Introducing a SQL-based Metrics Layer Powered by DuckDB

Metadata

• Author: DuckDB
• Full Title:: Introducing a SQL-based Metrics Layer Powered by DuckDB
• Category:: 🗞️Articles
• Document Tags:: Metrics, Metrics layer,
• URL:: https://www.youtube.com/watch?si=IZP6u6ERja1CpMcN&v=_IqvrFWY7ZM&feature=youtu.be
• Read date:: 2025-02-27

Highlights

metrics really are the right size building blocks for analytics. Fact tables are too raw. If, you know, I try to look at that Parquet file, I’m not sure I would get very far in terms of insight. But on the other hand, most reports are too baked. If I just gave you, you know, a PDF of a bunch of graphs, and frankly, we all look at GitHub analytics a lot. You don’t get a lot of insight when you just get a baked report, and so metrics are great because they’re flexible. A Metric is an aggregate function. This is quoting Julian Hyde, who’s been doing some work on metrics and SQL. But metrics are aggregate functions that can be evaluated in different contexts (View Highlight)