We haven't officially adopted Pulumi at my work yet. Due to the high price here, I was thinking of creating an S3 bucket to manage the state of our GitHub repos. That could be a good gateway for Pulumi at BENlabs in general, and maybe we'd come back and get a paid plan for other projects. For example, we use AWS CDK for all things AWS, however we use NewRelic, Auth0, and PagerDuty for all of our (many) REST APIs. I could certainly see us using Pulumi to manage those. And since it's easier to call that "production usage", I could see it being an easier sell to pay money for the state management of those things.

This does seem like a reasonable way to go, at least until Pulumi is adopted more broadly in your org. The S3 backend has a lot more features now than 3 or 4 years ago when I was using it, so it's much easier to recommend. And they've just released the "correct" org/project/stack naming for self-managed backends, so your projects are that little bit easier to migrate to the Pulumi service later. https://www.pulumi.com/blog/project-scoped-stacks-in-self-managed-backend/

You are amazing! Thanks for this resource! I hadn't been able to find much on self-managing the backend.

Thank Justin and Abhinav 🙂 I just relay information!

We create some initial infra using the Pulumi backend, but all big stuff goes into the s3 backend (as for example, there are hundreds of records in Route53). You can have different Pulumi projects with different backends. Set backend in

Pulumi.yaml

- https://www.pulumi.com/docs/intro/concepts/state/#logging-into-and-out-of-state-backends My dream is to be able to reference stacks that use a different backend too 🙂