University Of Illinois at Urbana-Champaign CS427(Summer 2001) Project

Architecture Framework For Active Network

What is the impact on layered reference model?

The web browsers allow users to interact with what they perceive to be "the network" without distinguishing among the millions of routers, domain name servers, and web servers that conspire to provide the service. It may be time for practitioners to re-evaluate their abstractions and start thinking about the network at a higher level.

Current thinking concerning network architecture has its roots in the layering of abstractions codified in the OSI Reference Model [11] . Although the model has proven quite useful it is beginning to show cracks that should be addressed:

Services at or below the network layer are presumed to be user- and application- neutral;

It deals poorly with upper layer services that are physically interposed between the communicating end points. Application relays can model these cases, but they are far from elegant;

It does not model the "recursion" that occurs at the network layer, i.e., the tunneling of networks over each other;

The upper layers, which have never been particularly satisfactory, are of diminished importance, given that active technologies enable the exchange of modules that implement application-specific protocols.

We are not certain what form a new model might take, but suggest that it will be more component-based than layered. It might distinguish primitive functions, such as cell relaying and IP "fast paths", from computationally active functions, including those that configure the fast path components. Architecturally, these two types of components might be viewed as peers rather than layered upon each other. Such an architecture might also give rise to new hardware activities, such as the development of switching technology that "caches" fast paths and is highly responsive to active capsules.[b7]

Index Page

Further info on model and related to network
transported by Go FTP FREE Client

	Services at or below the network layer are presumed to be user- and application- neutral;
	It deals poorly with upper layer services that are physically interposed between the communicating end points. Application relays can model these cases, but they are far from elegant;
	It does not model the "recursion" that occurs at the network layer, i.e., the tunneling of networks over each other;
	The upper layers, which have never been particularly satisfactory, are of diminished importance, given that active technologies enable the exchange of modules that implement application-specific protocols.