Question: Is DynamoDB a wide-column store?

Answer

Not officially. AWS classifies DynamoDB as a key-value database that also supports document data such as maps and lists, and it lists Amazon Keyspaces (for Apache Cassandra) as its wide-column database. DynamoDB still gets grouped with wide-column stores such as Apache Cassandra, ScyllaDB and Google Bigtable, because its data model looks similar.

What DynamoDB shares with wide-column stores

• Partitioned, sorted rows. Items with the same partition key form an item collection, stored together and ordered by sort key. That's close to a Cassandra partition with clustering columns, or a range of Bigtable rows.
• Sparse, flexible attributes. Apart from the key, each item can have different attributes, the way rows in a wide-column store can hold different columns.
• Hash distribution. DynamoDB hashes the partition key to spread data and traffic across partitions, much like Cassandra's partitioner.

Where it differs

• There are no column families and no per-cell versions. Bigtable, for example, can keep several timestamped versions of each cell.
• It has native document types: maps, lists and sets, which can be nested.
• You query through the primary key or secondary indexes with the DynamoDB API or PartiQL, not CQL.
• It's serverless: there are no nodes, clusters or compaction settings to tune.

In practice, Cassandra data modeling habits carry over well: design tables around your queries, denormalize, and choose partition keys that spread traffic. DynamoDB vs Cassandra compares the two in detail, and DynamoDB vs Amazon Keyspaces covers AWS's managed Cassandra-compatible service.

Source: AWS DynamoDB documentation.