The Planner's Guessing Game Inside PostgreSQL's ANALYZE

The Planner's Guessing Game: Inside PostgreSQL's ANALYZE is a deep, source-code-level guide to how PostgreSQL's query planner estimates row counts before a single row is read. Every plan the planner picks — index scan or sequential scan, nested loop or hash join — rests on a guess, and this book explains exactly where that guess comes from: the sampling math behind ANALYZE, the MCV lists and histograms stored in pg_statistic, the estimators that turn a WHERE clause into a number, and the autovacuum machinery that keeps it all fresh. Written and verified against PostgreSQL 17, every claim in the book traces back to one of four sources — the PostgreSQL source code itself, the official documentation, the research papers Postgres's own comments cite by name, or the pgsql-hackers mailing list archives — with direct links so readers can verify everything themselves rather than take the book's word for it. Across 12 chapters and a glossary, readers build a working mental model of the statistics system from the ground up, with runnable SQL examples throughout and honest boundaries around what the book deliberately leaves out. It's for engineers, DBAs, and architects who want to stop guessing why the planner chose the plan it chose.