A database extension is an installable package that adds capabilities outside a database’s core. In DuckDB, extensions support workflows such as querying remote files, connecting to cloud databases, and extending query behavior. section 2.4 section 7.1.1
Manage an extension through a short lifecycle:
- Inspect available, installed, and loaded extensions with duckdb_extensions(). section 2.4
- Install a package with INSTALL name; installation stores it in the database but does not necessarily load it. section 2.4
- Load it with LOAD name when needed; both commands are idempotent, and DuckDB can autoload installed extensions when it recognizes a required capability. section 2.4
Query a remote CSV with httpfs
DuckDB cannot query internet-hosted files by default. Install and load the official httpfs extension, then query the remote file directly: INSTALL httpfs; LOAD httpfs; SELECT count(*) FROM read_csv_auto('https://github.com/duckdb/duckdb/raw/main/data/csv/ips.csv.gz'); The extension enables HTTP(S), S3, and selected cloud-storage access without first downloading the file locally. section 2.4
The MotherDuck extension adds cloud database management, sharing, S3 access, and hybrid execution. Its MD_RUN parameter controls whether read functions execute locally, remotely, or automatically route remote URLs to the cloud; choosing the location can avoid unnecessary transfer and compute costs. section 7.3 section 7.3.5
Listing 2.4 The format of duckdb_extensions output
DESCRIBE SELECT * FROM duckdb_extensions();
Inspect the schema returned by duckdb_extensions() before selecting fields for extension inventory queries. section 2.4
