The startup's Postgres survival guide
Points and comments are a snapshot, not live.
A startup engineer's practical guide to moving beyond basic Postgres knowledge.
The guide, written by an engineer at the startup Hatchet, distills two years of production Postgres experience into actionable advice. It covers schema design (identity columns, timestamptz, foreign keys), query optimization (compound indexes aligned with ORDER BY, understanding seq scans vs. index scans), write performance (batching, short transactions, CREATE INDEX CONCURRENTLY), and connection management. Intermediate topics include query planner statistics, autovacuum tuning to prevent transaction ID wraparound, and handling bloat from dead tuples.
What commenters are saying
Commenters largely agreed on the guide's value but noted two major omissions: backups and monitoring. Several recommended pgBackRest over custom pg_dump scripts for point-in-time recovery. A thread split on managed databases: some argued RDS/cloud DBs are worth the cost for battle-tested HA and backups, while others countered that self-hosting with tools like pgBackRest offers more flexibility and lower cost, especially on providers like Hetzner. Others stressed the importance of proper schema design, recommending 'Database Design for Mere Mortals', and noted Postgres's utility for queues and caching.