Database
Amazon DocumentDB
Fully managed document database compatible with MongoDB: you can use the same application code, drivers, and tools as with MongoDB. It handles operating and scaling MongoDB workloads in the cloud.
Exam phrases that point here
«managed document database that is compatible with MongoDB»
Amazon DocumentDB is compatible with the MongoDB API: existing applications use the same drivers and tools, and AWS handles operations: automated backups, health monitoring, and failure recovery.
When not: Not all MongoDB APIs are supported, so you should verify compatibility. If the scenario does not mention MongoDB and asks for key-based access at any scale with no servers, Amazon DynamoDB is usually a better fit.
«migrate an existing MongoDB database with minimal downtime»
AWS DMS can move a self-managed MongoDB database (on premises or on EC2) to DocumentDB with practically no downtime.
When not: If the source database is relational (for example, Oracle or MySQL), the logical target is Amazon RDS or Aurora, even though the migration tool is still AWS DMS.
«JSON documents with a flexible schema and ad hoc queries, indexes, and aggregations»
A document model stores data whose structure changes (catalogs, user profiles, content) and lets you query it by any field with indexes and aggregations.
When not: DocumentDB has its own text indexes, but with limits (English only, one text index per collection); if text search is the focus of the scenario, think of Amazon OpenSearch Service, a search and analytics engine. If access is always by key and you want a serverless millisecond-latency service, DynamoDB.
«low-latency reads in other Regions and recovery from the failure of a Region»
Global clusters replicate the cluster from a primary Region to read-only secondary Regions, normally in under a second, and a secondary can be promoted if the primary fails.
When not: In a global cluster only the primary Region accepts writes. If the scenario asks to write in several Regions at once (active-active), it points to DynamoDB global tables.
«scale writes horizontally with sharding»
Elastic Clusters distributes data across shards by hash and lets you increase or decrease shards without changing the application, with little or no downtime.
When not: If only reads need to scale, adding replicas to a regular cluster is enough. An existing cluster is not converted to an Elastic Cluster: you must migrate the data with AWS DMS.