Feels like this ought to be faster... doing the immediate return and delayed update dance helps alleviate the user's concern that the server has crashed, but it doesn't make the experience faster.
Can we either:
a) make it faster?
b) return success earlier, and delay the datastore write until after the user is notified of success? (This would be a sort of eventual consistency... would we run the risk of conflicting updates? In fact, do we already have such? Do we need to do some sort of front-end-level locking?)
Feels like this ought to be faster... doing the immediate return and delayed update dance helps alleviate the user's concern that the server has crashed, but it doesn't make the experience faster.
Can we either: a) make it faster? b) return success earlier, and delay the datastore write until after the user is notified of success? (This would be a sort of eventual consistency... would we run the risk of conflicting updates? In fact, do we already have such? Do we need to do some sort of front-end-level locking?)