Amazon DocumentDB

Amazon DocumentDB is AWS's managed document database for apps built on MongoDB APIs and drivers, run on serverless capacity, instances or sharded clusters.

Amazon DocumentDB keeps application data as documents and answers queries sent through the MongoDB API, so code written for MongoDB drivers runs against a database that AWS operates. Its full name is Amazon DocumentDB (with MongoDB compatibility). It takes patching, backups and monitoring off the team that owns the application. Amazon builds and runs it through Amazon Web Services, which also makes DynamoDB, RDS and ElastiCache. A cluster runs in one of several modes. DocumentDB Serverless adjusts compute in fine-grained steps as demand rises and falls, while provisioned clusters run on memory-optimized instance classes chosen in advance. Elastic Clusters split collections across shards by a shard key and redistribute the data when shards are added. Global Clusters copy a cluster from its primary Region to read-only secondary Regions for disaster recovery and nearby reads. Inside each cluster the engine handles multi-document ACID transactions, aggregation pipelines, change streams, text search and vector search with HNSW or IVFFlat indexes. Storage is replicated across three Availability Zones, and point-in-time restore returns a cluster to any moment within its backup retention period. The service is aimed at development teams that already build on MongoDB drivers, frameworks and tooling and want to move those applications to AWS, typically without code changes or downtime. Serverless targets highly variable workloads that would otherwise be sized for their peak. Banking, healthcare and public sector organizations are a stated audience, with security and compliance controls built for high-sensitivity data. Role-based access scoped to collections and client-side field level encryption let security teams limit who can read a sensitive field. A cluster is created from the AWS console under the same account that runs a company's other AWS services. Content management systems, e-commerce catalogs, mobile and web back ends and generative AI applications are the jobs it is built around. A retailer can keep a catalog whose products carry different attributes, stream every change to downstream services and expire stale session records with TTL indexes. Operators watch clusters in Amazon CloudWatch and grant access through AWS IAM, the same controls they use across their AWS accounts. When an application needs richer search, documents replicate to Amazon OpenSearch Service through a zero-ETL integration. Analysts can also build machine learning models on the same documents in Amazon SageMaker Canvas without writing code.