Amazon DynamoDB

Amazon DynamoDB is AWS's serverless NoSQL database for key-value and document data. You create tables, write items, and read them back by key with single-digit millisecond latency, which is the vendor's own performance claim. There are no servers to size, no maintenance windows and no cold starts, according to AWS. Billing follows either per-request on-demand pricing or provisioned capacity.

For global applications, global tables replicate data across Regions with multi-active writes and up to 99.999% availability. The product page lists vector search for agent memory and retrieval, ACID transactions, DynamoDB Streams, Time to Live and bulk import. AWS positions it for shopping carts, session stores, user profiles, game state, ad-tech bidding and device data, along with workloads moving off relational systems.

• On-demand and provisioned modes. On-demand bills per read and write request with automatic scaling, and AWS recommends it in most scenarios. Provisioned mode lets you set read and write capacity in advance, which fits steady traffic and unlocks committed-use discounts.
• Global tables. A multi-Region, multi-active setup designed for up to 99.999% availability. AWS says it supports multi-Region strong consistency, so applications read the same data from any Region and can aim for zero recovery point objective.
• Transactions and streams. ACID transactions cover multi-item changes that must succeed together. DynamoDB Streams emits item changes to trigger near real-time processing, and Time to Live automatically expires old records, which helps with IoT telemetry and session data.
• Native vector search. AWS lists vector search for agent memory, recommendations and retrieval augmented generation. Embeddings are stored beside operational data, so an agent can run similarity search without a separate vector database.
• Two table classes. Standard suits frequently accessed data, while Standard-Infrequent Access lowers storage cost for data read less often. Request prices differ between the two classes, so the choice depends on how often the data is touched.
• Backup and restore. Point-in-time recovery, on-demand backups with warm and cold storage, and restore to a new table are priced as add-ons. Cold backup storage is supported for backups managed by AWS Backup, with a 90-day minimum.

• A mobile game studio launching a new title. Launch day traffic is hard to predict. On-demand mode absorbs spikes, and AWS lists player profiles, game state, leaderboards and matchmaking as typical data. Costs stay near zero when the game is idle.
• An online retailer with seasonal peaks. Carts, inventory, order history and sessions are the vendor's examples. Key-based access keeps response times flat during sales, and global tables can keep a second Region ready for recovery if the first becomes unavailable.
• A team building AI agents. Agent memory and retrieval can sit in the same table as operational records, using the built-in vector search. That avoids synchronizing a separate vector store, though the team should check query features against its needs.

With an AWS account you create a table in the console, choose a partition key and a capacity mode, and start writing items through the SDKs for languages such as Java, Python, JavaScript, .NET and PHP. AWS points to a developer guide on capacity modes and table classes, a free tier, hands-on training, and documentation.

Learning the access-pattern approach, in which you design tables around the queries you run, is the main hurdle for people used to relational databases. Bulk import exists for moving data in from other systems. The vendor does not state a typical setup time.

DynamoDB bills reads and writes, either on demand in request units or as provisioned capacity, plus storage. The free allowance is 25 GB and 25 read and write units a month.

You can export a table to S3 in DynamoDB JSON or Ion without using read capacity, and a downloadable version exists for local testing. The service itself runs only on AWS, so portability means rewriting access code.

In our database management systems ranking Amazon DynamoDB stands at #12 of 22: Oracle Database sits directly above it and FerretDB directly below. From place 4 the established names follow in their ranked order; we publish no numeric score.

North America is home to 15 of the 22 vendors, Amazon DynamoDB among them (the United States).

Pricing as recorded: Per request or provisioned capacity, published; monthly free allowance.

AWS offers on-demand pricing per million write and read request units, and provisioned pricing by capacity, each with Standard and Standard-IA table classes. Storage, global table replication, backups, restores, streams and data transfer are separate lines. The fetched page loaded unit prices dynamically, so amounts are not quoted here. The free tier, per Region and per payer account each month, gives 25 WCUs and 25 RCUs, 25 GB of storage, 2.5 million stream reads and 1 GB of data transfer out. AWS says that covers roughly 200M requests a month, and adds up to $200 in Free Tier credits for six months. Database Savings Plans save up to 18% on on-demand and 12% on provisioned throughput; reserved capacity saves up to 54% (1 year) or 77% (3 years).

Checked against aws.amazon.com/dynamodb for Best Database Management Systems in 2026, review date 7 November 2026.

Exports to S3 do not consume read capacity.

The downloadable local version exists so developers can test without the web service; production runs on AWS.

AWS says DynamoDB offers a broad set of security controls and compliance standards for governments and global banks. The financial services section names SOC 1/2/3, PCI, FINMA and ISO. Fine-grained access control and IAM permissions are referenced.

You choose the AWS Region where each table lives, and global tables replicate to the Regions you add. The pages checked do not describe an export format, though bulk import and point-in-time recovery are listed.

Integrations the vendor names (9):

• AWS IAM
• Amazon CloudWatch
• AWS Backup
• Amazon Kinesis Data Streams
• DynamoDB Streams
• DynamoDB Accelerator (DAX)
• ExtendDB for edge IoT
• AWS SDKs
• Database Savings Plans

Shared tags and a similar price model decide who is listed here; where none match, the neighbours in the ranking fill in.

• CockroachDB (#4 of 22): Distributed SQL database sold per vCPU-hour, with a revenue-capped free licence. Best for: Teams needing SQL that stays available across regions or clouds. (United States)
• Couchbase (#8 of 22): JSON document database with SQL++ queries, sold as Capella or as Server. Watch out: the free Capella cluster has forum support only. (United States)
• MySQL (#9 of 22): Suits: Teams running widely supported web stacks who accept Oracle as owner.
• Amazon DocumentDB (#17 of 22): AWS-managed document database with a MongoDB-compatible API. Suits: AWS teams that want to keep MongoDB drivers under an AWS invoice. (United States)

Read from the vendor’s own pages on 5 November 2026.

• aws.amazon.com/dynamodb: product description, benefits, use cases, industries, global tables, compliance names
• aws.amazon.com/dynamodb/pricing: capacity modes, table classes, free tier, savings plans and reserved capacity percentages, backup add-ons

We test for four things rather than for features: setup time, price in practice, exit and audience. Amazon DynamoDB ends up at #12 among 22. Details: how we rank software.