DataEngBytes 2023 - BRI-T1-08 - Ryan Boyd
summary
Argues cloud warehouses act like modern mainframes and demonstrates a hybrid pattern using DuckDB locally (WASM, Python) with MotherDuck coordinating cloud copies, showing a laptop running SQL against parquet files that returns aggregates on 1.5 billion rows in about 1.5–3 seconds, converts remote gzip CSV to parquet in one SQL statement, and produces query plans that split execution between client and server while caching and filtering results in the browser.
AI-written summary · about
chapters (17)
- 0:00Hybrid SQL execution intro
- 1:05Mainframes to cloud warehouses
- 3:20Cloud data warehouse as mainframe
- 4:28DuckDB introduction
- 7:39DuckDB for analytics
- 9:26Why DuckDB is loved
- 10:09Simplified data access
- 14:08DuckDB in browser
- 15:11DuckDB architecture
- 18:02DuckDB disk and memory scaling
- 20:33DuckDB compression
- 21:36DuckDB large datasets
- 22:42DuckDB billion-row performance
- 23:45Open-source community and rapid innovation
- 25:31MotherDuck mission
- 26:36Hybrid querying explained
- 28:42MotherDuck features
related talks
27mDataEngBytes 2023 - MEL-T1-07 - Ryan Boyd
33mDataEngBytes 2023 - SYD-T1-01 - Ryan Boyd
40mDuckDB: Harnessing in-process analytics for data science and beyond
51mMotherduck: Data Analytics in the Post Big-Data Era
38mFrom Big Data to DuckDB: Ryan Boyd (MotherDuck) on Hyper-Tenancy, Duck Lake, and Making Data Fun.
28mAnalyze data easily with duckdb - and the implications on data architectures
01h 02mAnalytics for not-so-big data with DuckDB
47mGábor Szárnyas | Data Track | Compass AI & Data Summit 2025