Mustafa Atik

In a distributed system, any operation may fail at any time leaving the change applied in some systems but hasn’t reached all the systems. So, this type of, half-done, changes are regarded as inconsistency.

In write repeat, calls from the clients are meant to fix, if there is, inconsistent state of data that is associated with the call.

  • smarter clients
  • retried until the call has been successfully completed.
  • idempotency is the key
  • retryable/non-retryable errors

Ref:

https://medium.com/airbnb-engineering/avoiding-double-payments-in-a-distributed-payments-system-2981f6b070bb

--

--

Mustafa Atik

Mustafa Atik

You can find my notes I take when learning something new or reading, watching. So, they only help me to refresh and remember what I’ve consumed.