Rebalance Storm — Scale Out or Tune?Kafka · hard · ~15 minThe clickstream-enrich consumer group (20 instances, 60 partitions) is in a rebalance storm. Throughput has collapsed to a third of normal and lag is climbing. Every few seconds the group rebalances, pausing all consumers while partitions are reassigned.An engineer has already opened a PR to double the group to 40 instances "to catch up." Your infra lead wants to know if that's the right call — and what it costs — before approving. Decide what actually fixes this.Evidence

Metrics

Consumer Logs

Config & Cost

Checkpoints

What is the core problem?

Why does one slow consumer take down the whole group's throughput?

The proposed fix is to double from 20 to 40 consumers ($0.10/instance/hour). What is the ADDED cost per 30-day month for the extra 20 instances? (answer in dollars, 730 hours/month)

Should you approve the scale-out, and what actually fixes the storm? Give the interview answer, including the cost reasoning.