Anonymous
The design involves using a queue like Kafka - first I will put the message to Kafka and process - in order to integrate - I will have a unique ID associated with the ticket and when an update to the ticket happens I will use the uniqueID to find the record and update. In order to make sure that we have no write overlaps we will use row level locking - this might not scale well in the distributed setting - if so we will evlove this with redis distrubted lock.
