[ndnSIM] Abilene topology information

[Alex Afanasyev]

Hi Saeid,

I haven't used MPI, I just run different runs/scenarios in parallel (using ./run.py script, which in current form runs as many instances as the number of CPUs).  I don't quite remember the exact timeline, I think it was about two days with 32 CPUs to finish all simulation runs for 4 different scenarios.

---
Alex

On Jul 1, 2013, at 10:00 PM, Saeid Montazeri <saeid.montazeri at gmail.com> wrote:

> Dear Alex,
> 
> Thanks a lot for the reply. May I have an estimate of the execution time of the Interest Flooding scenario (Please let me know the simulation time). Did you use MPI? If yes what was the number of processes?
> 
> Best Regards,
> Saeid
> 
> On Tue, Jul 2, 2013 at 1:20 AM, Alex Afanasyev <alexander.afanasyev at ucla.edu> wrote:
> Hi Saeid,
> 
> We have used Abilene topology, including link delays in https://github.com/cawka/ndnSIM-comcom-stateful-fw.  
> 
> Since Abilene topology is provided in several files, here is an example how to read it:
> 
>   RocketfuelWeightsReader rf_reader;
>   string weights   (topology+".weights");
> 
>   string latencies (topology+".latencies");
> 
> 
>   rf_reader.SetFileName (weights);
>   rf_reader.SetFileType (RocketfuelWeightsReader::WEIGHTS);
> 
>   rf_reader.Read ();
> 
>   rf_reader.SetFileName (latencies);
> 
>   rf_reader.SetFileType (RocketfuelWeightsReader::LATENCIES);
> 
>   rf_reader.Read ();
> 
> 
> As for your first question.  I would say that the choice really depends on what exactly you're simulating.  Abilene topology (if you're talking about the core topology) is really small and you don't have too much choice, even for random.  In Interest Flooding simulation (https://github.com/cawka/ndnSIM-ddos-interest-flooding) we have used larger topology from RocketFuel (unfortunately no real link delays or bandwidths, but we did some randomization for that), where we selected different tiers of nodes based on their topological parameters (leaf nodes are customers, next level are gateways, the rest is a backbone) and then installed consumers on random subset of leaf nodes.  Depending on where producer is installed (leaf, gw, backbone), you can get different results, as well as different real world scenario that it could represent.
> 
> Hope it helps.
> 
> ---
> Alex
> 
> On Jul 1, 2013, at 1:12 AM, Saeid Montazeri <saeid.montazeri at gmail.com> wrote:
> 
>> Dear All,
>> 
>> I am going to use abilene topology for my evaluation. I have a question:
>> How should I distribute the consumers and producers in the topology. Obviously, the simplest way is random. However, I am thinking about something more reasonable!
>> 
>> Meanwhile, it will be really helpful to have a reference for the link delay of the topology since the ndnSIM version does not have the link delays. 
>> 
>> Best Regards,
>> Saeid
>> _______________________________________________
>> ndnSIM mailing list
>> ndnSIM at lists.cs.ucla.edu
>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
> 
> 
> _______________________________________________
> 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/20130702/9c062c68/attachment.html>