DP-750 · Deploy and maintain data pipelines and workloads

A data engineer is building a Databricks job with three notebook tasks: 'ingest_raw' loads files into a bronze Delta table, 'clean_data' deduplicates and validates the bronze table into a silver table, and 'build_aggregates' computes daily summaries from the silver table into a gold table. The engineer wants the job to guarantee that 'clean_data' never starts before 'ingest_raw' finishes, and 'build_aggregates' never starts before 'clean_data' finishes, even if a run is retried or a task takes longer than usual. How should the engineer design the order of operations in the Databricks Jobs UI?

• ASchedule the three tasks as separate jobs with staggered cron start times spaced far enough apart that earlier tasks are likely to finish first.
• BPut all three notebooks in one task and call each notebook sequentially with dbutils.notebook.run inside a single notebook task.
• CAssign all three tasks to the same job cluster and rely on cluster autoscaling to process them in the order they were added to the job.
• DAdd each task to the job and set the 'Depends on' field for 'clean_data' to 'ingest_raw', and for 'build_aggregates' to 'clean_data', so the job scheduler builds a task DAG.