CAREER: From Protocol Support to Applications: Elevating Multicast to a Ubiquitous Internet Service

University Of California-Santa Barbara, Santa Barbara CA

Investigators

Abstract

The investigation of multicast communication as a service in the Internet has been an issue for nearly a decade. And in that time, much progress has been made. From its early inception as a means of streaming audio and video to researchers around the world, the Multicast Backbone (MBone) has grown into an infrastructure supporting commercial applications. But, even though great strides have been made, there are still numerous, serious technical problems to be solved. The rapid growth in multicast users, and the rapid growth in the interest in multicast means not only more and more multicast-capable sites are being added but more demands are being made for a richer set of multicast-base services. As multicast grows in use, the need to make good on the promises of scalability becomes even more important. Furthermore, the complexities of multicast protocols are significantly greater than for unicast protocols. Multicast protocols must communicate state information to manage the dynamic growth of multicast trees and ensure traffic flows only on necessary links; they must implement congestion control among many receivers instead of just one; and they must support management fuctions in an environment with more complexity than traditional management protocols were designed to handle. If multicast implementors have learned anything, it is that fuctions that are easy for unicast protocols are not so easy for multicast. Once one problem is solved, there is another set of problems to be identified and then solved. But even with all the hurdles to overcome the potential utility of a ubiquitous multicast service is immense. Once the protocol issues are more fully solved, the potential for new multicast-based applications and services will carry the Internet to a new level of utility. While multicast is quickly evolving and industry is beginning to take notice, there are still problems to be solved by the academic community. In the last several years, multicast has developed into more than just a research initiative. Multicast is now of interest to many companies. These companies see the value in multicast and also see the value in having multicast as a premiere Internet service. However, just because multicast is being touted as a service ready to roll-out, does not mean all of the technical problems have been solved. More importantly, it would be dangerous for the academic community to believe industry will assume full responsibility for solving the most important problems. The role of academia, and the goal of the research plan described in this proposal, is to support the investigation of problems limiting the effectiveness of the Internet-wide deployment of multicast. Furthermore, results of this research will hopefully be applied in next-generation high speed regional (Ca1REN-2) and national (vBNS and Internet2) networks. The basic goal of the research outlined in this proposal is to continue evolving multicast into an Internet service. A number of multicast research topics and deployment issues are targeted. Because multicast communication is both the subject of ongoing academic research and deployment issues, this CAREER proposal is an investigation of issues along two axes: (1) networking issues ranging from the data link layer up to the application layer, and (2) the type of research ranging from basic research to applied research. Significant thought has been given to organizing this CAREER proposal in a manner that reflects the changing philosophy of academia. The first axis is developed based on the philosophy that applications drive the need to develop new Internet services. In the network, Internet service enable new paradigms and applications to be developed. The second axis is developed based on the philosophy that a new era has begun in which successful faculty must be capable of addressing both fundamental research issues and applied research issues across a range of areas. The proposed education plan relates very closely to the proposed research plan. The educational mission is to teach undergraduates and graduates the fundamental aspects of networking and then to relate them to how things work in the Internet. The key strength of the educational mission is its close ties to the research plan. In particular, the digital classroom is the enabling concept which bring education and research together. The digital classroom is an ideal application. It embodies much of my research philosophy and is an excellent bridge for students. They can see the need for distance learning at the macro-level and can use that knowledge to put into context the micro-level projects. The best way to evaluate a digital classroom is to use it, and the best way to build a digital classroom is to have students (not as chep labor but as eventual "useer") build it. The classroom provides a traditional environment for students to learn theoretical topics, the classroom can also be a place where traditional topics using non-traditional methods, and the classroom is an ideal place to motivate students to think beyond textbooks and closed-end problems. While distance learning is the subject of much research, academic faculty are in the unique position of using, evaluation, proposing, and developing the techniques and tools that will someday define and enable distance learning.