arXiv: 2607.13276 · PDF: 2607.13276.pdf

Authors: Marc Brooker, Marc Bowes, et al. (Amazon Web Services)

TL;DR

Aurora DSQL is AWS’s new serverless, PostgreSQL-compatible OLTP database designed for multi-region active-active writes. The architecture disaggregates compute (Firecracker MicroVMs running stateless SQL), storage, and transaction coordination into independent horizontally-scalable services. It uses MVCC with precision timestamps for coordination-free reads and optimistic concurrency control for writes, deferring all coordination to commit time via distributed adjudicators and a Journal replication tier.

Why it matters

Every serverless SQL database eventually collides with the same wall: coordination overhead scales linearly with cross-region latency, and traditional 2PC-style multi-region writes make commit p99s unusable. DSQL’s bet is that if you push coordination out of the hot path — statements execute optimistically against a consistent snapshot, and only commit pays the cross-region cost — you can offer strong consistency and full ACID without giving up elastic zero-to-millions-of-TPS scaling.

The key architectural moves:

  • Stateless query processors in Firecracker. No local state means scaling is a matter of spinning MicroVMs.
  • Precision timestamps anchor MVCC across regions without a single sequencer bottleneck (the classic Spanner-vs-Calvin trade-off, resolved differently here).
  • Journal + adjudicators absorb the commit-time contention check. Writes only synchronize at commit, so read-heavy workloads and low-conflict writes pay almost nothing.
  • Continuous availability through AZ/region failures follows from the disaggregation — each tier fails independently.

Read this if

You’ve been tracking the “post-Spanner” wave of cloud OLTP systems (CockroachDB, YugabyteDB, TiDB, PlanetScale, Neon), or you care about the architectural trade-offs behind “serverless SQL” claims. This is the reference doc for how AWS approached the problem, and it clarifies which parts of Aurora v1 they kept vs. re-architected.

Bibliography

@article{Brooker2026_dsql,
  title = {Aurora DSQL: Scalable, Multi-Region OLTP},
  author = {Brooker, Marc and Bowes, Marc and others},
  year = {2026},
  eprint = {2607.13276},
  archivePrefix = {arXiv},
  primaryClass = {cs.DB},
  url = {https://arxiv.org/abs/2607.13276}
}