Notes

Outline

Internet2 K20 SEGPs and IP Multicast Joanne Hugi (hugi@oregon.uoregon.edu) Computing Center University of Oregon

IF the I2 K20 SEGP program is to be about more than cheap best effort IPv4 service... … we need to focus on high performance applications which have not been able to be easily accommodated on the commodity Internet. In the K20 context, application scalability is particularly crucial given that the size of the K20 audience is >> size of university audiences (e.g., 600 K vs. 60 K in Oregon)

IP Multicast: An Exemplary Advanced Application & Is Scalable IP multicast is a suitable advanced application. -- IP multicast has a proven track record as a success in the traditional I2 university space; it can be just as successful for K20 -- It is designed from the get-go to easily scale to support K20+ sized audiences

Deploying IP Multicast Is a Learned Craft... Deploying IP multicast isn’t “rocket science,” but on the other hand, neither is it putting together “Legos” (yet). It is moderately hard, but can be implemented if you use reasonable diligence. I2 member sites and gigapops have developed IP multicast expertise, and can transfer that practical knowledge to the K20 SEGP network community

Doing Multicast Doesn’t Require Training “Everyone” Opportunities for intensive training of K20 staff and end users are often quite limited, and technology topics compete with non-technical material which must also be covered. Training time can be a real issue. The best technologies for K12 are those which do NOT require intensive or extensive training for “everyone.” The best applications “just work.”

Once Native IP Multicast Is Enabled, “It Just Works” Once the K20 network engineers have done their part with the help of their I2 partners, to a first approximation, many IP multicast applications will “just work” Example: Cisco’s IP/TV IP multicast video viewer, delivering TV-quality video over existing LANs to vanilla Windows desktops

There May Be Some Gotchas to Work Through... By sheer luck of the draw, some K20 sites may have older network equipment, or equipment from vendors which poorly supports native IP multicast Other K20 sites may need to build out better wide area connectivity to take advantage of I2 connectivity (some of our Oregon partners are in the process of deploying their own fiber to our gigapop); 1xT1 or 2xT1 just won’t do it.

Gotchas (continued) Our experience has also been that K20 tends to be MORE firewall oriented than universities, and firewalls often interact poorly with IP multicast traffic Content, content, content. As always, finding interesting and relevant content which can lawfully be distributed can be a challenge.

Clarifying Expectations K20 users (like all users) do need to have some of their (possible) expectations appropriately calibrated/clarified, e.g.: -- one way MPEG1 video delivery is not     two-way H.323-style video conferencing -- one way MPEG1 video delivery is not     RealAudio style streaming video

Expectations (continued) -- no multicast via most dialups or third     party ISPs (including most DSL and most        cable modem providers) -- sourcing content is significantly different     from just viewing content -- debugging (can be) non-trivial

Conclusion In spite of potential issues, IP multicast is an advanced application with tremendous potential in the Internet2 K20 SEGP space. We would strongly encourage all I2 sites working with K20 SEGP users to consider making IP multicast a deployment priority for their K20 SEGP partners.

Conclusion The University of Oregon will develop multicast suitable documentation tools for the K20 environment. These materials will be aimed both at the end user audience as well as network engineers.