- A type of db that stores file or Binary Large Objects
- Can’t just use normal db’s b/c they are better suited for fast queries, less storage
- Big files bloat pages, slow every query
- replicas and backups copy blobs, huge lag and downtime
- DB storage costs 10X more than object storage
- How does Object Storage work?
- Flat namespace = quick file lookups
- Immutable writes = no locks or races
- Redundancy = “11 nines” of durability
- client GET /file1 -> Metadata service (lookup file) -> stream back to client
- What should you know for an interview?
- Large files stored in Object storage, metadata in “traditional” db
- Examples:
- Photos/ videos for a social-media or messaging app
- User-uploaded docs for a file-sharing or collab tool
- Static web assets (images, CSS, JS) served behind a CDN
- Log and event archives feeding analytics or security pipelines
- Backup snapshots and db dumbs for disaster recovery
- Machine-learning training datasets (images, audio, etc)
- You can upload directly to your object storage via presigned urls
- Large files are uploaded in chunks and object storage will “stitch” the file back together
- Examples of popular Object (BLOB) stores?
- Amazon S3
- Google Cloud Storage
- Azure Blob Storage