[ohstopitu]

maybe I'm doing this wrong, but if I use RDS / Cloud SQL / Compose etc - a normal base cluster + another failover cluster + hourly/daily backups (based on your release schedule) all add up to around 30 - 50 based on various cloud providers.

[StavrosK]

Hmm, can't you just install a database on one of the servers and later migrate to one of the managed datastores if you need it? That migration should be pretty simple with minimal downtime.

[Kpourdeilami]

You can start with one server with postgres installed on it and when you need to scale to multiple application servers, you can run postgres on another server.

For backing up postgres, all you have to do is setup a cron job that backs up the postgres' data directory to S3/Google Drive/Dropbox every hour/day.

If you want proper replication and failover then you can probably use 2 digital ocean droplets each for $5/month and another $5 VM for the application server itself.