CONCEPT Cited by 1 source
Atomic distributed transaction¶
Definition¶
An atomic distributed transaction is a multi-participant write protocol (typically 2PC or a 2PC variant) that provides all-or-nothing atomicity across multiple database participants — every participant either commits the full change-set or rolls back. The protocol costs a coordinator round-trip per participant per commit phase plus durable coordinator state for recovery. This is the strong-atomicity endpoint of the cross-shard-write design space; the Consistent Lookup Vindex + ordered-commit-without-2PC shape is the weaker-but-cheaper endpoint used when one participant is authoritative and the others are derivatives. In Vitess, atomic distributed transactions are the option for application-level cross-shard writes that require strict atomicity, while ordered-commit is the routing-metadata option.
Seen in¶
- sources/2026-04-21-planetscale-announcing-vitess-21 — Vitess 21 reintroduces atomic distributed transactions with a revamped design, "deeper integration with core Vitess components and workflows, such as Online DDL and VReplication (including operations like MoveTables and Reshard)", simplified configuration. Still experimental. VTAdmin gains a dashboard to view and conclude distributed transactions. Canonical wiki datum that Vitess offers both endpoints of the cross-shard-write design space in the same release.