Trino (formerly PrestoSQL) is an open-source distributed SQL query engine originally forked from Facebook's internal Presto project by the engineers who designed the original system. While both Trino and PrestoDB trace their lineage to the same Facebook codebase, they have diverged significantly in governance, community focus, and ecosystem breadth. In 2026, Trino is considered one of the most production-ready engines for interactive, high-concurrency SQL analytics on Apache Iceberg data lakehouses.
Trino's Apache Iceberg Connector
By 2026, the Trino Iceberg connector is considered highly mature. It supports the full spectrum of Iceberg features required for production-grade lakehouse management:
- Full DML: Reliable support for
INSERT,UPDATE,DELETE, andMERGE INTOvia Copy-on-Write semantics, enabling Trino to serve as both an analytics and a data engineering engine. - Time Travel:
SELECT * FROM table FOR TIMESTAMP AS OFand snapshot-based time travel queries. - Schema and Partition Evolution: Including support for changing nested map and array types within Iceberg tables.
- Iceberg v3 (Experimental, 2026): Trino introduced experimental write support for Iceberg v3 tables, including new column default values and row lineage tracking capabilities that form the foundation of next-generation data auditing.
Catalog Integration
Trino's strength is its versatile catalog system. A single Trino deployment can simultaneously connect to Hive Metastore, Nessie, AWS Glue, Apache Polaris, and various other REST Catalog implementations. This makes Trino the natural "Swiss Army knife" for organizations managing complex, multi-source data estates where analysts need a single SQL interface across heterogeneous systems.
Trino vs. PrestoDB
The critical distinction in 2026 is community model and adoption velocity. Trino benefits from a broad, rapidly growing open-source community with faster adoption of new Iceberg features and a wider set of community-maintained connectors (including JDBC, Kafka, Delta Lake, and MongoDB). PrestoDB prioritizes extreme performance at Meta-scale and is investing heavily in its C++-based Native Engine (built on Velox). Organizations should evaluate both based on their scale, team expertise, and the breadth of data sources they need to query.
