Registry Architecture and Components
What you'll learn: How container registries are structured internally to store and serve images efficiently.
Now that you understand what a container registry is, let's peek under the hood to see how it actually works.
The Three Core Components
Think of a container registry like a specialized library system. It has three essential parts working together:
1. Blob Storage
This is the actual warehouse where image layers (the "blobs" or binary large objects) are stored. Each layer you learned about earlier gets saved here as a content-addressable blob—meaning its filename is its SHA256 digest. When you push layer abc123..., blob storage keeps it safe and retrieves it on demand. This is typically backed by object storage like S3, Azure Blob Storage, or Google Cloud Storage.
2. Manifest Storage
Manifests are the catalog cards of your library. They're JSON documents that describe which layers make up an image, in what order, and with what configuration. When you pull myapp:v1.2, the registry first retrieves the manifest to learn which blobs to fetch. Manifests also store metadata like architecture (amd64, arm64) and OS.
3. The Registry API
This is the librarian—the HTTP/REST interface that handles all push and pull requests. It implements the OCI Distribution Specification, providing standardized endpoints like /v2/<name>/manifests/<tag> for fetching manifests and /v2/<name>/blobs/<digest> for downloading layers. This API authenticates users, validates requests, and orchestrates the flow between clients (like Docker) and storage backends.
When you docker push, the CLI uploads blobs to blob storage, then sends the manifest. On docker pull, it fetches the manifest first, then downloads only the layers not already cached locally.
Key Takeaway: A container registry consists of blob storage (layers), manifest storage (image metadata), and a standardized API that orchestrates image push/pull operations—working together to serve images efficiently and securely.