Talk:Named data networking
|WikiProject Computing / Networking|
Call for comments
I just created this page because I couldn't find much about this new idea on the Web, after seeing Jacobson's highly interesting Google Tech Talk. The talk failed to address a few points though, and that is what I'm writing about in the last paragraph. Please extend the article so I can learn more from it ! :-)
Also, I'm interested to hear what people think about the idea itself. It seems quite revolutionary to me, and I didn't put most of what I think in the article as that would be too much original work for an article. So here they are, in no particular order :
- Would that kind of network mean the end of web applications as we know them at the moment ? It seems to me that it would make most sense to diffuse the raw data (so that it can be re-used and cached by many) and leave the user-specific formatting for the endpoint to do. Also, programs could be diffused first, using a platform-agnostic form (the revival of the Java applet ?), and then only the data would have to move around.
- Does that sound like what's happening in Grid computing to you too ? In the grid middleware I know best (EGEE's gLite), resources have abstract names that are resolved to the (physically) closest match, with a resource broker possibly moving stuff closer to you on your behalf.
- What about what I said in the last paragraph about real-time use cases ? Would it be best to keep both a diffusion-based infrastructure and a conversation-based one, or turn conversation-based uses into special cases of diffusion (with only one listener) ?
Eventually I'd like Jacobson to see that but I guess he's a fairly busy man :) Aftereight 23:18, 6 May 2007 (UTC)
I only just saw Jacobson's presentation on Google video, and loved it.
However, there is prior art, in the form of Andy Tanenbaum's Amoeba distributed operating system project, especially its approach to storage. Data storage was abstracted into the network, in that every object was assigned a unique (encrypted) handle, and once you got the handle, you could retrieve the data from the file service, an agent contacting whichever block/file server actually held a matching copy. This is similar to what Jacobson refers to as a Name referring to a single, immutable object.
An amoeba file handle is supposed to be an opaque string, and to find the handle for an object, you'd have to search for it in a directory service which adds an extra layer on top of the file service.
Ted Nelson/Project Xanadu
Most of these ideas were pioneered by Ted Nelson and Project Xanadu between 1979 and 1981 - see Literary Machines and "The Future of Information" (1997). It's great to see Van Jacobson bringing new life to these concepts. -- Andrew Pam <email@example.com> 00:19, 26 February 2009 (AEST) —Preceding unsigned comment added by 22.214.171.124 (talk)
Personally I do not see the ideas/vision of ted Nelson the same as that of Named-data networking. -- Lixia Zhang
I also don't see why Ted Nelson is mentioned here. Project Xanadu has nothing to do with CCN. The only thing in Xanadu that comes close to CCN are "Rules" 12-14 in the Xanadu philosophy, but rule 12 is true for a lot of other ideas (with DNS I also don't need to know where a website is stored), rule 13 is caching as it is done by any web proxy and rule 14 is just a plain backup system. If no one is objecting I'll be removing the reference to Ted Nelson soon. Stipa (talk) 13:45, 21 February 2011 (UTC)
The official project name would seem to be Named Data Networking as funded by the National Science Council. The term content-centric networking is used by PARC one of the collaborators of the project, but not anywhere on the NDN official site.
- Since no one else has commented, I have gone ahead and done this. SpinningSpark 15:50, 13 November 2012 (UTC)
What is this protocol FOR?
I find this article very confusing, and the stuff on the consortium website isn't helping me much. I'm hoping the article can eventually express what functionality the protocol has, for 99% of users, that TCP does not. It looks to me like it could help with traffic flow inside of networks belonging to content providers (netflix, youtube, etc.) and immediate CDN providers for those companies. It also would seem to preclude anonymous publication, and build DRM in at a networking level. I can see why Comcast is in the consortium, along with Verisign and several networking hardware manufacturers, since all of them might make a boatload of money if it replaces TCP. I can see reasons why the gov't might want to fund a protocol which forbids anonymous publication of content. But, for the life of me, I cannot see why anyone outside of the content industry would want to use this protocol. If there are any reasons, this article might want to explain them. If there are not, it might mention that instead. 126.96.36.199 (talk) 20:28, 6 September 2014 (UTC)
Work In Progress
I am in the process of redoing this page since there seems to be a lot of confusion. For those interested in knowing more and/or contributing to the re-factoring of this page, I believe this paper NDN Sigcomm Paper will likely contain the vast majority of information that will be distilled and added to this article. For those REALLY interested in being fully informed on the information needed to be learned in NDN, I believe the required reading list NDN Community Meeting (NDNcomm 2014): Architecture, Applications, and Collaboration, the paper listed previously is the first member of this list, will have more information than is appropriate for a Wikipedia article. — Preceding unsigned comment added by Mickeysweatt (talk • contribs) 23:11, 9 September 2014 (UTC)--Mickeysweatt (talk) 18:33, 10 September 2014 (UTC)
Update, I completed a first draft which has gone through a first round review by researchers for correctness. Waiting on collaborator feedback, and requests to determine how to continue this effort. Mickeysweatt (talk) 23:59, 5 October 2014 (UTC)
Is this protocol running anywhere?
NDN looks like a pipe dream. If there is anyone actually running this (on bare hardware, not as an overlay network on top of TCP/IP) then someone should add a section to the article, about where the running network(s) actually are (and perhaps how to join it). If there is no real network, then the article should make it clear that this is a paper design, not a protocol that's running on any network known to man. Gnuish (talk) 01:39, 22 August 2015 (UTC)