Guided track
Performance and Planning
3 weeks3 modules12 steps
Mid-level engineers responsible for slow query diagnosis and tuning.
Start hereReading EXPLAIN: how the planner describes a queryFree
Module 1
Planner mental model
- LessonFreeReading EXPLAIN: how the planner describes a query
- ConceptRead EXPLAIN plans like a pro
- LessonProStatistics and selectivity estimation
- ConceptExtended statistics: correlated columns
You can predict why the planner picked a bad path and what evidence to gather.
Module 2
Executor and memory
- LessonProwork_mem and spills: where sorts and hashes go to disk
- ConceptTuplesort: the spill-to-disk boundary
- ConceptHash aggregate spill
- ReferenceThe GUC tuning reference
You can connect resource settings to observed execution behavior.
Module 3
Join strategies and parallelism
- ConceptJoin algorithms: nested loop, hash, merge
- LessonProHow the cost-based optimizer chooses a plan
- LessonProParallel query: splitting work across workers
- ConceptParallel query
You can explain why the planner picks a join and when parallelism helps or hurts.
What you will be able to do afterwards
- • Interpret planner choices instead of cargo-culting indexes.
- • Tune memory and executor settings with workload context.
- • Validate changes using before/after measurements that hold up in review.