Multi-Master replication stands for the ability to write to any node in the cluster, and not to worry that eventually it will get out-of-sync situation, as it regularly happens with regular MySQL replication if you imprudently write to the wrong server. This is a long-waited feature and there has been growing demand for it for the last two years, or even more.
With Percona XtraDB Cluster you can write to any node, and the Cluster guarantees consistency of writes. That is, the write is either committed on all the nodes or not committed at all. For the simplicity, this diagram shows the use of the two-node example, but the same logic is applied with the N nodes:
All queries are executed locally on the node, and there is a special handling only on COMMIT. When the COMMIT is issued, the transaction has to pass certification on all the nodes. If it does not pass, you will receive ERROR as a response on that query. After that, transaction is applied on the local node.
Please note that applying the transaction on remote nodes does not affect the response time of COMMIT, as it happens in the background after the response on certification.
The described behavior of COMMIT also has the second serious implication. If you run write transactions to two different nodes, the cluster will use an optimistic locking model. That means a transaction will not check on possible locking conflicts during the individual queries, but rather on the COMMIT stage, and you may get ERROR response on COMMIT. This is mentioned because it is one of the incompatibilities with regular InnoDB that you might experience. In InnoDB usually DEADLOCK and LOCK TIMEOUT errors happen in response on particular query, but not on COMMIT. It’s good practice to check the error codes after COMMIT query, but there are still many applications that do not do that.
If you plan to use Multi-Master capabilities of XtraDB Cluster and run write transactions on several nodes, you may need to make sure you handle response on COMMIT query.