Amazon DocumentDB

One of the key features of a document database is its ability to natively store JSON data. Being able to store data the same way that it is modeled in the application in the database reduces friction and help improve productivity.

Amazon DocumentDB enables you to write ad hoc queries to answer questions that you have of your JSON documents. The API enables single document lookups, index scan, regular expression queries, and aggregations.

With Amazon DocumentDB you are able to create single field, compound, and multi-key indexes on your collection that enable you optimize the performance of your query patterns. Reads from the indexes on the primary instance are read-after-write consistent and you can delete or create new indexes at any time.

Amazon DocumentDB was built as an enhancement to the Aurora PostgreSQL relational database system [4]. Its architecture separates storage and compute so that each layer can scale independently, though the system is limited to a single writable master. Amazon DocumentDB is uses the Aurora Storage Engine, a purpose-built storage system originally built for the MySQL relational database. This storage engine is distributed, fault-tolerant, self-healing, highly available, and durable, which it maintains by replicating data six ways across three AWS Availability Zones (AZs). Amazon DocumentDB databases cannot span AWS Regions or span cloud providers, nor can Amazon DocumentDB run on-premises. The system can support up to 15 low-latency readable replicas and continuously backs up all changes to Amazon S3.