Moving data out of Snowflake into PostgreSQL looks easy until the types come out wrong!
That's the kind of silent change I wanted Hydra ETL's new Snowflake source to rule out. Tested on a real Snowflake account, 50,000 orders into PostgreSQL:
→ 38-digit IDs and decimal amounts stay exact
→ DATE stays a date, TIMESTAMP_TZ keeps its time zone
→ VARIANT lands as JSONB
→ Snowflake's uppercase column names can be lowercased for PostgreSQL
Then I broke it on purpose: an ID too large for PostgreSQL's bigint, in the middle of 50,000 updates. The run stopped, and PostgreSQL still held exactly the previous data, to the cent. Nothing half-loaded.
Coming in Hydra ETL 0.20.
The example, with a local Snowflake emulator if you don't have an account, is in the first comment.
Example: https://github.com/bejaouibechir/Hydra/tree/main/examples/snowflake-to-postgres
Top comments (0)