Episode 4. Indexing. Day 4 of your 12 day sprint.

Scenario: "This query got slow as the table grew. Fix it." Open with EXPLAIN ANALYZE, never with a guess.

Say: "First I read the plan shape and the buffers line. Then I decide whether the access path or the index is wrong."

Say the covering rule: "Index the filter you run. A covering index turns heap fetches into an index-only scan when visibility allows."

Say the cost rule: "Every index charges on write and bloats on update. An unused index is pure tax."

Lab: take one slow query, capture EXPLAIN ANALYZE before, add or fix one index, capture after. Quote both timings.

Trap to rule out: Say: "I ruled out the stale-statistics plan before touching the schema. ANALYZE first, DDL second."

Drill: list three index traps from memory: write cost, bloat, unused tax.
Say: a covering index earns its keep only when the filter runs hot.
Say the buffers line once more: heap fetches avoided is the number I quote.
Done-check: recite your before-and-after timings for tonight's query.
Depth tag tonight: Debug. You captured both plans yourself.

Teaser: tomorrow, Episode 5, the planner. Why good indexes still lose when statistics lie.