Explain how you'd handle a multi-region Kafka deployment for disaster recovery.

Options: 1) Active-Passive with MirrorMaker 2: primary cluster handles all traffic, MM2 replicates to DR cluster. RTO = time to switch apps to DR cluster (minutes). RPO = MM2 lag (seconds). 2) Active-Active: both clusters accept writes, MM2 does bidirectional replication. Requires handling cross-cluster deduplication (cycle detection via source header). 3) Stretched cluster: brokers across AZs (not truly multi-region). For DR: run MM2 as a separate Connect cluster, monitor lag continuously, test failover quarterly. Offset translation: MM2 checkpoints consumer group offsets so consumers can resume on DR cluster from approximately the right position.

Expected key concepts: Active-Passive, MirrorMaker 2, Active-Active, bidirectional, RTO, RPO, MM2 lag, cycle detection, stretched cluster, failover, offset translation, checkpoints