What is the Kafka Mirror Maker 2 (MM2) and how does multi-cluster replication work?

MirrorMaker 2 (MM2, KIP-382) is Kafka's tool for replicating data between Kafka clusters, built on the Kafka Connect framework; the original MirrorMaker (MM1) was removed in Kafka 4.0. Its connectors consume from source clusters and produce to target clusters. The docs state it replicates topic data together with topic configuration and ACLs, and automatically detects new topics and partitions on the source. By default, replicated topics get the source cluster alias as a prefix (topic foo from us-west becomes us-west.foo); the separator and naming can be changed with a custom ReplicationPolicy. Consumer groups and their offsets can be replicated so applications can migrate between clusters: checkpoints carry offsets for each source consumer group, and with sync.group.offsets.enabled the translated offsets are periodically written to __consumer_offsets in the target cluster, as long as no active consumers of that group are connected to the target. Replication flows are directional: active/passive uses one direction (A->B) and active/active uses flows in both directions (A->B, B->A). MM2 can run as dedicated MirrorMaker processes or inside an existing Kafka Connect cluster.

Expected key concepts: MirrorMaker 2, MM2, Kafka Connect, KIP-382, prefix, checkpoints, sync.group.offsets.enabled, active/passive, active/active, ReplicationPolicy