Architectural Support for Interoperability

Although it would be nice if the world converged on a single suite of protocols, that is unlikely to happen in the near future. New networking technologies as well as new applications are constantly being developed and deployed; eventually this leads to a need for new protocols. The Internet today is not a single-protocol network. While TCP/IP remains the primary protocol suite, other protocols often share transmission facilities, either natively or encapsulated as data within IP. (The transition to IP version 6 is a prime example of this.)

This fact has several consequences. First, end systems end up supporting multiple protocol suites, none of which was really designed to operate in a multiprotocol environment. Applications designed to use only one protocol suite may restrict the set of end systems with which they can be used. Moreover, when applications are designed to run over multiple protocols, the problem arises how to choose one for each particular communication instance. Should this problem be solved by the user, by the application program, or by the protocol implementations themselves? Ideally this problem should be solved in a manner invisible to the user. This project deals with both issues: helping applications deal with multiple protocol stacks, and lowering the barriers to new protocol deployment and penetration. We investigate ways to maximize interoperability at minimum cost among systems, old and new, that support a variety of protocols, architectures and implementation environments.

Publications:

Students

Ken Calvert
Last modified: Tue Dec 30 13:34:50 PST 1997