Tag

#distributed-systems

2 posts

Exactly-once is a memory problem
You will receive the same message twice. The fix isn't a stronger delivery guarantee — it's remembering what you already did, and one database primitive is enough to do it.
Splitting events by type breaks the only ordering you had
Three ways to lay out domain events across Kafka topics. One of them looks like the tidy option and quietly costs you the ability to rebuild state at all.

← All posts