instant Messaging to voice to videoconferencing in a very smooth way.

Perhaps a market just as important to VoIP as businesses is the consumer market. Does MCI have any VoIP consumers yet?
It starts with people who are almost hobbyists and early adopters. That's certainly where VoIP was in 1996. What we're starting to see now is increased interest in providing these bundled services to consumers. It will probably go hand-in-hand with broadband service.

What about carriers, the third part of this market?
Most of the major carriers are showing a lot of interest. But it shouldn't matter to my customers how I do it. I would be happy to route calls over an IP backbone and some over internal narrowband switches. How we do it is not very relevant to consumers.

It must translate somehow to the consumer?
It's relevant in something like our Neighborhood Service. If you buy a bigger pipe, you pay more. So we're pricing per pipe instead of pricing per packet. A voice service that's operated through this kind of network could legitimately be priced on this basis.