Today, I gave yet another presentation on SimGrid, but this one was a bit different. I did not wanted to sell the framework to the attendee (since they are the co-workers of Arnaud, I assume that they already bought it). Instead I wanted to exhibit the coherence in my research work about this framework. I thus first spoke of its diversity. I contrasted our OS-inspired design for parallel simulation to our semantic decomposition of the simulated applications, allowing to automatically model-check them. Then, I went I took some altitude, exploring how we could come up with a coherent work bench to evaluate distributed applications using all available methodologies, if only we were more numerous working on these questions. The presentation was very well received, and I really hope that this research agenda will gain even more momentum soon.

Download PDF