Temp block writes
Query audit
Postgres temp writes check for spill-heavy queries
Database Optimization Tool surfaces PostgreSQL temp write pressure from evidence-focused statistics so teams can identify queries likely spilling to disk before changing work_mem or indexes.
Spill-heavy query ranking
Sort and hash risk signals
EXPLAIN (ANALYZE, BUFFERS) next step
What temp writes usually mean
Temp writes often appear when sorts, hashes, materialization, or large intermediate results spill beyond memory. The audit ranks candidates by observed temp-block pressure and query cost.
Why workload context matters
A batch report that spills once a day is different from an endpoint that spills thousands of times. Database Optimization Tool compares temp writes with total time, calls, rows, and read pressure.
Safer diagnosis before tuning
The report points to EXPLAIN (ANALYZE, BUFFERS), query shape review, index review, and controlled work_mem testing instead of recommending broad production memory changes.
evidence-focused boundary
Database Optimization Tool does not monitor in real time, change work_mem, or rewrite queries automatically. It provides evidence-gated findings for engineers to verify.
FAQ
Frequently asked questions
These answers describe the product focus: careful database evidence, clear findings, and team-approved next steps.
Does every temp write indicate a problem?
No. Some analytical or maintenance queries can write temp data acceptably. The risk rises when temp writes are frequent, user-facing, or tied to high total runtime.
Can Database Optimization Tool change work_mem automatically?
No. It only reports evidence. Memory tuning needs workload review because a global increase can raise concurrency risk and make incidents worse.
What evidence should be checked after a temp write finding?
Use EXPLAIN (ANALYZE, BUFFERS), inspect sort and hash nodes, compare rows estimated versus rows returned, and review whether an index or query rewrite reduces the spill.
Do temp writes always come from missing indexes?
No. Missing indexes can contribute, but temp writes can also come from joins, aggregation, sorting, CTE shape, bad estimates, or large result sets.