Document Stores
Agile schemas, JSON blobs, and horizontal scale. Often misused as a substitute for database design.
The Schema-less Trap
"Schema-less" is a marketing term. In reality, it means "schema-on-read." The structure of your data isn't enforced when writing to the database, meaning your application code must handle the complexity of missing fields, different data types, and legacy formats every time it reads a record.
When to ACTUALLY use Document Stores:
- Storing highly polymorphic data (e.g., product catalogs where a TV has different attributes than a T-shirt).
- High-volume data ingestion where strict validation can happen asynchronously.
- You need horizontal scaling out-of-the-box (sharding).
Major Players
MongoDB
The undisputed king of the document space. It has matured significantly, adding multi-document ACID transactions, but still excels when data is accessed together, stored together.
Compare with PostgreSQLCouchbase
Memory-first architecture. Extremely fast for key-value and document operations. Often chosen for caching tiers that evolve into persistence layers.