Hi all. When it comes to separating resources by their lifecycle, should I be using different projects or different stacks? Based on the online docs it looks like separate projects would be the correct approach, but that just seems counterintuitive. As an example, if I had a project on AWS which involved an s3 bucket and network which should be kept alive, then an EC2 instance I'd like to be able to spin up/down as needed, how would I best organize that?

It depends 🙂 in your specific case, my initial thought would be to have those as two different projects as it sounds like they aren't related anyhow. You could keep all 3 resources in a single project and make the EC2 resource conditional upon e.g. a config key or similar. What's feeling counterintuitive to you?

Okay, that's what I was tracking based on the docs as well. I think it's the terms "projects" and "stacks". I was using the AWS CDK before trying Pulumi, and had those resources in separate stacks (a stateful and a compute stack), so I probably just need some time to get used to the new definition.

Yeah, for sure. Take your time fully understanding the concept of Pulumi stacks as they're really the foundation for... well, just about everything in Pulumi 😄

That is fair, the idea of stacks does make sense, and had I not come from CDK prior to this I'd probably not have the same holdups.

you can 'tag' stacks and then group them in Pulumi Cloud; that's really handy

I haven't looked at Pulumi Cloud just yet, but it is in my todo list

you can def still use tags, you'd not have the fancy view of them in a UI is all

That's a fair point, and in this case I don't have an answer, probably because it really doesn't make sense in the hypothetical 🙃 . I'll keep working and see what makes sense.

That's another "it depends" from me 😄 In my current experience, for AWS, Classic is fine. AWS Native is in preview but has resources available that Classic does not. It's perfectly fine to have and use both in the same program.

Makes sense, but if a resource is in both is there an advantage to using one over the other? Like the word "Classic" makes me think that one may eventually be deprecated, so would I be better off using Native where I can for better future proofing?

No worries! 😆 "Classic" does sound a lil scary, but I'd say don't worry re: future proofing. The one major difference - IMO - is that the documentation is currently "better" for Classic. As someone just starting out, you will probably benefit most from that. Classic: https://www.pulumi.com/registry/packages/aws/api-docs/ec2/instance/ Native: https://www.pulumi.com/registry/packages/aws-native/api-docs/ec2/instance/ You might also like https://www.pulumi.com/docs/clouds/aws/guides/ and the

awsx

package https://www.pulumi.com/registry/packages/awsx/ as it gives you some best-practice building blocks so you can move quick

Much appreciated, thank you. I would suggest (in my humble opinion) adding something to the docs of the providers to that effect, as the note in each just says to use the other if a resource isn't available, but nothing about what to use where they are both available.