If your data is distributed across multiple
partitions in the data grid, you can read and update several partitions
in a single transaction. This type of transaction is called a multi-partition
transaction and uses the two-phase commit protocol to coordinate and
recover the transaction in case of failure.