We are currently evaluating a move away from the Databricks warehouse-based destination to the ADLS / Unity Catalog destination. Cost benefits from reduced warehouse DBUs need to be offset against increased costs in table processing, in some cases. Specifically, in the case of a Delta Live Table or Spark Structured Streaming pipeline using a Fivetran table as a source, it needs either an append-only source or CDF enabled when there are update/delete commits (i.e. merge commits) in order to leverage the incremental processing benefits available.

If you could enable CDF for the ADLS destination, the cost benefit of moving to ADLS would be more clear: losing the warehouse DBUs would compound with DBU savings from incremental table processing.