If S2PX is generating Parallel jobs, why don’t those jobs run in Parallel mode? (i.e. Why are all stages configured to run sequentially?)
The design philosophy behind S2PX (driven by the goal of minimising the total manual conversion effort) means that it…
-
prioritises the delivery of a working Parallel job (if at all possible) above all else,
-
does not attempt to prematurely optimise your job designs, and
-
does not attempt to guess the original Job designer’s intentions
…but seeks to replicate, as far as technically possible, the job design unambiguously specified in each Server job it converts.
The fundamental reason S2PX-generated Parallel jobs run in Sequential mode is that Server job designs don’t provide all the context required to identify the keys necessary to support hashed partitioning. Theoretically, S2PX could have attempted to interpret existing Job designs and make a guess at your partitioning keys, but this will always involve sufficient ambiguity to risk S2PX generating a non-functioning job, contrary to the above design priorities.