[Ndn-interest] The best-route strategy

[Klaus Schneider]

Yeah, good point. I guess I was thinking about the old CCNx 
"load-balance" strategy.

Something similar in NFD: 
https://github.com/schneiderklaus/NFD/blob/NFD-0.3.4-ndnSIM-pcon/daemon/fw/pcon-strategy.cpp

>     // Choose face according to current forwarding percentage:
>     double forwPerc = 0;
>     for (auto face : eligbleFaces) {
>       forwPerc += measurementInfo->getforwPerc(face->getId());
>       if (r < forwPerc) {
>         outFace = face;
>         break;
>       }
>     }

Where r is a random double between 0 and 1.

Best regards,
Klaus



On 4/30/19 5:01 AM, Lan Wang (lanwang) wrote:
> Klaus,
> 
> I thought the multicast strategy just sends the same packet to all the faces in a FIB entry, not splitting traffic among them.
> 
> Lan
> 
>> On Apr 30, 2019, at 1:17 AM, Klaus Schneider <klaus at cs.arizona.edu> wrote:
>>
>> Yeah, if costs are equal it uses some tie-breaker, I think the face-id.
>>
>> The best-route strategy will never split up traffic between multiple faces. If you want that, you should look into the multicast-strategy (https://github.com/named-data/NFD/blob/master/daemon/fw/multicast-strategy.hpp) or write your own.
>>
>> Best regards,
>> Klaus
>>
>>
>>> On 4/29/19 3:22 PM, Yasmine Belaguid wrote:
>>> Hello,
>>> Thank you for your reply Sir.
>>> For the grid topology all costs are apparently equal, so why is it using only one path? is it relying on the interface in this case (it takes the lowest one which is 0)?
>>> Best regards.
>>> Le lun. 29 avr. 2019 à 03:35, Klaus Schneider <klaus at cs.arizona.edu <mailto:klaus at cs.arizona.edu>> a écrit :
>>>     Hello,
>>>     I think you can find the answers to your questions in the NFD Developer
>>>     Guide:
>>>     https://named-data.net/publications/techreports/ndn-0021-10-nfd-developer-guide/
>>>     On 4/27/19 4:19 AM, Yasmine Belaguid via Ndn-interest wrote:
>>>      > Hello,
>>>      >
>>>      > I am a student at INTTIC (Algeria), I am working on QoS-based
>>>     routing. I
>>>      > have some questions about the best-route forwarding strategy.
>>>      >
>>>      > Since I tried the grid topology with 4 nodes using the best-route
>>>      > strategy I observed that the consumer sends Interests to the
>>>     producer
>>>      > choosing only one route, so my questions are :
>>>      >
>>>      >   * For the first time, does the consumer flood Interests to discover
>>>      >     the network? or what are the metrics that he uses to choose and
>>>      >     calculate this route?
>>>     No, the consumer does not flood the first Interest but sends it on the
>>>     shortest path (= lowest cost face).
>>>     You can set the link metric in your simulation file (search for
>>>     "metric"
>>>     here http://ndnsim.net/current/examples.html), and the default will
>>>     be 1.
>>>      >   * The cost used by this strategy is it hop-count?
>>>     Yes, if you didn't set any other link metrics.
>>>     Best regards,
>>>     Klaus
>>>      >
>>>      >
>>>      > Thank you!
>>>      > Best regards.
>>>      >
>>>      >
>>>      > Belaguid Milouda Yasmine
>>>      > Student at INTTIC
>>>      >
>>>      > M +213 699 90 73 25 <tel:+213(0)+699+90+73+25>
>>>      >
>>>      > E yasmine.belaguid at gmail.com <mailto:yasmine.belaguid at gmail.com>
>>>     <mailto:yasmine.belaguid at gmail.com <mailto:yasmine.belaguid at gmail.com>>
>>>      >
>>>      > <http://github.com/yasminebelaguid>
>>>      > <http://www.linkedin.com/in/milouda-yasmine-belaguid/>
>>>      >
>>>      >
>>>      >
>>>      > _______________________________________________
>>>      > 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