10.3.2.2 GAR Regional Repositories

GAR regional repositories allow an Artifact Registry repository to be created in a specific Google Cloud region, placing image storage geographically close to where it will actually be consumed, an important consideration for both pull latency and certain data residency requirements.

Creating a Repository in a Specific Region

The region is specified explicitly at the time a repository is created.

gcloud artifacts repositories create myapp-repo --repository-format=docker --location=us-central1

This repository's content is stored specifically in the us-central1 region, rather than some other location.

Why Regional Placement Affects Pull Performance

A workload running in the same region as the repository it's pulling from generally experiences lower latency and potentially faster pulls than one pulling across a greater geographic distance.

docker pull us-central1-docker.pkg.dev/my-project/myapp-repo/myapp:1.0

A pull like this, performed from a workload also running in us-central1, benefits from this regional proximity.

Why Multiple Regional Repositories May Be Appropriate for a Global Deployment

A deployment spanning several regions might maintain a separate, regionally placed repository (or use a multi-regional configuration) in each region to ensure every deployment location has low-latency access to its images.

gcloud artifacts repositories create myapp-repo --repository-format=docker --location=europe-west1

A second repository like this, placed in a different region, serves deployments running there with the same regional proximity benefit.

Considering Data Residency Requirements

For organizations with specific data residency requirements, choosing a repository's region deliberately, in line with those requirements, is an important consideration beyond pure performance.

gcloud artifacts repositories create eu-only-repo --repository-format=docker --location=europe-west1

Why GAR Regional Repositories Matter

Deliberately choosing a repository's region — balancing pull performance for nearby workloads against any specific data residency considerations — is an important aspect of designing an efficient, compliant registry setup for a deployment spanning Google Cloud's various regions.