DuckDB

DuckDB is an analytical database engine that can run embedded in an application process. It queries stored tables and supported external data such as Parquet, without requiring a separate server for a local analysis.

A file query can select columns, filter rows, and aggregate before returning a result. Inspect plans and measure actual work rather than assuming that every filter skips storage reads. Pin the input file set when repeatable results matter.

Supported operations can spill to temporary storage, but RAM, temporary disk, operator behavior, and final result size still limit a workload. A configured buffer-manager limit does not cap every allocation in the process.

Concurrent work within one process is supported; arrangements across processes require a supported access architecture. Remote storage still requires network access and credentials. A passing local SQL example does not validate another engine’s dialect or production recovery.

See Single-Node Analytics: DuckDB, Polars, and Arrow for worked examples.

Reference: Official documentation.