Amazon DocumentDB

Amazon DocumentDB is a fully managed NoSQL database service on Amazon Web Services that runs MongoDB-compatible databases and stores JSON document data structures.

AI Practitioner focus

  • DocumentDB supports MongoDB-compatible document workloads and vector search, so an application can keep JSON business records and embeddings in a managed document database.
  • It is useful when the existing data model is document-oriented; do not confuse it with Amazon DynamoDB key-value access or Amazon Neptune graph traversal.

Key points

  • Non-relational database service, purpose-built for document data structures (document databases store JSON data).
  • Builds clusters of 0–16 database instances (1 primary and up to 15 read replicas) plus a cluster storage volume.
  • Provides 99.99% availability by copying the cluster’s data across three different Availability Zones.
  • Automatic failover to one of up to 15 replicas in other Availability Zones, or to a new instance if no replicas are provisioned.
  • Storage and compute scale independently.
  • Provides backup capability with point-in-time recovery for the cluster; backup retention period of up to 35 days.
  • Best suited for content-management, catalog, and user-profile workloads with flexible, semi-structured schemas; supports ACID properties for transactions across one or more documents. Not a time-series database — that role belongs to Amazon Timestream.
  • Distinct from Amazon DynamoDB, a key-value NoSQL database offering sub-millisecond latency; DocumentDB is a document database purpose-built for JSON.
  • Monitored via Amazon Cloudwatch to track health and performance of the instances in a cluster.

Sources