Trino5 min read
Trino Federation: Query the Lake Without Melting Prod Postgres
Trino stores nothing. A federated SELECT * on prod Postgres is a database incident. Use Iceberg for the lake, and keep interactive SQL off OLTP.
- trino
- sql
All stacks · Warehouses & analytics
Trino
Federated SQL across Iceberg, Hive, and warehouses without moving data.
TrinoAbout Trino
Trino (formerly PrestoSQL) is the federated SQL engine: one query across Iceberg, Hive, Postgres, and a warehouse without first copying everything. Starburst is the common commercial distribution.
It is a query engine, not a warehouse. There is no durable storage of your own. Cost is cluster hours plus the pain you inflict on the sources you federate.
What you'll learn here
- Coordinator vs workers and why a single fat coordinator dies
- Catalogs, connectors, and Iceberg as the default lake connector
- Why SELECT * and SELECT DISTINCT are cluster killers
- When to federate vs when to land a table in the warehouse
Frequently asked questions
Trino or Spark SQL?
Trino for interactive, federated, ad-hoc SQL. Spark for heavy ETL, streaming, and ML. Many lakes expose the same Iceberg tables to both.
Is Trino a warehouse replacement?
No. It does not store your gold tables. It queries them. Governance, time travel, and cost attribution still live in the table format and the catalog — or in Snowflake/BigQuery if that is the store.
Why did we melt the OLTP database?
A federated connector with no predicate pushdown and a dashboard that scans the app database. Never point Trino at prod Postgres without a replica and a row limit culture.
New Trino posts, straight to your inbox
One email a week with our latest tutorials. No spam.
Newsletter signup is not live yet. Use the contact form if you want to be notified.