Honest comparison

Firedrill vs a DIY pg_dump cron job

A cron job running pg_dump into S3 costs almost nothing and teaches you how your database works. We're not going to pretend it's crazy — half of us ran exactly that for years. Here's what it doesn't give you, learned the hard way.

Where DIY pg_dump scripts is genuinely good

Side by side

DIY pg_dump scriptsFiredrill
Silent failureCron doesn't tell you it stopped; most DIY scripts fail silently for monthsEvery scheduled run heartbeats a dead-man switch — silence itself raises an alert
Restore testingAlmost nobody's script restores the dump and checks itEvery backup restored and checked: exit code, schema hash, tables, row counts, amcheck, sample queries
Version matchingpg_dump from apt may not match your server majorVersion-matched binaries 14–17, selected automatically
Encryption & key handlingUsually a bucket-level setting, if thatClient-side AES-256-GCM, per-project data keys, envelope encryption
RetentionA find -mtime +30 -delete if you rememberedGFS retention (7 daily / 4 weekly / 12 monthly), never prunes the last verified backup
Cost~$0 plus the incident where it turns out the backup was emptyFrom $19/mo

If you'll genuinely maintain the script, test restores quarterly, and monitor the monitor — DIY is defensible. If you'd rather that be someone's whole job, that's ours. Either way: restore-test whatever you have this week. Our free verification will do it for you once a month, no signup.