In the de-duplication approach, we give every message a unique identifier, and every retried message contains the same identifier as the original. In this way, the recipient can remember the set of identifiers it received and executed already. It will also avoid executing operations that are executed.

It is important to note that in order to do this, we must have control on both sides of the system: sender and receiver. This is because the ID generation occurs on the sender side, but the de-duplication process occurs on the receiver side.

Example

Imagine a scenario where an application sends emails as part of an operation. To send an email is not an idempotent operation. If the email protocol does not support de-duplication on the receiver side, we can’t be sure that every email displays exactly once to the recipient.