SQL Performance Playbook for Real Systems

Performance work starts with measurement, not assumptions.

What I check first

1. Query plan shape and index usage.
2. Cardinality mismatch in joins.
3. Repeated heavy reads with no caching strategy.
4. Overly wide result sets.

Practical example

EXPLAIN ANALYZE
SELECT id, status, created_at
FROM orders
WHERE account_id = $1
  AND created_at >= NOW() - INTERVAL '30 days'
ORDER BY created_at DESC
LIMIT 100;

When the plan is stable and the index strategy is aligned with access patterns, latency drops quickly.