information-centric networking

Think about how you ask for things in everyday life. When you want a particular book, you usually do not care which shelf in which warehouse it sits on — you just want that book. But today's Internet is built around addresses of places: every packet is stamped with the IP address of a host to talk to, and getting your data means setting up a conversation with that specific machine. Information-centric networking is a family of research designs that asks: what if the network were organized around the information itself — named content — rather than around the location of hosts?

Information-centric networking is the umbrella idea, and named-data networking is its best-known concrete design. The shared principle is that you request data by name, and the network is responsible for fetching it from wherever a valid copy exists. This naturally builds in three things that are awkward to retrofit onto IP. First, in-network caching: because requests name content, any router can keep a copy and satisfy future requests, so popular data does not travel the long way every time. Second, security on the data: content is signed by its producer, so you verify the bytes directly instead of trusting the server or the channel they came over. Third, easy mobility and multipath: since you are not tied to one host's address, the nearest available copy will do.

Why it matters: it is a serious attempt to redesign the Internet around how we actually use it — overwhelmingly to retrieve named content (videos, pages, files) rather than to hold point-to-point conversations. The candid limits: it is a research and clean-slate effort, not a deployed Internet replacement. Open challenges include scalable routing over a huge, unbounded name space, privacy (the names you request can reveal a lot about you), and trust management for signatures — plus the sheer momentum of the IP Internet, which is good enough that wholesale replacement is unlikely. Its ideas, though, keep leaking into mainstream systems like CDNs.