GCP optimization has a constraint that surprises people who come from AWS: the project structure is load-bearing in ways that are hard to change
In AWS, most resources exist in a single account with region as the primary organizational boundary. Moving things around within an account is generally possible, sometimes annoying
In GCP, the project is the fundamental billing and access boundary. Many resources are project-specific and cannot be moved between projects. Cloud SQL instances, GKE clusters, persistent disks - these live in a project and stay there
This matters for optimization because the instinct when you find a project that should be consolidated or reorganized is to move its resources. Often you can't move them. You can recreate them in the right project and migrate the data, but that's a different and much more involved operation than moving them
Before designing any optimization that involves restructuring GCP projects, understand which resources are portable and which aren't. Build the migration plan around the immovable resources, not the other way around
Organization policy constraints are the second GCP-specific caveat
In GCP organizations with mature security posture, Organization Policies control what can be done across all projects. Policies that restrict which regions resources can be deployed in, which machine types are allowed, which APIs can be enabled
These policies exist for good reasons. They encode security and compliance requirements at the organization level so they can't be overridden by individual project administrators
They also mean that the cheapest compute option might be in a region that's blocked by policy. Or that the machine type recommended by the Recommender isn't on the allowed list
Understanding the organizational constraints before designing optimizations saves the frustration of designing a solution you can't actually deploy. Ask the organization admin for the active constraint list before you start planning
Work within the constraints. They're usually there for a reason
