What happened
On 31 July 2026 AWS expanded Amazon Aurora DSQL multi-Region cluster support to four more Regions — Stockholm, Spain, Mumbai and Singapore. Generally available since May 2025, Aurora DSQL is a serverless, PostgreSQL-compatible distributed SQL database that AWS describes as offering active-active high availability and multi-Region strong consistency: a single logical database with a writable endpoint in each peered Region that remains available if one Region fails, with stated availability of 99.99% single-Region and 99.999% multi-Region.
Why it matters
Strongly-consistent, active-active writes across regions have historically meant operating a CockroachDB- or Spanner-class system yourself, or settling for eventual consistency. Delivering it as a serverless managed database behind a Postgres endpoint changes the build-versus-buy math for global applications — the hard part, cross-region consistency with regional failover, becomes someone else's operational problem.
'Managed' relocates the distributed-systems trade-offs; it doesn't repeal them. Aurora DSQL uses optimistic concurrency control — no locks, conflicts detected at commit — so it shines under low contention but surfaces commit-time conflicts (and cross-region write latency) under hot-row contention. It also isn't drop-in Postgres: foreign-key constraints aren't supported, so referential integrity moves into your application. The Staff move is to model your real contention and consistency needs before assuming strong multi-region is now free.
A team wants strongly-consistent writes across two regions. Walk through the latency and failure trade-offs, and when you'd choose a system like Aurora DSQL over async replication or a single-region primary.
Probes CAP/PACELC reasoning, cross-region write latency, and consistency-vs-availability judgment.