I am interested in using parallel processing in a workspace and I understand that the Group By parameter must be uswed. If using a transformer that has two input feature classes such a Clipper or AreaOnAreaOverlayer is it required that the field used, in this case Value, in the Group By parameter be in the attribute table of each feature class? So in order to use parallel processing with the Clipper transformer both the clipper and the clippee would need to have the field Value along with the corresponding attributes to work correctly. Is that correct?
Page 1 / 1
Yeah you need to sync the data using grouping strategy.
Wich is why tiling strategies are strong in that respect.
Workspace efficency concerning set theory, like in Oracle will aslo help quite a lot.
Controlling the processingstreams in the workbench is also a factor to consider.
(see the fme 2015 live counters...) Like blocking transformers must wait for their input etc.