because its not a one time operation. also, how else do you sensibly deploy and manage a multi-stage programme on >500 nodes? I mean we use AWS batch, which i

because its not a one time operation. also, how else do you sensibly deploy and manage a multi-stage programme on >500 nodes? I mean we use AWS batch, which is far superior for this sort of thing. SLURM might work for real steel, as would tractor from pixar.