Data Engineer Professional · 7% of the exam

Data Ingestion and Acquisition: free practice questions

5 sample questions from our 5-question bank for this domain — answers and explanations included. These are the same scenario-based style as the real Databricks exam.

1. New JSON files land continuously in a cloud storage location and must be ingested incrementally with automatic detection of new files and schema evolution. Which Databricks feature is designed for this?

  • A. Auto Loader (cloudFiles)✓ Correct
  • B. COPY INTO run once manually
  • C. A static batch read of the whole directory each hour
  • D. Delta Sharing
Explanation

Auto Loader (cloudFiles) incrementally and efficiently detects and ingests new files as they arrive, with schema inference/evolution and scalable file discovery. A one-time COPY INTO (B) isn't continuous, a full directory batch read (C) reprocesses everything, and Delta Sharing (D) shares data across orgs.

2. For a streaming ingestion pipeline, what MUST be configured so the stream can recover after a failure and continue exactly-once?

  • A. A checkpoint location for the streaming query✓ Correct
  • B. A larger driver node only
  • C. A ZORDER on the source
  • D. A Delta Sharing recipient
Explanation

A checkpoint location persists the stream's offsets and state, enabling fault-tolerant recovery and exactly-once semantics. A bigger driver (B), ZORDER (C), and Delta Sharing recipients (D) don't provide streaming recovery.

3. Auto Loader encounters files whose schema has an added field not seen before. With schema evolution enabled, what happens?

  • A. Auto Loader can evolve the schema to include the new column (and may restart the stream to pick it up), rather than failing✓ Correct
  • B. It silently drops all data
  • C. It deletes the target table
  • D. It ignores schema entirely and stores raw bytes
Explanation

With schema evolution, Auto Loader adds newly detected columns to the schema (often restarting the stream to apply the change) instead of failing on the mismatch. It doesn't drop data (B), delete the table (C), or store raw bytes ignoring schema (D).

4. A team needs an idempotent batch ingestion command that will not re-ingest files it has already loaded into a Delta table. Which command fits?

  • A. COPY INTO (which tracks loaded files and skips duplicates)✓ Correct
  • B. INSERT INTO run repeatedly with no tracking
  • C. A plain spark.read of the folder each time
  • D. MERGE with no condition
Explanation

COPY INTO is idempotent — it records which files it has loaded and skips them on re-runs, making retries safe. Repeated INSERT INTO (B) or a folder read (C) would duplicate data, and a conditionless MERGE (D) isn't an ingestion-dedup mechanism.

5. How does Auto Loader efficiently track which files it has already ingested at very large scale?

  • A. It maintains its ingestion state in a checkpoint (and can use file notification/directory listing), so already-processed files aren't reprocessed✓ Correct
  • B. It rereads every file in the directory each run
  • C. It relies on the file names being sequential integers
  • D. It stores state only in the driver's memory
Explanation

Auto Loader records processed files in its checkpoint (using scalable file discovery, including file notifications), so it ingests only new files and scales to millions of files. Rereading all files (B) doesn't scale, sequential names (C) aren't required, and driver memory (D) wouldn't survive restarts.

0 more questions in this domain

Practice the full bank with instant grading, flashcards, and a timed mock exam.

Start practicing free