Database
Amazon DynamoDB
Serverless NoSQL at any scale
Amazon DynamoDB is a fully managed key-value and document database delivering single-digit-millisecond performance at virtually any scale. Capacity modes: on-demand (per-request billing) and provisioned (with auto scaling). DAX adds microsecond in-memory caching; global tables replicate across Regions.
When to use it
- Key-value/document access at unpredictable or internet scale
- Session stores, user profiles, shopping carts, gaming leaderboards
- Serverless architectures wanting per-request billing
- Multi-Region active-active writes via global tables
When NOT to use it
- Complex joins, ad-hoc SQL, relational integrity (use RDS/Aurora)
- Analytics aggregations over large datasets (use Redshift)
- Graph traversals (use Neptune)
Key exam points
- On-demand vs provisioned capacity modes trade flexibility vs predictable pricing
- DAX caches reads in memory for hot keys
- Global tables = multi-Region, multi-active replication
- Partition key design drives performance; hot partitions throttle
Compared with
- Amazon RDS — RDS is relational SQL with vertical scaling and read replicas; DynamoDB is horizontally scalable NoSQL.
- Amazon ElastiCache — ElastiCache is an in-memory cache in front of a database, not a durable primary store.