trino-cost-optimization

What problem does it solve?

This Skill helps you control Trino warehouse spending by diagnosing what drives query cost, such as CPU time and physical data scanned, and by applying practical strategies to reduce those drivers.

Core Features & Use Cases

• Query scan cost analysis: Identifies top expensive queries and operators using Trino system runtime tables (e.g., CPU time and bytes read).
• Scan reduction and governance: Applies partition-filter requirements and physical scan limits to prevent costly full scans and runaway analyst queries.
• Storage and worker cost optimization: Uses compaction economics (fewer files/splits), storage retention checks, autoscaling patterns, and spot-instance strategies to lower compute and S3 request costs.
• Cost attribution and break-even: Estimates total cost using a simple model, supports chargeback reporting, and evaluates materialized view break-even to decide when precomputation pays off.
• Use Case: When a team’s analytics workload is driving rapidly rising infrastructure bills, use the Skill to attribute spend to teams/users, reduce S3 GET pressure via compaction, and enforce scan limits so expensive queries become rare exceptions rather than the default.

Quick Start

Ask your agent to analyze your last 24 hours of Trino queries by CPU time and bytes scanned, then propose the top three scan-reduction actions (partition pruning, compaction, and scan limits) for the highest-cost tables.