Trino
Trino is a distributed Structured Query Language (SQL) query engine (data analytics) designed for interactive analytic queries across heterogeneous data sources without data movement.
- Distributed Massively Parallel Processing (MPP) SQL query engine for large-scale analytics (data analytics)
- Query federation across diverse data sources through a pluggable connector model (data virtualization)
- American National Standards Institute (ANSI) SQL-compatible engine with support for complex analytical queries and joins (SQL processing)
- Cluster-based architecture with coordinators and workers for parallel query execution (distributed computing)
- Integration with data lakes, object storage, and traditional databases for unified querying (data lake analytics)
More About Trino
Trino is a distributed SQL query engine (data analytics) designed to run interactive analytic queries against data stored in a variety of systems, including data lakes, object stores, and relational databases. It focuses on querying data in place instead of requiring extract-load-transform pipelines, enabling query federation across multiple heterogeneous sources. Trino is generally classified as a query engine rather than a storage system, and it is used to provide a single SQL interface over disparate data platforms.
The project is built on a MPP architecture (distributed computing). A Trino cluster consists of a coordinator node and one or more worker nodes. The coordinator parses SQL statements, plans queries, manages metadata, and schedules tasks, while workers execute the plan fragments in parallel and exchange data over the network. This architecture allows Trino to process large datasets and complex joins by splitting work into smaller units and distributing them across the cluster.
Trino exposes a SQL interface (SQL processing) with broad support for ANSI SQL, including joins, aggregations, window functions, and subqueries. It is frequently used for interactive analytics and business intelligence workloads where users need to issue ad hoc queries and receive results with low latency relative to batch processing systems. Because Trino operates as a compute layer, organizations can connect BI tools, notebooks, or custom applications to Trino using standard JDBC and ODBC drivers (data access integration).
A central capability of Trino is its connector framework (data virtualization). Connectors allow Trino to query data from multiple backends such as distributed file systems, cloud object storage, relational databases, NoSQL systems, and other analytic platforms. Each connector translates Trino’s query plan into operations appropriate for the underlying system, including predicate and projection pushdown where possible. This enables a single query to join and aggregate data across different systems without copying data into a central warehouse.
In enterprise environments, Trino is deployed as part of data lake analytics and federated query architectures (data platform). It often runs alongside object storage-based data lakes using open table formats, and alongside traditional data warehouses and operational databases. Enterprises use Trino to serve analytical workloads for reporting, interactive dashboards, data exploration, and data science, by providing one query endpoint over diverse data assets. Trino integrates with enterprise security and governance via authentication, authorization, and role-based access controls exposed through its configuration and connector-level options (security and governance).
From a directory and taxonomy perspective, Trino fits into the categories of distributed SQL query engines, federated query engines, and data lake query layers. It is relevant for architectures that separate storage and compute, for hybrid and multi-cloud data access, and for organizations standardizing on SQL as a query language across various data systems. Its connector ecosystem and MPP execution model position it as a query federation and virtualization layer in modern data platforms.