What it is
Object storage treats each file as a self contained object: the raw bytes plus a bundle of metadata, addressed by a single unique key. Unlike a filesystem there are no nested directories, no inodes, and no in place editing. You put a whole object or get a whole object.
Why it scales
- The namespace is flat, so the system never walks a deep directory tree to find a file.
- Objects are immutable in practice: you replace the whole thing rather than seeking into the middle. That makes replication and caching simple.
- Metadata travels with the object, so a request needs no separate stat call.
What you give up
- No partial writes or appends in the middle of a blob.
- No real folders, only a key that may look like a path.
- Strong filesystem semantics like locking are replaced by simpler put and get.
Key idea
Object storage swaps a mutable directory tree for a flat keyed store of immutable blobs, trading rich filesystem semantics for massive scale.