What is Amazon Aurora DSQL?
Amazon Aurora DSQL is a fully serverless, distributed relational database. AWS introduced the service at re:Invent 2024 and made it generally available on May 27, 2025. DSQL stands for Distributed SQL and addresses a fundamental limitation of classical relational databases: the challenge of executing consistent, scalable transactions across multiple geographic regions. Aurora DSQL enables active-active operation in multiple AWS regions simultaneously, where all regions execute both read and write operations with full ACID consistency and multi-region strong consistency.
The technical approach differs fundamentally from classical relational databases. Aurora DSQL uses optimistic concurrency control (OCC) instead of conventional pessimistic locking. Transactions run without row locks; at commit time, the system automatically checks for conflicts. When a write conflict occurs between concurrent transactions, one transaction is aborted and must be retried by the client. This approach enables very high throughput at low latencies, but requires conflict-tolerant application logic. The distributed architecture separates compute, commit, and storage and operates without a single primary node: there is no single point of failure. Aurora DSQL is compatible with a subset of PostgreSQL 16 and requires no manual sharding or replication configuration.
Core Features
- Active-active multi-region operation: All regions process read and write traffic simultaneously with strong consistency and automated zero-data-loss failover.
- Fully serverless: No cluster management, no capacity planning, no maintenance windows. Compute, read, write, and storage capacity scale automatically and down to zero when idle.
- PostgreSQL compatibility: Compatible with a subset of PostgreSQL 16, including support for PostgreSQL views. Existing drivers, ORMs, and tools remain usable.
- Optimistic concurrency control: Lock-free transaction processing for high throughput without deadlocks.
- Security and governance: IAM authentication, KMS encryption with customer managed keys, AWS PrivateLink (also over Direct Connect), AWS Backup, CloudFormation, CloudTrail, and resource-based policies for cross-account access.
- AI and tooling integration: A Model Context Protocol (MCP) server helps AI agents with schema design, performance optimization, and database operations. Query plans provide statement-level cost estimates.
Typical Use Cases
- Globally distributed applications: Low latency for users in multiple regions through local reads and writes.
- Financial platforms and e-commerce: Transactional systems that require end-to-end consistency and high availability across regions.
- Serverless and event-driven architectures: Databases for highly variable workloads that should scale to zero automatically.
- Microservices with distributed transactions: Consistent data storage without manual sharding or custom replication logic.
- SaaS platforms: Multi-tenant applications with unpredictable growth and EU data residency requirements.
Benefits
- High availability: 99.99% SLA in single-region operation and 99.999% in multi-region operation, with no single point of failure.
- Usage-based billing: Pay for Distributed Processing Units (DPUs) and storage (GB-month). No DPU charges accrue when idle. An ongoing free tier covers the first 100,000 DPUs and 1 GB of storage each month.
- No operational overhead: AWS handles patching, scaling, and replication automatically.
- EU data residency: Available in Frankfurt (eu-central-1), Ireland, London, Paris, and Stockholm. Multi-region clusters can be built within the EU.
- Future-proof architecture: Distributed design without sharding complexity and with native AI tooling integration.
Integration with innFactory
innFactory supports you in evaluating Aurora DSQL for your use case, adapting your application architecture to optimistic concurrency control including retry logic, and planning a multi-region strategy on AWS. We assess the PostgreSQL compatibility of your existing application, plan data residency in EU regions, and integrate Aurora DSQL into your infrastructure-as-code and CI/CD processes. As an AWS partner, we support you from proof of concept through to production multi-region operation.
Typical Use Cases
Frequently Asked Questions
What is Amazon Aurora DSQL?
Aurora DSQL is a serverless, distributed SQL database from AWS. AWS introduced the service at re:Invent 2024 and made it generally available on May 27, 2025. Aurora DSQL provides active-active operation across multiple AWS regions simultaneously, PostgreSQL compatibility, and fully serverless scaling without manual sharding or replication configuration.
What distinguishes Aurora DSQL from Aurora Serverless v2?
Aurora Serverless v2 is a single-region service with a primary write node. Aurora DSQL is designed from the ground up for multi-region active-active operation: all regions process read and write operations simultaneously, with no primary node and no single point of failure. Aurora DSQL uses optimistic concurrency control instead of classical locking and is specifically built for highly distributed, latency-optimized workloads.
What is Optimistic Concurrency Control (OCC)?
With OCC, database rows are not locked for the duration of a transaction. Instead, the system checks for conflicts at commit time. On conflict, the transaction is aborted and must be retried. This enables very high throughput without deadlocks, but requires conflict-tolerant application logic with a retry mechanism.
Is Aurora DSQL PostgreSQL compatible?
Yes, Aurora DSQL is compatible with a subset of PostgreSQL 16. Existing PostgreSQL drivers, ORMs, and tools work in principle. However, there are limitations: not all PostgreSQL extensions are supported, the use of SELECT FOR UPDATE or other lock-based patterns must be reviewed, and each DDL statement runs in its own transaction.
How much does Aurora DSQL cost and how is it billed?
Aurora DSQL bills based on actual usage: database activity in Distributed Processing Units (DPUs) plus storage in GB-month. DPUs cover compute, read and write I/O, change data capture, and multi-region replication. When idle, the service scales to zero and incurs no DPU charges. An ongoing free tier covers the first 100,000 DPUs and 1 GB of storage each month.
In which regions is Aurora DSQL available?
Aurora DSQL is available in many AWS regions, including several EU regions: Frankfurt (eu-central-1, since October 2025), Ireland, London, Paris, and Stockholm. For EU data residency, multi-region clusters can be built within the European regions. Additional regions are available in the US and Asia Pacific.
When should I choose Aurora DSQL over Aurora Global Database?
Aurora Global Database provides multi-region read access with a primary write region, suited for read-intensive workloads. Aurora DSQL enables true active-active writes in multiple regions simultaneously with multi-region strong consistency and without Global Database routing overhead. For new applications with global distributed write requirements, Aurora DSQL is the more modern choice.