This note tracks the Apache Druid work for dynamically auto-scaling Kafka stream ingestion tasks.
Operational question
When event volume changes, ingestion task count should follow the workload without manual babysitting or wasteful over-provisioning.