Brokers, storage and cross-AZ transfer β where a Kafka bill really goes.
Broker hours are visible; cross-AZ replication is the hidden majority of a Kafka bill. Model the transfer side on the cross-AZ transfer calculator and the storage on the object storage calculator.
The Managed Kafka Cost Calculator estimates your monthly bill for a managed Kafka service such as Amazon MSK or Confluent Cloud. It combines the three charges that make up most of that bill: compute, calculated from your number of brokers multiplied by the $ per broker-hour rate across the hours in a month; storage, based on the GB retained; and data transfer, driven by cross-AZ GB per month. Broker count and hourly rate are usually the largest drivers, but cross-AZ replication traffic can quietly become a major line item as throughput grows.
The key trade-off to watch is between broker sizing and redundancy costs. Fewer or smaller brokers lower compute spend but reduce headroom for traffic spikes, while multi-AZ replication improves availability yet adds cross-AZ transfer charges on every replicated message. Estimate your real retention needs and inter-zone traffic before committing, since storage and transfer often outlast the initial compute estimate.
Three things: broker instance hours, storage per GB, and data transfer. On a multi-AZ cluster the cross-zone replication traffic is often the single largest component β frequently more than half of total infrastructure cost β because every message is replicated across availability zones.
Compress messages, use fetch-from-follower to read within a zone, right-size replication, and consider rack-aware or single-AZ topologies where durability requirements allow. Networking, not brokers, is usually the biggest lever.
Cost at different volumes:
| Volume | Monthly | Yearly |
|---|