The cost quickly adds up for small underfunded projects though. You also need to factor in how the application does scale. What kind of ingress/load balancing is required. What kind of stateful storage is required. Network policy. Resource monitoring. Config management, CI/CD pipeline. I setup a basic cluster on gke to start but haven’t gotten around to building these just yet. I’ve got a goal of attempting the most performant, scalable, and cheapest instance out there.

That’s the easy part. Next part is identifying which app endpoints are the most intensive and splitting them off into separate deployments.

The hard part is scaling the database, as I’ve only seen one tool that can autoshard effectively, but it’s for MySQL, not Postgres.