Amazon Aurora DSQL: The Launch of Infinitely Scalable Serverless SQL Database

AWS’s new distributed database technology breaks traditional database boundaries

https://youtu.be/aim5x73crbM?si=Ur3aLJQdxnsL6RmC&t=5162

Amazon Web Services has announced the general availability of Aurora DSQL, a database that has been in development for an extended period. This innovation, first introduced as a preview at re:Invent 2024, offers truly revolutionary features in the database world.

What is DSQL and Why Does It Matter?

Aurora DSQL (Distributed SQL) is designed with a modular approach, unlike traditional databases. This system, embracing Unix philosophy, ensures each component performs a single task perfectly. Independent components like query processor, adjudicator, journal, and crossbar communicate through well-specified APIs and scale independently based on workload.

The system’s most striking feature is undoubtedly its availability rates: 99.99% in single region and 99.999% availability guarantee in multi-region deployments. This means only a few minutes of downtime per year — an almost unimaginable level for most enterprise applications.

Technical Architecture Deep Dive

The story behind DSQL is actually a reflection of Amazon’s struggle with scale problems. To overcome the complexity created by traditional two-phase commit (2PC) approaches, they adopted a strategy of writing the entire commit to a single journal. While this satisfied atomic and durability requirements, it made the read path more complex.

The Crossbar component developed as a solution separates the scaling of read and write paths. Storage nodes can subscribe to specific key ranges, and Crossbar routes updates to the correct nodes. While it sounds simple, serious performance issues like garbage collection pauses were encountered during implementation.

The Journey from JVM to Rust

Perhaps the most interesting part is the project starting 100% with JVM and completing with 100% Rust. When simulation tests aimed for 1 million TPS with 40 hosts but only achieved 6,000 TPS, and tail latency increased from 1 second to 10 seconds, it pushed the team to make radical decisions.

The decision to switch to Rust seemed risky, but the results were surprising. Code written by Java developers new to Rust performed 10 times faster than carefully optimized Kotlin implementation. The jump from 2,000–3,000 TPS to 30,000 TPS made accepting the “hangover first” principle worthwhile.

PostgreSQL Integration and Memory Safety

The DSQL team chose to use extension points rather than hard forking PostgreSQL’s millions of lines of C code developed since 1986. While writing extensions in C initially seemed logical, discovering memory safety issues during code reviews supported the decision to switch to Rust.

According to Android team research, the vast majority of bugs come from new code. Preventing memory safety bugs from the start by using Rust was seen as a more sensible long-term strategy.

New Features and Integrations

The features coming with GA are quite comprehensive:

• Improved console experience and multi-region cluster management
• PostgreSQL views, unique secondary indexes, and Auto-Analyze support
• AWS Backup, PrivateLink, CloudFormation, and CloudTrail integrations
• AI-powered natural language queries with Model Context Protocol (MCP) server

The MCP feature is particularly noteworthy. By integrating with Amazon Q Developer CLI, it offers the ability to explore database schema and execute complex SQL queries using natural language.

Pricing and Availability

DSQL uses a normalized billing unit called Distributed Processing Unit (DPU). The AWS Free Tier offers 100,000 DPUs and 1 GB storage free monthly.

Currently available in US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Osaka, Tokyo), and Europe (Ireland, London, Paris) regions.

Conclusion

Aurora DSQL is being evaluated as the biggest product launch since Lambda and DynamoDB. It appears ideal especially for microservices, event-driven architectures, and multi-tenant SaaS applications.

Of course, it may not be the right choice for every project — modern Java implementations offer sufficient performance for many services. What matters is making architectural decisions based on project-specific requirements, team capabilities, and operational environment.

Ultimately, it’s no exaggeration to say that Amazon’s bold move will establish new standards in the distributed database world.

Sources:

1. Amazon Aurora DSQL, the fastest serverless distributed SQL database is now generally available — https://aws.amazon.com/blogs/aws/amazon-aurora-dsql-is-now-generally-available/
2. Amazon Aurora DSQL Goes GA: Distributed, PostgreSQL-Compatible Serverless Database — https://www.infoq.com/news/2025/06/amazon-aurora-dsql-ga/
3. Just make it scale: An Aurora DSQL story — https://www.allthingsdistributed.com/2025/05/just-make-it-scale-an-aurora-dsql-story.html
4. Amazon Releases Distributed Aurora Database — https://www.i-programmer.info/news/197-data-mining/18077-amazon-releases-distributed-aurora-database.html
5. Amazon launches serverless distributed SQL database, Aurora DSQL — https://sdtimes.com/data/amazon-launches-serverless-distributed-sql-database-aurora-dsql/
6. ​​AWS re:Invent 2024 — Dr. Werner Vogels Keynote — https://youtu.be/aim5x73crbM?si=Ur3aLJQdxnsL6RmC&t=5162

Note: This article has been compiled from the news sources listed in the references section. For more detailed information, please refer to the source articles. For the Turkish version of this article, please visit https://hobimiz-teknoloji.com/amazon-aurora-dsql-sonsuz-%C3%B6l%C3%A7eklenebilir-serverless-sql-veritaban%C4%B1n%C4%B1n-genel-kullan%C4%B1ma-a%C3%A7%C4%B1lmas%C4%B1-8ecd4f8b8891.