[Nfd-dev] Possibility to create unicast face inside forwarding strategy

Mahyuddin Husairi mahyuddin at gmail.com
Mon Nov 27 21:37:34 PST 2017


Thanks for the invitation.

My other question is, during flooding (discovery and announcement), is
there any specific broadcast suppression technique applied?
If more than one consumer doing content discovery (interest packet
flooding) or more than one producer doing content name prefix announcement
(data packet flooding) using blind flooding, it will lead to flooding
overhead and probably lead to broadcast storm (
https://en.wikipedia.org/wiki/Broadcast_radiation) on the local network.
The only control mechanism that i found in the source code and nfd dev
guide is interest suppression which are design for different purposes.

On Tue, Nov 28, 2017 at 8:14 AM, Junxiao Shi <shijunxiao at email.arizona.edu>
wrote:

> Hi Mahyuddin
>
> Teng (CC'd) is still designing the solution. You may join an upcoming NFD
> call (see invitation on this mailing list) to discuss.
>
> Yours, Junxiao
>
>
> On Mon, Nov 27, 2017 at 8:56 AM, Mahyuddin Husairi <mahyuddin at gmail.com>
> wrote:
>
>> If the suggested feature on issue #4283
>> <https://redmine.named-data.net/issues/4283> is not accepted, what is
>> the way that self learning strategy switch the forwarding mode from
>> multicast to unicast after discovery?
>> We just assume the relevant unicast face exist? and if not exist how?
>>
>> On Mon, Nov 27, 2017 at 9:11 PM, Junxiao Shi <
>> shijunxiao at email.arizona.edu> wrote:
>>
>>> Hi Mahyuddin
>>>
>>> I need your opinion about enabling forwarding strategy to create face.
>>>> At least unicast face (raw ethernet or udp) for dynamic unicast face
>>>> features in my NDN based MANET solution.
>>>>
>>>
>>>> 1. What is the possible dirty hack way to enable it?
>>>>
>>> Do not look for or use "dirty hack way". The harm of doing so is
>>> described on https://yoursunny.com/t/2017/academic-papers/ "program
>>> quality" section.
>>>
>>> AFAIK, to create face, it is part of NFD management protocol API. Is
>>>> that possible to access that API just like face-module.cpp in NFD tools?
>>>>
>>> No, because forwarding plane is not an application. It cannot send
>>> packets to itself.
>>>
>>> 2. Or is there any better way to implement it?
>>>>
>>> The better way is to avoid the program in the first place.
>>> A face is a generalization of an interface. Therefore, an Ethernet face
>>> shall support both multicast and unicast on the same physical interface.
>>> NFD currently implements Ethernet unicast like tunnel, which is wrong.
>>> You may achieve unicasting by implementing https://redmine.n
>>> amed-data.net/issues/4283 . I heard from last NFD call that NFD team is
>>> planning to reject this issue, but I believe this is the right way to go.
>>>
>>> Yours, Junxiao
>>>
>>
>>
>>
>> --
>> Human knowledge Belongs to The World
>>
>
>


-- 
Human knowledge Belongs to The World
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/nfd-dev/attachments/20171128/f8e7009b/attachment.html>


More information about the Nfd-dev mailing list