We tend to use state machines for massively asynchronous operations like order management (with days between updates), as well as a home-built library for handling the task. It's incredibly fault tolerant and offers really easy abstraction and automation.