By Bill Brogden
It is appropriate from time to time to step back and look at what the more far-out thinkers are speculating on. The revolutions which took us from dial-up time sharing and BBSs to the present internet took some imaginative jumps by people who were "far-out" at the time. People like:
- Ted Nelson - who brought us the concepts of "hypertext" and "hypermedia" way back in the 60's
- Vinton Cerf - one of the "fathers of the internet."
- Tim Berners-Lee - who put hypermedia and the internet together to create the World Wide Web.
Several approaches to large scale content-centric networking have been proposed. Right now, Project CCNx, a PARC sponsored open-source project seems to be the most active. PARC Research Fellow Van Jacobson is the visionary behind this research. I recommend his Google Tech Talk video presentation on the evolution of the concept.
Jacobson articulates the following problems with the existing URL oriented approach:
- Bandwidth wasting Duplication Data from a single source gets duplicated for every user. Jacobson reports seeing 90% of traffic as duplicate packets for popular sources.
- No Built-in Security Authentication, data signing and trust are add-ons to the URL model.
Widely recognized for his work on improving TCP/IP networking, Jacobson has become convinced that current network research is at a dead end in terms of significant cognitive jumps in architecture so it is time to step back and think about alternatives. Jacobson calls the present approach a "conversation model" which depends on a connection between endpoint addresses. Named data objects would create a "dissemination model" which does not depend on endpoints.
Content-centric networking hearkens back to Ted Nelson's attempt to let the user ignore the details of how content is located. Jacobson points out that the move from the telephony model of dial-up to the packet switching model of the Internet required a big increase in network intelligence to support DNS lookup and packet addressing. Moving to CCNx will also take a huge increase in network computing power but this is clearly already happening, for example with "cloud" computing.