DuckDB: SQL on logs without a server

DuckDB: SQL on logs without a server

The new favorite for ad-hoc log work — query CSV/Parquet logs in-process. Hand-picked references.

⭐ Pick of the week: the DuckDB docs on read_csv_auto — point it straight at a log file and run SQL, no import step.
Takeaway: SELECT status, count(*) FROM read_csv_auto('access.log') GROUP BY status;

→ DuckDB's regex functions — parse Combined format into columns inline with regexp_extract.
Takeaway: no pre-processing pipeline needed.

→ MotherDuck's blog posts on log analytics — practical patterns for crawl-frequency rollups.

→ The Parquet conversion trick — convert once, query a month of logs in milliseconds.
Takeaway: gzip-to-Parquet beats re-grepping daily.

Credit to the DuckDB team and MotherDuck's engineering blog.