[Battlemesh] No results again

[leonardo]

On Mon, 2016-05-09 at 18:02 +0200, nemesis wrote:


>  We either let both participate and try different scenarios, hence 
>  getting even more results, because we can do more tests if we work
> well, 
>  or we should clarify that the tests need to be performed with a
> single 
>  system, like wibed, therefore those who don't find it comfortable to
>  work with it will have to decide if coming just for the talks or not
>  come at all.

Let me add my 0.2$,

I think we need some diversity, regardless of the approach used for the
testbed. 

When I tried to help with plotting data on Wednesday, I realized that
there was still no data do plot, and that there were already at least
three/four people trying to make Wibed work and I did not have a clue
on how to help them. In my experience, if it takes X days to set-up a
testbed, it takes at least X more days to get some meaningful results.
Results must be repeated, collected, plotted, interpreted. And
everytime you find something unreasonable in your results, you have to
step back and understand why (and in this, contrarily to Juliusz I see
the role of an automated tool). So, I think it is reasonable to have
people setting up the scenarios, preparing scripts that generate
traffic, decide where they run, do some small scale tests, collect and
plot data etc... all this can be done in parallel on a few nodes
configured by hand, and as soon as the testbed is ready it can be
deployed on the full-scale testbed, whether it is Wibed or not. 

leonardo.

> _______________________________________________
> Battlemesh mailing list
> Battlemesh at ml.ninux.org
> http://ml.ninux.org/mailman/listinfo/battlemesh