There are a number of techniques you can use to improve the performance of the parallel jobs generatd by S2PX:
Use partition parallelism
There are good reasons why S2PX must generate single Parallel jobs which run in sequential mode. The foremost reason is that S2PX can’t guess the job’s partitioning keys, and the implication of selecting incorrect keys is the genreation of a non-functioning job. The good news is that it’s
Identify keys and set appropriate stages to run in Parallel mode, with the goal of running all of a job’s stages in Parallel mode.
Remove redundant stages
S2PX necessarily introduces extra stages into your jobs to make them work with a Parallel execution engine. Where a particularly time-sensitive job justifies the investment you can use humans to make design decisions which .
Eliminate redundant function calls
Optimising DataStage expressions (such as those in Transformer derivations) typecasting