Store Avatar Metadata in a Database, Not in Bucket Listings
The article argues that avatar bytes belong in object storage while ownership, content type, dimensions, and the active object key should live in the application database. It explains that a prefix-only bucket listing cannot serve as a metadata search index, so account pages should read the database row, check authorization, and issue a signed read for the exact private key. To handle retries after interrupted uploads, it recommends deterministic object keys and a single database row naming the active key, updated in a transaction. Bucket listings are presented as a recovery and reconciliation tool rather than a customer request path. The author also notes that globally unique hashes alone make tenant-level recovery harder because they remove useful grouping.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.



Discussion (0)
Log in to join the discussion and vote.
Log in