work_mem and Spills: Where Sorts and Hashes Go to Disk
work_mem is how much memory one sort or hash step is allowed to use.
The catch is that a single query can run many such steps at once, and each gets its own budget, so the real memory used multiplies fast. When a step needs more than its budget, it spills the overflow to disk and slows down.
Pro members see the rest of this lesson
- A 4-step state diagram: Fits → Spill → Merge → Scale
- The full mechanism: what PostgreSQL does internally, in source terms
- 3 SQL queries you can run, each labelled by how it was verified
- 1 transcript captured on PostgreSQL 17.10
- The closing insight, the mistake it prevents, and what it changes in your work
Card required. Cancel before day 7 and you are not charged.
Check it against the source1 citation in postgres/postgres · file, symbol and line verified on REL_17_STABLE
Primary symbol: tuplesort_begin_common · line 645
Anchored to postgres/postgres on REL_17_STABLE and cross-checked against the manual for PostgreSQL 15–18. Every query here was run and its output captured on a throwaway PostgreSQL 17.10 lab; corrections are noted inline. §20.4.1 Memory (work_mem) in the official docs →
Connected
Where this lesson sits
What comes before, after, and alongside it.
Part of these pathways
Same learning track
Finished a free lesson?
Pro opens the rest of the engine course
You felt one mechanism. Pro is the full bodies, interview depth, and the tracks that build on this session.