Databricks Databricks-Certified-Data-Engineer-Associate Exam With Confidence Using Practice Dumps

Option D matches the heterogeneous sources and governance requirements. A relational database, object storage, and Kafka require source-appropriate ingestion configurations rather than a single connector attempting to handle every source. Database ingestion can use incremental or change-data-capture processing, Amazon S3 files can use file-oriented ingestion, and Kafka requires a streaming connector. Publishing the outputs into Unity Catalog schemas provides centralized permissions, discovery, and governance boundaries. Supported Databricks ingestion workloads also record lineage between source and destination objects in Unity Catalog. Delta table history and time travel provide access to earlier table versions when the required transaction-log and data-file retention settings are maintained. Disabling version history or repeatedly performing complete reloads would weaken auditability and increase processing overhead. Therefore, the separate, incrementally processed, Unity Catalog-governed design in option D is correct.

================

Option D retains only orders whose customer identifier matches a row in customers_df. Because customers_df contains only active customers, an inner join acts as a membership filter: orders belonging to active customers match and remain, while orders belonging to inactive customers have no corresponding row and are removed. A left join with orders_df on the left would preserve every order, including orders from inactive customers, unless an additional NULL filter were applied. A left join with customers_df on the left changes the driving dataset and can exclude the required order structure or produce different cardinality. A cross join generates the Cartesian product of both DataFrames and is unrelated to key-based filtering. Therefore, the inner join in option D precisely implements the stated active-customer filtering requirement.