The next wave of data infrastructure is in the Lake
See how a dbt project can move ingestion and individual models onto a data lake, engine by engine, without a rip-and-replace migration.
Overview
Addresses limits of the modern data stack for AI-agent operations and vendor lock-in. Introduces Open Data Infrastructure (ODI): open formats like Iceberg in controlled object storage for direct ownership, engine choice per task, and one unified source for all teams and agents, built with Fivetran and dbt.
What you'll take away
- 01
Apache Iceberg as an open table format, exposed through an Iceberg REST catalog, lets multiple engines read and write the same data without copying it, avoiding the "death march" migrations of the past.
- 02
Fivetran's new "Move" capability migrated a Snowflake-native table to Iceberg in place in a couple of minutes, no changes required to dependent dbt models or dashboards, and cut 15-20% off the Snowflake compute bill for that workload.
- 03
Lake Compute lets specific low-complexity dbt models run on cheaper lake-native compute model by model, targeting the roughly 75% of a typical stack's spend that goes to ingestion and transformation.
Session speakers
Russell Christopher
dbt LabsCasey Karst
Fivetran