Skip to content One engine between your data and your tools. It reads every source in place and exposes them through the same SQL.
Beacon documentation
Point a server at your archive. Connect a client. Write queries.
Your data, in place
Datasets storea local dir or one S3 bucket
NetCDF, Zarr, Parquet…read in place, never copied
SQL databasesPostgres, MySQL
Other Beacon serversremote catalogs
Beaconthe server
HTTP + Flight SQLAccess controlAdmin UICrawlersExports
Query engineinside it
Rust, on Apache Arrow and DataFusion. One SQL dialect, the format readers, and the managed tables in beacon.db.
One process. One SQL dialect. One catalog.
Query from anywhere
Python & notebookspandas, Polars, Arrow
SQL clientsDataGrip, DBeaver, JDBC
Dashboards & BIover HTTP or Flight SQL
Exports & pipelinesParquet, NetCDF, CSV, ODV
Choose a version
2.0.1StableRead this for a 2.0.1 server. This is the current release.Read the docs 1.8.0OlderRead this for a 1.8.0 server.Read the docs 1.7.3OlderRead this for a 1.7.3 server.Read the docs
Beacon 2.0.1 is the current release. A 1.8.0 server does not upgrade in place. Read Upgrade from 1.8.0 before you change the image tag.
Common tasks
Deploy a serverPoint Beacon at a directory or a bucket. Set the ports, the storage and the limits.Connect a clientUse Python, TypeScript or the terminal client. DataGrip and DBeaver use Arrow Flight SQL.Write SQLQuery your files with SELECT, JOIN and UNION BY NAME. Beacon adds 123 spatial functions.Call the APIOne endpoint accepts SQL or a JSON query. It answers in Arrow, Parquet, CSV, NetCDF or ODV.Read your formatsBeacon reads NetCDF, HDF5, Zarr, Icechunk, Atlas, Parquet, GeoParquet, CSV, ODV, Arrow, GeoTIFF, Iceberg, Delta and BBF.Tune and fixSet every BEACON_* variable. Find the answer for a common error.