[Verse 1] ZooKeeper's been the guardian for so many years Managing metadata, calming all our fears But now there's a new way to handle the load KRaft consensus takes us down a different road No more external systems to maintain and scale Kafka's got its own story to tell [Chorus] Leader election, log replication Committed entries through the nation KRaft controller quorum standing strong No ZooKeeper needed, we've moved along Metadata topics flowing free This is how it's meant to be [Verse 2] Raft protocol with leaders at the helm Three phases guide us through this realm Follower, candidate, then leader takes control Heartbeats keep the system on a roll Log entries replicated across the quorum Majority consensus, that's the forum [Chorus] Leader election, log replication Committed entries through the nation KRaft controller quorum standing strong No ZooKeeper needed, we've moved along Metadata topics flowing free This is how it's meant to be [Bridge] Migration path from old to new Rolling upgrade sees you through First the controllers make the switch Then the brokers join without a glitch Dual-write mode helps you transition Legacy support during intermission [Verse 3] Kafka four point oh brings the change ZooKeeper mode is out of range Faster failover, seconds not minutes Partition scaling with no limits Metadata log stores the state Performance improvements truly great [Chorus] Leader election, log replication Committed entries through the nation KRaft controller quorum standing strong No ZooKeeper needed, we've moved along Metadata topics flowing free This is how it's meant to be [Outro] From ensemble old to quorum new KRaft's the future coming through Controllers handle all the load Welcome to the KRaft mode
← 5 Cross-Cluster Replication and Active-Active | 2 Apache Kafka Fundamentals →