Can't I batch process in a single workspace by merging in source data and fanning it out on the destination side?

From fmepedia

Q) Can't I batch process in a single workspace by merging in source data and fanning it out on the destination side?

A) Yes, you can. However, be aware that a Dataset Fanout carries huge costs related to temporarily writing data to disk. I assume this is because there is no guarantee that features arrive at the fanout in a single dataset group. Therefore FME has to write all of the datasets to temporary storage then fan them out afterwards.

Feature Type Fanout has no similar issues, so you may get away with it if your destination format is directory based.