Blob Store System Design

Practice: commit and recover a blob upload

Design a hypothetical blob store for large media objects. A user can create a container, upload a new object version, read that version, list objects, and delete a version after its retention period. Keep the object bytes immutable after commit.

Start with the API contract. Define which response means an upload is complete, what a retry returns after a lost response, and whether a listing immediately includes the new version. Separate authentication from authorization to the requested container.

Draw the metadata path and the data path. Show the frontend, manager, metadata store, data nodes, and replicas. Mark the point at which an uploaded version becomes visible to readers. Explain why the manager does not need to carry every payload byte.

Work through these failures in order: