MCI's Cerf muses on IP multicast

By Chris Nerney
Network World, 2/9/98

The second IP Multicast Summit opens today in San Jose, Calif. It is an event designed to promote widespread deployment of the technology that supporters say efficiently transmits multimedia content over the Internet. One of IP Multicast's backers is Vinton Cerf, senior vice president of Internet architecture and engineering at MCI Communications Corp. and a man many refer to as the "Father of the Internet." Cerf, who will deliver the two-day summit's opening keynote speech, recently spoke with Network World Senior Writer Chris Nerney about where IP Multicast is at and where it is going.

At last year's summit, the message to the audience was to persuade Internet service providers to support IP Multicast. How much progress has been made toward that goal?

There's been a substantial amount of progress. More ISPs are beginning to announce multicast services. [MCI] did, for example, and UUNET did, and I think several others.

Also, the vendors who make software that will allow routers to do multicast have come some distance in improving implementation and efficiency.

But there's still an awful lot of work to be done to make multicast an efficient mode of operation for routers. It still consumes a significant amount of router resources to implement [multicast technology].

Where do you expect to see IP Multicast initially deployed?

You'll see IP Multicast facilities used more in enterprises than you will on the public Internet, simply because the resources of the virtual private net or the corporate net can be better controlled and adapted to multicast use.

Currently, what are the biggest roadblocks to widespread deployment of the technology?

To get multicast to be reliable for purposes of software distribution is a big challenge, but if it's a matter of HTML page distribution or video or audio or something which is more forgiving, we are much closer to having that sort of thing regularizable, if that's a word. Being absolutely 100% reliable using multicast techniques still begs a lot of technical questions.

What are the most important IP Multicast developments we can look for this year?

One of them will be simply increased deployment and demonstration of services. As people begin to see new applications that take advantage of this technology, that will stimulate interest on the business consumer side or the mass market side. And it will also stimulate interest among the ISPs.

What are the most prevalent uses of IP Multicast today?

I would expect to see a good deal of use on the enterprise level for distribution of price lists and other information that changes on a regular basis. On the end-user side it could be news information that's being distributed. Whether the modality is video, audio, text or Web pages is almost irrelevant. The same mechanism will deliver the bits regardless of the application.

What is MCI doing with IP Multicast?

Last year we looked at implementing IP Multicast in the public Internet and concluded not to do it during 97. We did implement it, however, in our very broadband network service, which is a special service we developed for the National Science Foundation.

Are there any emerging technologies that pose a threat to IP Multicast? Are there alternatives that also save on bandwidth and deliver multimedia information?

If you use cable or digital satellite, you can do a very credible multicast. I would expect the basic Internet to be augmented by the use of digital broadcast satellite for multicasting or digital one-way cable. You don't even need two-way cable to make it work.