Skip to content

Paper on Wrapper Engine

This issue is for drawing the lines of the paper that should come out from %Wrapper Engine Optimizations.

Roughly, I am thinking to have two main contributions

  1. Quantitative analysis of using all of these tools (#33 #23 #24 #39 (closed) #38 (closed))

I am thinking here just to measure the makespan of running tasks using the tools (kinda of following the validation part of the paper on Parsl). Hopefully, if I can, use multiple platforms (MN5, Fugaku, Lumi).

  1. Qualitative analysis of all the tools considered (#33 #23 #24 #39 (closed) #38 (closed) https://earth.bsc.es/gitlab/mgimenez/phd-project-management/-/issues/37 https://earth.bsc.es/gitlab/mgimenez/phd-project-management/-/issues/36 )

Talk about how it was to install them on the platforms, configure, and integrate on Autosubmit.

Also, above all, I find it insteresting that the motivation for wrappers can be find across different workflow managers. In Balsam and Radical Pilot they just allocate a chunk of resources and pump tasks to them and Nitro devs mention that small tasks would clutter the platform's scheduler.