Databricks Certified Data Engineer - Associate

Explanation

Delta Lake tables store data primarily in Parquet format. Here's why:

1. Delta Lake is built on top of Parquet: Delta Lake is a storage layer that sits on top of Parquet files, adding ACID transactions, schema enforcement, and time travel capabilities.

2. Parquet as the underlying format: All data in Delta Lake tables is stored in Parquet files, which are:

   • Columnar storage format
   • Highly efficient for analytics workloads
   • Compressed for storage optimization
   • Supports predicate pushdown for faster queries

3. Delta adds metadata: While the actual data is stored in Parquet files, Delta Lake adds:

   • A transaction log (in JSON format) to track changes
   • Metadata about the table structure and statistics
   • Versioning information for time travel

4. Why not the other options:

   • Delta (A): Delta is not a file format itself, but rather a storage layer/format that uses Parquet files.
   • CSV (B): CSV is a row-based text format, not the primary storage format for Delta Lake tables.

Key takeaway: Delta Lake enhances Parquet files with transactional capabilities, making Parquet the fundamental storage format for all data in Delta Lake tables.