[Ndn-interest] Questions about broadcasting an interest

Christos Papadopoulos christos at colostate.edu
Thu Dec 15 07:10:32 PST 2016


Laqin

I think we need more information before we can give you an answer. Why 
is the broadcast happening? Is only on the first hop from the user? Are 
you talking about a MAC layer broadcast or a message sent to all faces 
at the NDN layer?

The following may or may not relate to your question.

The way I understand the question is that, the NDN stack at the end user 
decides to broadcast an interest over all available faces, perhaps 
because it has no route or as part of an experiment looking for a better 
route to the content. I would not expect implosion at the end point, 
since typically there are not that many next hop potential responders 
(implosion would happen if there were many next hop NDN routers that 
receive the broadcast *and* they had the requested content). While I can 
construct scenarios where that may potentially happen, I can't imagine 
they will be very common.

I expect the broadcast (at the MAC or NDN layers) to be handled by the 
strategy layer. Once responses start to arrive, the strategy layer will 
select one or more faces (possibly those that responded faster) and send 
future interests there.

Christos.


On 12/15/2016 12:56 AM, Marc.Mosko at parc.com wrote:
> Unless some extra mechanism is used, one could have a Data object 
> implosion problem when many other systems have a Data to satisfy the 
> interest.  The sender’s PIT will filter out multiple responses, but it 
> is expensive network-wise.  I am not sure if NFD will unicast or 
> broadcast the response Data.
>
> The old CCNx 0.8 forwarder would apply a randomized delay before 
> answering a broadcast and if it saw another response during that 
> period, it would suppress its transmission.  This is like a gossip 
> style suppression sometimes used in MANETs.  It does rely on the 
> response Data being broadcast too so everyone else sees it.  In a 
> high-speed LAN, using this kind of broadcast-broadcast communications 
> is pretty inefficient as a large switch often runs much slower with 
> broadcast traffic.  Over other styles of networks, like wifi, it might 
> make sense, though one would lose RTS-CTS and ACKs, which introduces 
> other problems.
>
> Personally, I think broadcasts should be use sparingly, such as for 
> initial discovery or to elect a cluster head, etc., and then use unicast.
>
> Marc
>
>> On Dec 14, 2016, at 10:34 PM, Laqin Fan (lfan1) <lfan1 at memphis.edu 
>> <mailto:lfan1 at memphis.edu>> wrote:
>>
>> Hello all,
>> I have questions about broadcasting an interest, if a consumer asks 
>> for content by broadcasting its interest over all available 
>> connectivity, any node hearing the interest and having the satisfied 
>> data will respond with a data packet, the consumer will receive the 
>> first data packet? and how? what about the others? And if there are 
>> two data packets arriving at the consumer at the same time, what happens?
>> Could anyone please make me clear? I really appreciate it.
>> Thanks,
>> Laqin Fan
>> _______________________________________________
>> Ndn-interest mailing list
>> Ndn-interest at lists.cs.ucla.edu <mailto:Ndn-interest at lists.cs.ucla.edu>
>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndn-interest
>
>
>
> _______________________________________________
> Ndn-interest mailing list
> Ndn-interest at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/ndn-interest

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndn-interest/attachments/20161215/b7c9d060/attachment-0001.html>


More information about the Ndn-interest mailing list