AWS NoSQL Databases: Types, Services & FAQs
What is AWS NoSQL?
NoSQL databases allow users to store data with a variety of data models and flexible schema. These databases are user-friendly for developers, and offer the functionality and high performance modern applications require.
There are several types of NoSQL databases among the AWS NoSQL database offerings. Users can select both managed and self-managed AWS NoSQL database services designed to support cloud-based workloads and integrate seamlessly with existing AWS resources.
Is Amazon Aurora a NoSQL database?
No. Amazon Aurora is AWS’s fully managed relational (SQL) database, not a NoSQL one. Questions about AWS Aurora NoSQL support usually come from teams comparing Aurora to true AWS NoSQL workloads that need flexible schemas, like Amazon DynamoDB or ScyllaDB.
Is Amazon Aurora a NoSQL database?
The AWS NoSQL vs SQL decision comes down to data structure and scale. Choose a NoSQL database when your data model is flexible or rapidly changing and you need horizontal scale; choose a SQL database like Amazon RDS or Aurora when you need strict relational integrity and complex joins. For most high-throughput, low-latency workloads, the AWS SQL vs NoSQL trade-off favors NoSQL options such as DynamoDB or ScyllaDB.
Does AWS have a NoSQL Workbench tool?
Yes. AWS NoSQL Workbench is a visual tool (built for DynamoDB) for data modeling, query building, and sample-data generation. It’s a helpful starting point for any AWS NoSQL project before deploying a NoSQL Workbench AWS data model to production.
How much does an AWS NoSQL database cost?
AWS NoSQL database pricing varies by service. DynamoDB uses pay-per-request or provisioned-capacity billing, while other AWS NoSQL pricing models (like Keyspaces or DocumentDB) are based on read/write capacity and storage, so compare AWS NoSQL pricing carefully before committing. ScyllaDB Cloud prices on capacity rather than per request, with Flex Credit for committed baseline plus on-demand burst.
What are AWS NoSQL best practices?
Following AWS NoSQL best practices starts with matching the database type to your access pattern, provisioning capacity for actual traffic instead of guessing, and monitoring latency at the p99 level, not just averages. These same principles apply whether you run DynamoDB or a DynamoDB-compatible alternative like ScyllaDB.
Does AWS support NoSQL databases?
Yes, there are a number of AWS NoSQL database options:
Key-value databases. Key-value databases allow users to store data in pairs containing a data value and a unique ID. Since values can hold any data structure or amount and are not assigned to a table, this provides a flexible storage structure. Key-value databases can handle huge volumes of requests.
Common use cases for these databases include eCommerce sites, gaming applications, and other high traffic applications. The relevant AWS service here is Amazon DynamoDB, a key-value database.
Document databases. Document databases and key-value databases are structured similarly, except that data is written in a markup language like XML, JSON, or YAML and stored in documents rather than in keys and values. Hierarchies of data can be stored in these databases by linking documents.
Common use cases for document databases include catalogs, user profiles, and content management. Amazon DynamoDB can be used here as well, but the main AWS service in this category is Amazon DocumentDB.
Wide column databases. Wide column databases lack a strict column format but are still based on tables. However, users can combine columns containing different data formats and segments of rows, and rows do not need a value in every column.
Common use cases for wide column databases include fleet management, route optimization, and industrial maintenance applications. Amazon Keyspaces for Apache Cassandra is the AWS service in this category.
Graph databases. Graph databases are structures made up of nodes and edges. Nodes are the unique data values and the relationships between them are edges. Graph databases enable the intricate tracking of closely related data in an organic network rather than a structured table.
The database is designed for big data applications and use cases for graph databases include social networking, recommendation engines, and fraud detection. The AWS graph database is Amazon Neptune.
Time series databases. Time series databases sort and store data in time ordered streams by the time of ingestion, collection, or other timestamps in the metadata, not by ID or value.
These allow users to query data based on time intervals for DevOps, industrial telemetry, Internet of things (IoT) applications, and similar use cases. Amazon Timestream is the relevant AWS service. Note that Timestream now ships as two offerings: Timestream for InfluxDB (managed InfluxDB, and the option available to new customers) and Timestream for LiveAnalytics, AWS’s original in-house engine, which has been closed to new customers since June 2025.
In-memory data stores. In-memory databases keep the working set in RAM instead of on disk, trading durability guarantees for sub-millisecond response times. Amazon ElastiCache covers caching and ephemeral state; Amazon MemoryDB adds durability with a multi-AZ transaction log, so it can serve as a primary database rather than a cache.
AWS NoSQL Databases Services: An Overview
AWS offers a range of AWS NoSQL services to fit different access patterns and scale requirements. Here are the AWS NoSQL options and their basic features:
Amazon DynamoDB. Amazon DynamoDB is a document and key-value database. It is a fully managed service that stores items as collections of attributes in DynamoDB tables, with point-in-time recovery, on-demand backup and restore, and encryption at rest. Optional in-memory caching is available through DynamoDB Accelerator (DAX), a separate service.
Amazon DynamoDB enables encryption at rest by default. ACID transactions are available through the TransactWriteItems and TransactGetItems APIs, which you opt into per request and which consume roughly twice the capacity of standard operations. Capacity adjusts automatically in on-demand mode, or in provisioned mode when you enable auto scaling.
Amazon ElastiCache. An in-memory data store, Amazon ElastiCache offers fully managed support for Valkey, Redis OSS, and Memcached in place of a disk-based database. AWS recommends Valkey for new deployments and prices it below Redis OSS. ElastiCache supports sub-millisecond response times and scales through memory sharding.
Typical use cases include caching, real-time analytics, queuing, and session stores.
Amazon MemoryDB. MemoryDB is a durable, Valkey- and Redis OSS-compatible in-memory database. Unlike ElastiCache, it persists writes to a distributed multi-AZ transaction log, which makes it usable as a primary datastore rather than a cache in front of one.
Amazon Neptune. Amazon Neptune is a fully managed graph database service that is optimized for storing data from billions of relationships. It supports two data models with three query languages: Apache TinkerPop Gremlin and openCypher for property graphs, and SPARQL for W3C RDF graphs. For property graphs, you can query the same data in either Gremlin or openCypher regardless of which language wrote it.
Neptune features include continuous backups, multi-zone data replication, point-in-time recovery, and read replicas. It provides encryption in-transit and at-rest and supports ACID transactions.
Amazon Timestream. Timestream is AWS’s managed time series offering, available in two forms. Timestream for InfluxDB runs managed InfluxDB instances and is the option open to new customers. Timestream for LiveAnalytics is AWS’s serverless in-house engine with automated compression, rollups, retention, and tiering, plus built-in interpolation and smoothing functions; it has been closed to new customers since June 20, 2025, and AWS recommends that existing users migrate to Timestream for InfluxDB.
The Timestream adaptive query processing engine includes features for automated data compression, rollups, retention, and tiering. It also includes built-in analytics for data approximation, interpolation and smoothing.
QLDB provides full transactional support, a SQL-like API, and a flexible document data model. It includes ACID compliant transactions, automatic scaling, data streaming with Kinesis Data Streams, and multizone availability.
Amazon DocumentDB. Compatible with MongoDB, Amazon DocumentDB is a fully managed document database designed with architecture that separates storage and compute resources for improved flexibility and scalability. It replicates data across availability zones for durability, supports up to 15 read replicas, and scales to 256 TiB of storage. AWS Database Migration Service is free for six months when you migrate to DocumentDB.
DocumentDB Elastic Clusters shard workloads across nodes for millions of reads and writes per second, and DocumentDB is compatible with the majority of MongoDB APIs, drivers, and tools through MongoDB 8.0.
Amazon Keyspaces. Compatible with Apache Cassandra, use the managed wide column database Amazon Keyspaces to use Cassandra native code and tools and migrate Cassandra workloads and applications. Its features enable autoscaling and using both provisioned and on-demand resources. Teams already running Apache Cassandra can reuse existing CQL drivers and tooling. Note that Keyspaces implements the Cassandra 3.11.2 API (CQL 3.4.4) and does not support every Cassandra feature – so audit your schema and query patterns before migrating.
Does ScyllaDB Offer AWS NoSQL Solutions?
ScyllaDB is a fast and scalable NoSQL database that is API-compatible with DynamoDB. With ScyllaDB’s open source DynamoDB API, DynamoDB-compatible applications can be deployed wherever you want: on-premises or on any public cloud.
Moreover, ScyllaDB Cloud or ScyllaDB Enterprise, both created for data-intensive applications that demand low latency and high performance, are available on AWS Marketplace
Many teams have found that ScyllaDB gives their AWS NoSQL stack DynamoDB-compatible performance with greater flexibility and at lower cost.
Learn more here.