[ndnSIM] Multicast Strategy , strange behavior

[Ahmed Sadek]

Thank you Junxiao for the detailed and quick answer.

1- Is there a way to turn off this feature to test the broadcast strategy ?
I tried to set :

*const time::nanoseconds DeadNonceList::DEFAULT_LIFETIME =
time::milliseconds(1);*

but didn't do the trick. It seem the only way is that there should be a
delay on the links from the start so interests don't arrive in parallel.

2- What if there are two clients contacting the server asking for the same
content (same interest) at the same time, this mean only one of them will
get the content as backbone router will suppress one of them.

Thanks!
Ahmed


On Thu, Mar 31, 2016 at 4:11 PM, Junxiao Shi <shijunxiao at email.arizona.edu>
wrote:

> Hi Ahmed
>
> After an Interest is multicasted onto multiple paths but those paths join
> at an upstream ("backbone" in the topology), the upstream would detect that
> the Interests have duplicate Nonce and suppress all but the first Interest,
> and send back the Data only along the path on which the Interest arrives
> first.
>
> Duplicate Nonce detection is done through the PIT entry and the Dead Nonce
> List. See NFD Developer Guide "Forwarding - incoming Interest pipeline" and
> "Tables - Dead Nonce List" sections for more information.
> Detection through PIT entry is reliable: it can catch every incoming
> Interest with duplicate Nonce and suppress it, as long as the PIT entry
> still exists. A PIT entry stays for about 100ms after the Interest is
> satisfied ("straggler timer"). When the three paths in the topology have
> the same or almost same delay, it's almost guaranteed that the PIT entry
> exists when the second and third Interest arrive, and thus they are
> suppressed.
>
> Detection through Dead Nonce List is selective: it's only used to prevent
> "persistent loop" (NFD Bug 1953).
> When the second and third paths have a delay that is more than 100ms
> longer than the delay of the first path, the duplicate Nonce cannot be
> detected by throug the PIT entry, and is usually not detected by the Dead
> Nonce List. In this case, the Interest would be permitted to lookup the
> ContentStore, find a match, and the cached Data is sent back.
>
> Yours, Junxiao
>
> On Thu, Mar 31, 2016 at 5:20 AM, Ahmed Sadek <don1559 at gmail.com> wrote:
>
>> Hi All,
>>
>> I am using the following topology to try multicast strategy (broadcast).
>> So I noticed that all requests are broadcasted but I am receiving only
>> replies from one link (client->R3->Backbone).
>>
>>
>>
>>
>> ​​
>> When I increase the delay on one of the other two links during the
>> simulation, I start receiving data replies on them also.
>>
>> I am assuming that on the intersection node (backbone) something stop
>> data replication on the three interfaces and when there is a delay between
>> interests this stop doesn't work.
>>
>> Can someone guide me, on what is happening here and why no data
>> replication happening from the start ?
>>
>> Thanks!
>> Ahmed
>>
>> _______________________________________________
>> ndnSIM mailing list
>> ndnSIM at lists.cs.ucla.edu
>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20160331/3f45a103/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: broadcast.png
Type: image/png
Size: 58518 bytes
Desc: not available
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20160331/3f45a103/attachment.png>