Networking of Information (NetInf)

This research was conducted within SICS Center for Networked Systems.

As of today the predominant style of communication on the Internet is for a client to request information from a explicitly named server. But we think it would be nice if the "server" could be eliminated from this request style operation. The general NetInf idea is to request the information from "the network" rather than from an explicitly named server. The client doesn't care much about where the information comes from, what really matters is that the information is authentic, i.e that the information is identical to the information requested.

Removing the server from the communication loop can make the network more robust. If your network is partitioned from the network where the server resides, you may still be able to retrieve a copy from within your network. If the server happens to be overloaded with requests NetInf communication will not focus communication to the server and will thus avoid overloading the networks surrounding the server.

NetInf is part of the CNS vision of the Reliable Internet. As part of the CNS NetInf project we investigate and evaluate different approaches to creating a network of information. We are exploring how publish/subscribe and content routing algorithms may be used to realise our vision and the limitations various solutions introduce. Our goal is to design a new platform for information distribution with a communication service that better suites applications than the current Internet. We will then use the experimental platform to validate our ideas on some application scenarios.

One application scenario we investigate is IPTV distribution.