Trino

Trino is a fast distributed SQL query engine for big data analytics, published by the Trino Software Foundation with connectors across many data sources, a developer guide, an open contribution model and an active community.

The defining capability is querying data where it lives rather than requiring it to be loaded first, and joining across sources in a single query.

That addresses a problem every organisation of moderate size has. Data is spread across a data lake, several relational databases, a warehouse and various operational systems, and the questions worth answering cross those boundaries. Which customers in the CRM have open support tickets and declining usage is a single business question that touches three systems.

The conventional answer is to build pipelines copying everything into one warehouse first. That works, and it costs engineering time to build, ongoing effort to maintain, and introduces a delay between something happening and it being queryable. It also means the answer to a question nobody anticipated requires building a new pipeline before it can be asked.

Trino queries the sources directly and joins across them, which makes exploratory work possible without a project. The honest limitation is that federated queries pull data across network boundaries and can put load on production databases, so it complements a warehouse rather than replacing one, and querying an operational system during business hours needs care.

Speed is the other emphasis, with the project describing itself as running at ludicrous speed. Trino was built for interactive analytics where an analyst waits for the result, which is a different design target from batch engines optimised for throughput on jobs nobody watches.

Being governed by an independent foundation rather than a single vendor matters for a query engine an organisation builds analytics on, since it reduces the risk of licence changes.

Trino is free and open source, with commercial distributions and managed services offered by third parties.