Spark pipeline merge rules

baobo

What kinds of visual receipts can be merged together during the job executions?

Clément_Stenac

Hi,

All visual recipes except split can be merged, provided that they all use Spark engine.

For each recipe, there are specific conditions that can prevent it from merging. In that case, you'll find the reason for non-mergeability in the logs.

baobo

Thanks Clément.
I notice there are some logs for non-mergeablity. But I cannot figure of the reason for the non-mergeablity.

Does spark pipeline in dataiku as same as the pipeline concept in Spark ML ?
if not, what is the difference between them?

Furthermore, can you please explain "pre-merge prune"? How can it optimize the job?