replicate the results in paper

[yfle0707]

Hi,

We want to replicate the results of workload W5 in Figure 12 in paper (https://arxiv.org/abs/1803.09615) to sanity check if we use Homa simulator properly. As the figures also shows the results of comparison with NDP, we noticed that the modification of NDP code is not in the open sourced code. Could you please also release the modified code and configurations that you use for NDP simulator and help us replicate the results? Thanks a lot!

[redtagsolution]

Hi @yfle0707 , were you able to replicate rest of the results? I am running that python file i.e. PlotDigestor.py but that one is throwing some errors. Any assistance can be of great help.

Thanks

[redtagsolution]

Hi @behnamm ,

Can you please share the updated PlotDigester.py file. And can you also share the latex file for your paper, that would be super helpful.

Thanks
Salman

[yfle0707]

Hi @redtagsolution,

I fixed the PlotDigister.py. However, you still can use the following command to process the results.
python analysis/MetricsDashBoard.py homatransport/src/dcntopo/results/manyReceivers/comparison/linkCheckBytes__-1/WorkloadDCTCP-0.sca

Then, you can plot the figures by yourself by looking at the stats below this header:

                             ----------------------------------------------------------------
                            |End To End Message Stretch For Different Ranges of Message Sizes|
                             ----------------------------------------------------------------

[redtagsolution]

Hi @yfle0707 ,

Yes I fixed that too, but for some reason I am unable to convert and get the results for hostids > 90. But for 90 hosts I plotted the results and it was fine.

Thanks for the help though. Really appreciate it.

[yfle0707]

Hi @redtagsolution,

Thanks for your reply.
I want to understand how omnet++ is working.
Do you know how the switch, hosts are connected and where the switch code is located? Thanks!

[kvm2116]

@redtagsolution @yfle0707 I am trying to create the figures as shown in the paper for HOMA and pFabric. Can you share the files and/or commands you used?