PGConf.Russia 2019 Tatsuro Yamada «Auto plan tuning using feedback loop»
PGConf.Russia 2019 Tatsuro Yamada «Auto plan tuning using feedback loop»
As is often seen in OLAP and batch processing workloads, the more complex a query (containing many joins, filters, aggregates), the more there is a possibility of row count estimation errors, which leads to planner choosing an inefficient execution plan.
To address that problem, I developed a tool called pg_plan_advsr as a PostgreSQL extension, which corrects the estimation errors by repeatedly feeding back the information collected during query execution to the planner.
The tool has three features:
• Automatic plan tuning by repeatedly feeding execution information to planner
• Preserve all plans generated during plan tuning in a history table
• Create and store optimizer hints to be able to reproduce plans generated during tuning process
I verified the effectiveness of pg_plan_advsr by enabling it when running the join order benchmark (JOB) against PG 10.4 and observed its execution time shortening to 50% of the original. Therefore, it is useful for user who would like to do plan tuning for OLAP and batch processing.
I will talk about the following things in this presentation:
• Principles behind pg_plan_advsr and its architecture
• Detailed information about the measurements done with JOB
• Possible future enhancements
• Using aqo and pg_plan_advsr together (experimental)
More details: https://pgconf.ru/en/2019/242844
#PGConfRu #PGConfRussia #PostgreSQL #Postgres