A small SaaS can end up with Kubernetes, multiple environments, elaborate CI/CD pipelines, observability tools, and a surprisingly complicated cloud setup.

None of these are necessarily bad decisions.

But sometimes the infrastructure starts requiring more attention than the product itself.

I think the difficult part is distinguishing between preparing for growth and solving problems that don't exist yet.

For those working with smaller engineering teams:

What's one infrastructure decision you would deliberately postpone until the product reaches a certain scale?

And what would make you introduce it earlier?