What

A distributed NoSQL database where every row is grouped under a partition key, with named columns inside each partition.

Partition key
    │
    ▼
 user_123
    ├── name → "Jae"
    ├── age  → 29
    └── city → "NYC"

How it scales

The partition key is hashed, and the hash picks the node that owns the row.

           hash(partition key)
                  │
        ┌─────────┼─────────┐
        ▼         ▼         ▼
      Node A    Node B    Node C
      user_1    user_7    user_3
      user_4    user_2    user_8

Writes to different keys land on different nodes, so they run in parallel. Adding nodes adds write capacity.

When to use it

• High write throughput and horizontal scaling
• Query patterns known up front, since you model tables around the queries
• Events, logs, time series, feeds, messaging

Trade-offs

• No joins, no heavy relational queries
• Schema can still be fixed (Cassandra requires one)

Examples: Cassandra, HBase, Bigtable

Wide-column vs key-value

Key-value:    key → one value or blob
Wide-column:  key → structured columns

Wide-column is key-value style distribution with more structure under each key.