You will need a project plan--a testing phase that will cover every router in the network. When you're done with the assessment, the result may well be "more and more of our network will be multicast-enabled as we replace our older (Vendor X's) routers with the newer (Vendor Y's) routers." This seems to me to be the very definition of the term forklift upgrade.

Sounds Like a Project This project plan needs to start with the premise that until your entire network is multicast-enabled, multicast will be limited to certain pockets of multicast-enabl ed segments or be handled inefficiently. This is acceptable if the inefficiencies are limited to small parts of the network. This leads to some type of quantification of how much multicast and peer-to-peer communication will actually take place. Next, inventory the various technologies that make up the backbone and WAN--in particular different WAN technologies, satellite versus terrestrial, frame relay versus dedicated lease line, dial-up versus high bandwidth, ATM versus Switched Multimegabit Data Service (SMDS)--what you end up with on your list will have different implications regarding multicast traffic. This is probably the crux: Anything that touches "wide area" slow links is certainly the critical path for upgrades if it isn't multicast capable.

Are you interested in receiving Internet multicasts? First ask, "Has my Internet service provider (ISP) turned on multicast?" Most ISPs haven't. Not all of the Internet is multicast-enabled. As an alternative, you can tunnel through to other multicast site s using the Mbone. The Mbone uses standard unicast IP tunnels to encapsulate multicast packets. As a result, standard IP routers and switches can forward them to the destination multicast site.

One can create an iterative approach, as opposed to undertaking a forklift upgrade. As users just ify and fund the expense, create communities of multicast nodes. As the number of these communities grows, upgrade the intervening hops. The result is a high-quality switched infrastructure that fully understands multicasting. Ultimately, the next generation of Internet standards, such as quality of service (QoS) for support of real-time multimedia, will require the same infrastructure.

For additional information on multicast technologies, check out the IP Multicast Initiative Web site at www.ipmulticast.com and www.icast.com.

What's your experience been? Enabled multicast and lived to tell the story? Any examples of particularly atrocious useful content to bandwidth hogs? Using eight-year-old routers and proud of it? Send me some e-mail at bwalsh@nwc.com.

Brian Walsh is a senior consultant with Cap Gemini America in Portland, Ore. He can be reached at bwalsh@nwc.com.