Every backend developer has encountered a query that should be fast and is not. Most of the time the fix is obvious once you understand what the planner is doing - which is exactly what this workshop teaches.

Topics covered

• Storage internals: heap files, TOAST, and what VACUUM actually does
• Index types: B-tree, GiST, GIN, BRIN - when each is the right choice and when they make things worse
• Reading EXPLAIN ANALYZE output: what seq scan, index scan, hash join, and bitmap heap scan tell you
• Statistics and the planner: pg_statistic, ANALYZE, and why the planner makes surprising choices on small tables
• Locking and MVCC: row-level locks, deadlock patterns, and the cost of SELECT FOR UPDATE
• Connection pooling: PgBouncer, Supavisor, and the max_connections trap
• Hands-on lab: bring your slowest query; we diagnose it live

Prerequisites

You write SQL regularly. You have used EXPLAIN before but are not sure you understand the output. Bring a real slow query if you have one.