Date: 2025-09-30
Time: 14:00–14:50
Room: Room 3
Level: Intermediate
PostgreSQL performance measurement is complex, with metrics like average query runtime often hiding significant variability, particularly when analyzing short time windows or different times of the day. This talk explores the limitations of traditional performance indicators and introduces a novel workload fingerprint method designed for more reliable performance assessment, particularly in live production environments. This fingerprint approach provides a granular view of database activity, aiding both human database administrators and AI-powered tuning systems in understanding and effectively addressing performance tuning. Attendees will gain insights into effective tuning strategies and monitoring techniques, engaging in discussions that span both the technical implementation and philosophical considerations of database optimization, ensuring a comprehensive understanding of the challenges and solutions in maximizing PostgreSQL efficiency.