All Products
Search
Document Center

PolarDB:Distributed transactions

Last Updated:Mar 28, 2026

PolarDB-X 1.0 uses two-phase commit (2PC) to guarantee the atomicity and consistency of distributed transactions.

How it works

2PC splits each distributed transaction into two sequential phases:

  1. Prepare phase — Each data node (ApsaraDB RDS for MySQL instance) acquires all resources needed to commit the transaction, including resource locking and logging.

  2. Commit phase — After every data node has successfully completed the prepare phase, the PolarDB-X 1.0 server, acting as the transaction manager, sends a commit request to each data node. Each node then commits its portion of the transaction.