Opening Scenario
Your Kafka cluster lives in us-east-1 and serves ~80% of traffic. The product is expanding to Europe; the EU latency on producers and consumers is starting to hurt, GDPR makes the data-locality conversation unavoidable, and the SRE team is nervous about a single-region single point of failure. Three asks land on your desk in the same week: (1) European consumers should read locally, not across the Atlantic; (2) if us-east-1 goes dark, EU traffic should keep flowing — degraded but alive; (3) finance wants exactly one source of truth for revenue events, not "two slightly-different streams." The architect mentions MirrorMaker 2 and the words "active-active." Before you commit to a replication topology, you have to understand what's actually possible, what each topology costs, and what new failure modes you're signing up for.
Question Progression
1. Why is cross-region replication different from within-region replication?
Question. Kafka already replicates partitions across brokers via [[replication-and-isr]]. Why isn't "just add brokers in eu-west-1" the…
In the full tutorial
- When to Reach for This