IP was Middle School, Named Data Networking is College

With Cisco, Intel and a bunch of academics behind content-centric NDN Internet architecture, momentum picks up

Much of the Named Data Networking (NDN) project codebase is still at the Version zero-dot-something level. But things are nevertheless starting to get real for this content-centric architecture designed to blast past today’s host-based and point-to-point Internet scheme to one more suited for accessing applications across hugely scalable networks that are mobile and extend to all sorts of sensor-equipped things.

The NDN project team hosted more than 100 people from 63 institutions and 13 countries at UCLA last week to share progress and ideas at the second annual NDN Community Meeting (the first such meeting that we wrote about a year ago in "UCLA, Cisco & more join forces to replace TCP/IP" drew attendees from 39 institutions and 7 countries). A quick search of the IEEE digital library through early October turned up 42 papers from 2015 with Named Data Networking in their titles, whereas there were 52 new Named Data Networking papers for the entire span of 2014.

The open source NDN codebase, which is being run through its paces across an international test bed, was even put to use at the NDN Community meeting itself for conferencing purposes.

In addition to academic experts, influential network and computing vendors such as Cisco, Alcatel-Lucent, Huawei and Intel are behind this movement and took part in NDNComm, which featured sessions on topics such as advanced routing  -- “hyperbolic” no less -- serious science applications and trust/security/encryption.


The NDN Project even has a new Twitter account -- @nameddata -- so it must be the real deal, right?

That meeting was immediately followed, up the coast in San Francisco, by the 2nd ACM Conference on Information-Centric Networking, where NDN was one of the stars. And prior to all this, UCLA hosted the first NDN hackathon.

NDN's spiritual leader, Internet Hall of Famer and UCLA adjunct professor in computer science Van Jacobson, delivered keynote addresses at both NDNComm and the ACM event. NDN has its roots in content-centric networking, a concept that Jacobson started discussing publicly about 10 years ago while at Xerox PARC.

Jacobson actually took the podium at NDNComm after Jeff Burke from UCLA’s School of Theater, Film and Television. Burke set the stage by focusing on how researchers in the arts and beyond are intensely interested in where the Internet is going as a means for communicating stories, and not just as in movie plots, but in the broader sense of the stories of our lives (think personal health, and tapping into everything from genome to Fitbit data). Named-based data could be a godsend for exploiting Big Data, including information served up by a sensor-based Internet of Things, and for supporting emerging applications, such as video streaming like you’ve never seen before, he said. “[NDN can] take us far beyond the connection-oriented model that a lot of our applications started in,” Burke said, adding that people should be excited about “the opportunities to be designing for applications that are coming down the pipeline and that are at the fringes in the future, and that that’s going to intersect in exciting ways with NDN.” 

Jacobson took it from there (and you can see/listen to his presentation in the YouTube video below, a bit after the 1-hour mark): 

“When you’re telling stories you need a vocabulary, you need idioms, you need ways to express those stories, and we’re telling stories as computer scientists, particularly networking researchers, with our protocols, with APIs, with packet formats. And the thing that kicked off all the [Future Internet Architecture] efforts was this frustration in the community that telling our stories with IP is getting really hard. It’s like our tools are working against us, not for us. We spend like 80% of our effort trying to bash the tools into shape and only spend 20% on the problems. In vocabulary terms, IP is like a good middle school education. There’s a lot of things you can say and communicate in society but it’s not so great for writing a poetry volume or a thesis. The Web kind of brought us up to high school and the expressivity in things like XML and Ajax. You can work at a higher level, accomplish a lot more, but you’re still pretty limited by this conversational client-server model. The real goal of NDN was to get us into college. To have a set of idioms, vocabulary that would let us naturally express the high-level things that Jeff presented, that we want to see.” 

Jacobson went on to highlight new thinking in data transport, Internet security and routing. Routing, he said, has been long overdue for an upgrade. "For the past 40 years, unfortunately, we've been doing routing based on Paul Baran's original model of building paths through graphs and then binding identities to the ends of those paths. There's a lot of downsides in doing routing that way." He pointed to new geographic routing models prototyped in NDN that can be more efficient.

If you’re playing catch-up on NDN, don’t sweat it, as it really is still early despite the progress that's been made over the past few years. The good news is that a growing body of content is available online to help you bone up. NDNComm is making its presentations available in streaming and PDF formats, and a slew of tutorial videos are available from the ACM event.

This story, "IP was Middle School, Named Data Networking is College" was originally published by Network World.

NEW! Download the State of the CIO 2017 report