When a mapping runs on the Spark engine, the Spark engine translates unconnected Lookup transformations to join operations each time that you reference the transformation using a lookup expression. When there are more than five unconnected lookup expressions in a single mapping, the performance might degrade considerably depending on the size of the lookup source.

If the mapping logic requires multiple unconnected lookup expressions, consider running the mapping on the Blaze engine. The Blaze engine is optimized to generate the lookup cache one time and reuse the cache each time that an unconnected lookup expression references the same lookup source.