contentaddressed

Content-addressed refers to a design principle in which data is addressed by a representation of its content rather than by its location. In content-addressed storage (CAS), each object is identified by a cryptographic hash of its data, sometimes with a type or length prefix. The hash becomes the object’s address; identical content yields the same address, and changes produce a new one. This provides data integrity and supports immutability, since identity depends on the data itself.

How it works: data objects are stored in a key-value store keyed by their content-derived hash. Retrieval uses the hash to fetch the object; if missing, it may be retrieved from replicas. Deduplication occurs automatically because identical content maps to a single address. Systems often use Merkle trees or DAGs to enable efficient verification across large data sets.

Examples and use cases: Git stores objects (blobs, trees, commits) by content hash (SHA-1 or SHA-256). IPFS uses content addressing to fetch data from a peer-to-peer network. CAS is used in cloud storage, backup, and database systems to enable deduplication, versioning, and offline synchronization.

Advantages include data integrity, deduplication, and caching efficiency. Limitations include reliance on a collision-resistant hash function, potential performance costs for retrieval from distributed graphs, and the need to manage multiple content addresses for edits. Although hash collisions are rare, most designs mitigate this with strong hash functions and supplementary metadata.