[Verse 1] In the streaming world where data flows Kafka's architecture everyone should know Brokers hold the topics in their care Partitions split the load for all to share Segments store the messages in files Offsets track position all the while [Chorus] Brokers Topics Partitions all in line Segments Offsets keeping perfect time Producers Consumers in the streaming dance Kafka core gives your data flow a chance [Verse 2] Producer sends with acks of zero one or all Zero fire forget might lose the call One waits for leader confirmation All replicas for durability's foundation Idempotency keeps duplicates away Exactly-once semantics saves the day [Chorus] Brokers Topics Partitions all in line Segments Offsets keeping perfect time Producers Consumers in the streaming dance Kafka core gives your data flow a chance [Verse 3] Consumer groups divide the workload fair Round robin range or sticky assignment there When members join or leave the group Rebalancing kicks in to regroup Partition ownership shifts around Until stability is found [Bridge] Log compaction keeps the latest key Time retention clears what's old and free Schema Registry guards the format tight Avro Protobuf JSON in sight Evolution modes control the change Backward forward full compatibility range [Chorus] Brokers Topics Partitions all in line Segments Offsets keeping perfect time Producers Consumers in the streaming dance Kafka core gives your data flow a chance [Outro] From broker down to schema at the gate Kafka's architecture seals your streaming fate Core components working as designed Defense infrastructure peace of mind
← 5 Multi-Cluster and Active-Active Patterns | 2 KRaft — The ZooKeeper Replacement →