At a glance
| SQL support | ANSI SQL across all connectors; write support varies by connector |
| Storage and compute architecture | Coordinator and worker nodes; no storage of its own, reads and writes through connectors |
| Workload types | Interactive and batch analytics on a lake, federated queries, lakehouse ETL |
| Integrations | Iceberg, Delta Lake, Hudi, Hive, PostgreSQL, MySQL, SQL Server, Oracle, Kafka and more |
| Deployment | Self-hosted, Starburst Galaxy or Enterprise, Amazon EMR |
| Licensing | Apache License 2.0; Starburst Galaxy priced in credits from USD 0.50 |
Trino suits data teams that want one SQL layer over a data lake and several operational databases, especially with Iceberg or Delta Lake tables in object storage. It is a query engine, not a storage platform, so you bring the storage, catalog and operations, or buy a managed distribution such as Starburst Galaxy.
Overview
Trino describes itself as a fast distributed SQL query engine for big data analytics. A coordinator plans queries and workers execute them in parallel, reading through connectors rather than from storage of its own. It began as PrestoSQL, the fork of Presto led by Presto's original creators, and was renamed Trino in December 2020. The project is supported by the non-profit Trino Software Foundation and licensed under the Apache License 2.0.
Trino releases often: the release notes list Trino 484 on 15 October 2026, following 483 in July. The Trino 484 documentation lists 37 connectors, including Iceberg, Delta Lake, Hudi, Hive and a Lakehouse connector, relational databases such as PostgreSQL, MySQL, SQL Server and Oracle, and systems such as Kafka, ClickHouse, DuckDB, Snowflake and SingleStore. Commercial distributions and managed services come from Starburst, and Amazon EMR also ships Trino.
Key features
- Federated SQL that joins data across catalogs, such as an Iceberg table and a PostgreSQL database, in one query
- Connectors for Apache Iceberg, Delta Lake, Apache Hudi and Hive, plus a Lakehouse connector
- Connectors for relational databases and other systems, 37 in the Trino 484 documentation
- Fault-tolerant execution with QUERY and TASK retry policies for long-running queries
- Separation of compute from storage: Trino stores no data and scales by adding workers
- ANSI SQL dialect with a single SQL layer across all connected sources
Pros and cons
Pros
- Apache 2.0 licence with a non-profit foundation behind the project
- One SQL layer over many sources, without copying data into a warehouse first
- Broad documented connector list, including the main open table formats
- Managed and self-managed commercial options from Starburst, plus Amazon EMR
Cons
- A query engine only: storage, catalogs and table maintenance are yours to provide
- Frequent releases mean regular upgrade and configuration work
- Feature support, such as writes, differs from connector to connector
Pricing
Trino is free and open source under the Apache License 2.0; running it yourself costs the compute and storage it uses plus the effort of operating a cluster. Starburst, a separate company founded by Trino's creators, sells Starburst Galaxy, billed in credits: a Free tier (up to three clusters), Pro from USD 0.50 per credit, Enterprise from USD 0.75 and Mission-Critical from USD 1.00 (the page shows dollar prices), with a 30-day trial including USD 500 of compute. Starburst Enterprise, the self-managed distribution, has no published prices.
Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.
Who it is best for
Data platform teams that keep data in object storage and several databases and want one distributed SQL engine to query them in place.
Who should look elsewhere
- Teams that want a managed warehouse with storage included and nothing to operate: see Snowflake alternatives.
- Heavy data engineering and machine learning in code, where a general processing engine may fit better: see Trino vs Spark.
Verdict
Trino is a mature choice for SQL over a lakehouse and across databases, with an open licence, frequent releases and wide connector coverage. Because it stores nothing itself, its value depends on the storage, catalog and operations around it, so self-hosting suits teams with platform skills, while others may prefer a managed distribution.
Frequently asked questions
Is Trino free?
Yes. Trino is open source under the Apache License 2.0. Starburst sells managed (Galaxy) and self-managed (Enterprise) distributions, and Galaxy has a free tier of up to three clusters.
Is Trino a database?
No. Trino is a distributed SQL query engine. It stores no data of its own and reads and writes through connectors to object storage, table formats and databases.
What is the latest Trino release?
The Trino release notes list Trino 484, dated 15 October 2026, when checked on 7 November 2026. Trino releases frequently, so check the release notes before upgrading.
Sources
- Trino release notes
- Trino connectors documentation
- Trino fault-tolerant execution documentation
- Trino GitHub repository and licence
- Starburst pricing
Checked November 2026.
How we research reviews and what star ratings mean: our editorial method.