Amazon DynamoDB vs CockroachDB: Which Is Right for You?

Architecture and data model

Amazon DynamoDB is a fully managed, serverless NoSQL database offering key-value and document models with single-digit millisecond latency at effectively unlimited scale. There are no servers or clusters to manage; capacity is expressed as read and write units or handled automatically in on-demand mode. The design rewards modelling data around known access patterns, frequently using single-table designs and global secondary indexes, because DynamoDB does not support joins or arbitrary ad hoc queries the way a relational engine does.

CockroachDB is a distributed SQL database with a PostgreSQL-compatible interface. It stores data as automatically replicated ranges and uses the Raft consensus protocol to maintain strong consistency while scaling horizontally across nodes and regions. Unlike DynamoDB, CockroachDB preserves the full relational toolkit: multi-statement ACID transactions, secondary indexes, foreign keys, and flexible SQL queries that are not constrained to predefined access paths.

The choice therefore turns on data model rather than scale alone. DynamoDB excels when the workload is well-defined key-value or document access at very high throughput, while CockroachDB suits applications that genuinely need relational semantics and query flexibility but also require distributed scale and resilience.

Pricing comparison

DynamoDB on-demand pricing is $1.25 per million write requests and $0.25 per million read requests after a November 2024 reduction that cut on-demand throughput prices by 50 percent and global tables by up to 67 percent. Provisioned capacity costs about $0.00065 per write-capacity-unit-hour and $0.00013 per read-capacity-unit-hour, which is cheaper for sustained, predictable traffic. Global secondary indexes and DynamoDB Streams add cost, and AWS Database Savings Plans launched in December 2025 offer modest additional discounts.

CockroachDB self-hosted is licensed per core, roughly $1,500 to $3,000 annually depending on support tier, with contracts often starting near $50,000 and scaling into six figures for large clusters. CockroachDB Cloud offers consumption-based Standard and Advanced plans with data transfer, backups, and change data capture billed separately since the late-2024 pricing update. Pricing verified June 2026. Enterprise pricing for both requires a quote.

Fit, consistency, and operations

DynamoDB offers eventual consistency by default and strongly consistent reads as an option, and it removes operational burden almost entirely because there is nothing to provision or patch. Its main limitation is modelling rigidity: changing access patterns after launch can require expensive index redesign, and complex analytical queries usually need a secondary system. CockroachDB defaults to strong consistency and serializable isolation, which simplifies correctness for transactional applications, at the cost of consensus latency on distributed writes and the need to operate a cluster competently.

A genuine limitation of DynamoDB is that it locks the application into AWS and into a NoSQL modelling discipline that does not suit every team. A genuine limitation of CockroachDB is operational complexity and the write-latency overhead of multi-region consensus. Teams should pick based on whether their workload is naturally key-value at scale or relational with distributed requirements.

When to choose Amazon DynamoDB

Choose Amazon DynamoDB if your workload is high-throughput key-value or document access on AWS, if you want a serverless database with no infrastructure to manage, or if predictable single-digit millisecond latency at scale is the priority. DynamoDB is a strong fit for shopping carts, session stores, gaming state, IoT ingestion, and event-driven architectures where access patterns are well understood and relational querying is not required.

When to choose CockroachDB

Choose CockroachDB if your application needs relational SQL, multi-statement transactions, joins, and flexible querying alongside horizontal scale and strong consistency. It is a strong fit for distributed transactional systems such as payments, ordering, and inventory that must remain available through zone or region failures, and for organisations that want cloud portability rather than committing exclusively to one provider's managed NoSQL service.

Alternatives to both

User sentiment

Buyers frequently note that DynamoDB is valued for its near-zero operational overhead, consistent low latency at scale, and reliability for well-modelled workloads, with the November 2024 price cuts improving its cost story. The most common criticisms are the difficulty of changing access patterns after launch, the cost of global secondary indexes, and the absence of ad hoc querying, which pushes teams toward additional analytics systems. Reviewers of CockroachDB highlight its resilience, strong consistency, and the familiarity of a PostgreSQL-compatible SQL interface for distributed data. Recurring complaints involve the operational learning curve, multi-region write latency, and per-core licence costs that rise with cluster size. Across both, sentiment tracks the workload shape: DynamoDB satisfies teams with predictable key-value access at high throughput, while CockroachDB satisfies teams that need genuine relational semantics with distributed scale and survivability.