Page History

...

1. Create a MutatorClient to manage a transaction for the targeted ACID tables. This set of tables should include any transactional destinations or sources. Don't forget to register a LockFailureListener so that you can handle transaction failures.
2. Open a new Transaction with the client.
3. Get the AcidTables from the client.
4. Begin the transaction.
5. Create at least one MutatorCoordinator for each table. The AcidTableSerializer can help you transport the AcidTables when your workers are in a distributed environment.
6. Compute your mutation set (this is your ETL merge process).
7. Optionally: collect the set of affected partitions.
8. Append bucket ids to insertion records. A BucketIdResolver can help here.
9. Group and sort your data appropriately.
10. Issue mutation events to your coordinators.
11. Close your coordinators.
12. Abort or commit the transaction.
13. Close your mutation client.
14. Optionally: create any affected partitions that do not exist in the meta store.

See ExampleUseCase and TestMutations.testUpdatesAndDeletes() for some very simple usages.