AWS re:Invent 2025 - Build modern applications with Amazon Aurora DSQL (DEV308)

Building Modern Applications with Amazon Aurora DSQL

Overview of Amazon Aurora DSQL

• Amazon Aurora DSQL is a serverless, relational database service that provides the benefits of a traditional database with the scalability and flexibility of a serverless architecture.
• Key features of Aurora DSQL:
  • Fully managed, no infrastructure to provision or manage
  • Automatic scaling of compute and storage
  • 5 nines of availability in a single region, 5 nines across multiple regions
  • Strong transactional consistency and ACID compliance
  • Familiar SQL interface and compatibility with PostgreSQL

Architecture and Design of Aurora DSQL

• Aurora DSQL uses a distributed, disaggregated architecture with separate components for:
  • Connection management
  • Query processing
  • Isolation enforcement
  • Transaction journaling
  • Storage
• This allows each component to scale independently and be optimized for its specific role.
• The query processor acts as the "heart" of the system, handling SQL execution and managing transactions.
• Transactions use snapshot isolation, with the adjudicator component resolving conflicts between concurrent transactions.
• The journal component ensures durability of committed transactions, replicating data across availability zones and regions.
• Aurora DSQL uses a global, synchronized clock provided by the Amazon Time Sync Service to coordinate transactions and ensure strong consistency.

Performance and Scalability

• Benchmarks from the presenters showed:
  • Single-region cluster: 20ms for 3 inserts, 12ms for 2 selects
  • Multi-region cluster: 40ms for 3 inserts, 12ms for 2 selects
• No noticeable "cold start" latency when reactivating a dormant database
• Ability to scale to 10,000 concurrent connections without performance degradation

Authentication and Authorization

• Aurora DSQL uses token-based authentication instead of passwords
• Tokens are generated by the AWS SDK and are short-lived, reducing the risk of credential compromise
• Integrates with AWS Identity and Access Management (IAM) for fine-grained access control

Developer Experience

• Seamless integration with popular ORMs like Hibernate, SQLAlchemy, and Django
• Provides connectors that handle token-based authentication automatically, reducing boilerplate code
• Supports familiar SQL syntax and tooling (e.g., psql, DB Beaver)

Constraints and Considerations

• Limitations on transaction duration (5 minutes), rows modified per transaction (3,000), and total data modified (10MB)
• Lack of support for some PostgreSQL features like foreign keys, sequences, and triggers
• Need to design applications to work within these constraints for optimal performance and consistency

Business Impact and Use Cases

• Enables serverless, event-driven architectures by providing a scalable, highly available database backend
• Reduces operational overhead and infrastructure management for database-backed applications
• Suitable for a wide range of use cases, from microservices to SaaS applications, where the constraints fit the workload

Conclusion

• Aurora DSQL provides a compelling serverless database option that combines the benefits of traditional relational databases with the scalability and flexibility of a cloud-native architecture.
• While there are some limitations compared to fully-featured PostgreSQL, the service is well-suited for many modern application development use cases.
• Careful consideration of the service's constraints and design patterns is required to maximize the benefits of Aurora DSQL.