Contrary to the "classic" Internet architecture familiar to most people, today's Internet is a composition of a wide variety of networks. The IP protocol suite offers a general-purpose network design with a widely available implementation; as such, it is re-used to design and implement networks with many different purposes. Compositional architecture explains how, despite the fact that IP has not changed significantly since 1993, the Internet has evolved to meet many new requirements and challenges since then. In the following paper we argue that understanding and modeling network composition is the key to continued evolution of the Internet, and to meeting society's demands for Internet services that can be verified to meet their requirements, particularly for security and reliability. We first define networks, requirements on network services, and bottom-up reasoning that a network meets its service requirements. Next we define the composition operators of layering and bridging. Rich examples from everyday networking not only illustrate the operators, but also show reasoning across a composition hierarchy. In conclusion, we show how a shift toward more explicit use of composition would greatly enhance our ability to make the Internet better:
We have been thinking about these issues for awhile now, which is why we organized the workshop reported here:
The paper above gives motivation for a theory of networking in which the architecture of the "uber-network" we call "the Internet" is a composition of networks. "Using software engineering to teach networking" is a recent talk about it. "The geomorphic view of networking: An abstract model and its uses" is an older talk with more technical detail about the theory.
|In the past we have called it the "geomorphic view" because the complex arrangement of networks in an architecture resembles the arrangement of layers in the earth's crust.|
This paper is a gentle introduction to the theory, suitable for nonspecialists: