Apache Arrow Flight

High-performance RPC based on Arrow + gRPC.

Definition

Arrow Flight gRPC service exposes 5 methods: GetStream, DoGet (download), DoPut (upload), DoExchange (bidirectional), ListFlights, GetFlightInfo. Apache Arrow IPC stream columnar format. Performance 10-100x superior to JDBC/ODBC. Flight SQL adds SQL semantics (statements, prepared statements, results). Flight RPC adopted by Dremio, InfluxDB IOx, ChromaDB, Snowflake, BigQuery, Apache Spark, DuckDB.

Origin

Apache Arrow project (Dremio + Wes McKinney, Apache Software Foundation top-level 2016), Flight introduced 2019, Flight SQL 2022.

Usage

Modern evolution of analytics data transfer. Progressively replacing JDBC/ODBC in new data tools.

Related terms

• Apache Arrow — fiche socle.
• Arrow Flight SQL — évolution SQL.
• gRPC — transport sous-jacent.