Databricks Certified Data Engineer - Associate

Explanation

Option D is correct because it performs a streaming hop from raw data to a Bronze table. Here's why:

1. Streaming Source: Uses spark.readStream.load(rawSalesLocation) which reads from a streaming source (raw data location).
2. Streaming Sink: Uses .writeStream to write as a streaming job.
3. Checkpointing: Includes .option("checkpointLocation", checkpointPath) for fault tolerance.
4. Append Mode: Uses .outputMode("append") which is appropriate for streaming incremental data.
5. Bronze Table: Writes to a table (.table("newSales")) which represents the Bronze layer in the medallion architecture.

Why other options are incorrect:

• Option A: Reads from an existing table (spark.table("sales")) not raw data, and uses complete mode which rewrites the entire table each time.
• Option B: Uses batch processing (spark.read.load()) not streaming, and writes with .write not .writeStream.
• Option C: Similar to A, reads from an existing table not raw data, and performs transformations before writing.

In Databricks medallion architecture, a streaming hop from raw data to Bronze typically involves reading streaming data from raw sources and writing it to Bronze tables with minimal transformations, exactly what Option D demonstrates.