[Nfd-dev] Throughput Measurement limited to 10 Mbps?

[Klaus Schneider]

Hey Aman,

can you post the complete summary output of ndncatchunks? What is the 
delay between your links? Lastly, have you tried a larger file and 
checked if it speeds up?

Best regards,
Klaus


On 11/18/18 2:46 PM, Aman Maldar wrote:
> Hello All,
> Short Question -  Is my experiment limited by the fact that interest 
> packets are generated at 1ms duration? How do I generate fast interests?
> 
> Long Question -
> I have a simple consumer and producer topology. I am trying to measure 
> throughput.
> 
> producer -  ndnputchunks /ndn/d-site/d/100MB.txt < files/100MB.txt
> consumer -   ndncatchunks -d iterative -v /ndn/d-site/d/100MB.txt &> 
> /home/100mb.txt
> 
> Everything works fine. Now I know there are 23832 chunks generated by 
> ndnputchunks.
> So, my producer send 23832 requests to fetch those chunks. I timed this 
> experiment.
> Time = 11108.2489491 ms
> Total size of interest sent + data received = 109.75 Mb
> 
> The observation is the throughput ( 109.75 / 11.10 ) is around 10Mbps. I 
> have tried this with ndnping and ndnpingserver application as well,
> I still get throughput of 10mbps.
> 
> I am running my experiments on 2 machines connected over 100G link. I 
> have tried the same on mini-ndn as well.
> Is this experiment limited by the fact that interest packets are 
> generated at 1ms duration by ndncat and ndnping?
> How do I generate fast interests?
> 
> I appreciate your time and help.
> -- 
> *Regards,**/
> /Mr. Maldar Aman M.***
> *University of Massachusetts, Lowell*
> 
> _______________________________________________
> Nfd-dev mailing list
> Nfd-dev at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/nfd-dev
>