In watsonx.data; when should you choose Spark over Presto for a workload?

Spark is preferred for batch ETL; iterative ML training; and complex multi-stage transformations; while Presto excels at interactive SQL analytics with low latency on data already in the lakehouse. The incorrect options ("When you need low-latency interactive SQL queries for a BI dashboard", "When you need to query external Db2 or PostgreSQL databases through data virtualization", "When you need the fastest response time for simple SELECT queries on small tables") are distractors that don't fully capture the concept described.