A hierarchical execution model organizes a distributed query into nested units of work. In Trino, queries contain stages, stages contain tasks, and tasks process splits that can run concurrently across worker nodes. section 8.3.7
The hierarchy coordinates execution across distributed and federated environments:
- Queries represent complete analytical requests and are decomposed into stages. section 8.3.7
- Stages divide processing into tasks that can execute independently across cluster machines. section 8.3.7
- Tasks operate on splits, enabling parallel reads, aggregations, filter pushdown, and reduced data movement. section 8.3.7
The model suits federation because it coordinates work across object storage, relational databases, and other sources while supporting concurrency, workload isolation, and scalable coordinator–worker deployments. section 8.3.7 section 8.7
Figure 8.10 Trino clusters are made up of a coordinator node that distributes work among worker nodes. The worker nodes can be horizontally scaled by adding more workers to handle more queries.
Use the cluster diagram to connect coordinator-distributed work with horizontally scalable worker capacity. section 8.3.7
