trino-dbt-query-performance

What problem does it solve?

This Skill helps you reduce slow dbt-generated queries on Trino by improving dbt materialization choices, incremental filtering, and Trino execution characteristics.

Core Features & Use Cases

• Materialization strategy optimization: Selects between view, ephemeral, table, incremental, and materialized_view based on cost and freshness needs.
• Incremental performance safeguards: Designs bounded, partition-aware is_incremental() filters to prevent accidental full scans.
• Efficient incremental write strategies on Iceberg: Chooses MERGE vs delete+insert and introduces patterns to avoid small-file proliferation.
• Trino cost controls for dbt: Recommends session properties, thread tuning, join hints (e.g., BROADCAST), and targeted post-run ANALYZE for better CBO planning.
• Safety against production regressions: Shows how to prevent risky full-refresh usage on large tables and calls out common anti-patterns.

Quick Start

Configure your dbt Trino profiles and rewrite your incremental models using bounded watermarks and the right incremental strategy to stop full scans and reduce runtime.