Using DuckDB’s Native Parquet Engine for Zero‑Config Analytics
Learn how DuckDB lets you read, transform, and write Parquet files directly from SQL without extra services or complex setup.
Loading…
ReadMeFeed / Community knowledge
#parquet
Real questions. Useful conversations. Find the people who know your stack.
duckdbQuerying Parquet Files Directly in DuckDB with Predicate Pushdown and Column Pruning
Learn how DuckDB can query Parquet files directly, pushing down predicates and pruning columns for fast analytics without ETL.
Leveraging DuckDB's Automatic Column Pruning to Speed Up Wide Parquet Queries
Learn how DuckDB automatically prunes unneeded columns from Parquet files, see a concrete query example, verify the plan with EXPLAIN, and understand the limits of this optimization.
Zero‑Copy Parquet in DuckDB: Boost Data‑Science Performance without the Memory Cost
DuckDB’s zero‑copy Parquet reader lets you query terabyte‑scale datasets in Python with minimal memory, using column pruning and vectorised execution. This post walks through a practical example, shows how to verify the benefit, and discusses trade‑offs.
Querying Parquet Files Directly in DuckDB with Predicate and Projection Pushdown
Learn how DuckDB pushes WHERE predicates and column projections into Parquet scans, reducing I/O without any ETL step.
Choosing Between Parquet and Avro for Spark ETL Pipelines
A technical decision guide for choosing between Parquet and Avro in Apache Spark, comparing read/write performance, schema evolution, and columnar vs. row-based storage.
Zero‑Copy Analytics: Querying Remote Parquet Files Directly with DuckDB
Discover how DuckDB lets you query remote Parquet files directly, reducing data egress and speeding up analytics with zero‑copy, predicate pushdown, and an in‑process columnar engine.
DuckDB to Apache Parquet round-trip type fidelity for TIMESTAMPTZ and INTERVAL
Goal Determine whether DuckDB preserves TIMESTAMP WITH TIME ZONE and INTERVAL columns through a write-to-Parquet then read-from-Parquet cycle without changing the SQL type or losing values. Constraints DuckDB offers read_parquet() and COPY ... TO ... (FORMAT PARQUET), but INTERVAL and TIMESTAMP WITH TIME ZONE lack an exact Parquet primitive. They may be writ