Trino Query Optimizer

概要

Provides expert guidance for tuning Trino SQL queries specifically within Treasure Data, focusing on critical performance techniques like time-based partition pruning, columnar selection, and output optimization. It helps developers overcome common hurdles such as memory limits, timeouts, and slow aggregations by recommending approximate functions, efficient join strategies, and User-Defined Partitioning (UDP). This skill acts as a performance consultant, offering actionable advice to transform inefficient full-table scans into high-performance, cost-effective data workflows.

主な機能

• Memory management via approximate functions (HyperLogLog) and partitioned window functions.
• Implementation guidance for time-based partition pruning using td_interval and td_time_range.
• Step-by-step troubleshooting workflows for PAGE_TRANSPORT_TIMEOUT and memory limit errors.
• Advanced join optimization techniques including Magic Comments and colocated joins.
• 2 GitHub stars
• Optimization strategies for column selection and high-speed output methods like CTAS.

ユースケース

• Resolving slow-running queries or frequent timeouts in production data pipelines.
• Debugging memory-intensive operations like large-scale joins and exact distinct counts.
• Reducing query execution costs by optimizing data scanning and resource utilization.