ODBC takes an Arrow to the knee: ADBC
summary
ADBC standardizes a C API that transports zero-copy Arrow payloads so columnar OLAP systems avoid row↔column transposes; drivers exist for Snowflake, Flight SQL, Postgres, DuckDB and language bindings in Python Go Java Rust. Benchmarks show dramatic gains — DuckDB: 28s → 7s for a bulk select when Arrow is passed natively — and Python usage is as simple as pip installing adbc drivers and calling a fetch_arrow_table or using the driver manager to load shared-object drivers.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 4:21ODBC/JDBC analytics limits
- 8:16What is ADBC
- 12:55When ADBC outperforms ODBC and JDBC
- 14:43DuckDB ADBC benchmark
- 16:53ADBC Python API
- 21:08Simpler database drivers
- 23:12ADBC bulk ingestion
- 25:05ADBC learning resources
- 28:44Q&A
related talks
01h 04mWhere We’re Going, We Don’t Need Rows: Columnar Data Connectivity with Apache Arrow ADBC
41mFaster SQL with pandas and Apache Arrow
33mArrow in Flight - New Developments in Data Connectivity
35mPractical Applications of Apache Arrow
42mState of the Apache Arrow Ecosystem:How your project can leverage Arrow!
33mturbodbc: Turbocharged Database Access for Data Scientists
32mThe expanding Apache Arrow universe
30mDuckDB: Bringing analytical SQL directly to your Python shell