Question

A retail company runs an order-processing microservice that publishes account balance-update events to an Amazon Kinesis Data Stream. A downstream analytics team must materialize the current balance per customer in Amazon S3. During load testing, the team notices that events for the same customer are sometimes processed out of order, producing incorrect final balances. The producer currently uses a random UUID as the partition key. What is the MOST effective change to guarantee correct ordering for each customer's events?

Accepted Answer

Set the partition key to the customer ID so all events for a customer route to the same shard and preserve ordering within that shard.

Answer

Increase the number of shards in the stream so each customer's events get more throughput and process faster.

Answer

Enable enhanced fan-out on the consumer so each consumer receives a dedicated 2 MB/s throughput pipe.

Answer

Switch the delivery to Kinesis Data Firehose with a Lambda transform that sorts records by timestamp before writing to S3.