[Ndn-interest] Some questions about NDN forwarding and traffic control

Lixia Zhang lixia at CS.UCLA.EDU
Fri Mar 31 07:37:55 PDT 2017

> On Mar 31, 2017, at 5:18 AM, 董谦 <dongqian at cstnet.cn> wrote:
> Dear community,
> We are NDN beginners from Computer Network Information Center, Chinese Academy of Sciences. We are uncertain about the following issues, will you give us kind help.

thanks for your interest in looking into NDN.

> As we understand, the first stage is that routing module (NLSR or hyperbolic routing) propagates initial or long term coarse grained Name information, the second stage is that the adaptive forwarding plane obtains real time, precise, and fine grained Name information, so we get the FIB table.

I am not sure I fully understand the above description.. .
Think this way:
1/ the fundamental job of networking is to fetch the data pieces being requested by all consumers.

2/ FIB is one of the components in a forwarder helping achieve that goal.
Do note that it is the forwarding strategy that makes the forwarding decision, using FIB as one of the inputs.

3/ FIB can be populated by one or multiple routing protocols; FIB entries may also be filled by other means.

As we move forward with NDN research over the last few years, our understanding of NDN forwarding also advances.  You might be interested in checking out a few writeups, starting from section 2 in NDN TR#1 <https://named-data.net/wp-content/uploads/TR001ndn-proj.pdf> (back in 2010), stateful forwarding (2013) <http://named-data.net/publications/comcom-stateful-forwarding/>, the role of routing in NDN (2014) <http://named-data.net/publications/role_of_routing_ndn/>, etc.

> When an Interest is sent to a router, if its Name is not in CS or PIT, the router will check FIB and send it to another router (or drop).

Please see pointer 2/ in the above.

> On the other hand, Data will be forwarded by checking PIT in an intermidiate router.
> And, we have some questions:
> 1. What principle or mechanism to establish the original FIB? In the traditional TCP/IP network, the r! outing pr otocol establishes the RIB depending on the IP address, and the transmission process also uses a IP address. In NDN network, because the Interest forwarding or disseminate is based on name, whether the original FIB establishing is also based on name?

this question suggests that you may want to start with NDN TR#1 I mentioned in the above.

> For a new content, namely some new Interests with new names, how does the original FIB provide the right or effective interface for forwarding?

A similar question: when my department adds a new host, how would all routers know how to forward packets to it?

> Using flooding method to detect or maybe some active signals from the new Content located nodes?
> 2. In Multi-path and Multi-source scenario without considering In-network cache, do we just need to consider about how to divide Interests in consumer's faces?

you may want to check out the stateful forwarding paper mentioned above.

> Do Data chunks have any size range? Is it necessory to consider In-network collaborative or non-collaborative cache? 

People have built a big literature in NDN caching, Google can help find the papers.

> 3. In NDN, intermidiate routers decide how to forward packets, do we need a controller like SDN to propagates Name prefix information to mitigate the scale of FIB/! PIT and a dd global factor?

- As I have heard, there seem multiple people investigating into SDN-NDN relations.
- there are multiple ways to mitigate FIB scaling issue

my 2 cents,

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndn-interest/attachments/20170331/5cbab5fd/attachment-0001.html>

More information about the Ndn-interest mailing list