[Verse 1] In the world of Strimzi where Kafka needs to run ZooKeeper holds the state for everyone Three replicas minimum, that's the golden rule For production workloads, keep your cluster cool Memory allocation starts at one gig base CPU cores depending on your cluster's pace Network bandwidth matters when nodes communicate Size it right or watch your performance deteriorate [Chorus] Configure your Zoo, allocate with care Memory CPU storage everywhere Persistent volumes keep your data safe Tuning parameters set the perfect pace Three two one, replicas must be odd Ticktime heartbeat, sync limit's your rod Configure your Zoo, make it bulletproof Strimzi's foundation underneath your roof [Verse 2] Persistent storage is where your logs will live Dynamic provisioning has so much to give Storage class selection for your volume claims SSD performance when you're playing big games Data directory mounted at var lib zookeeper Transaction logs need their own keeper Separate volumes for optimal I O flow Watch your disk space as your clusters grow [Chorus] Configure your Zoo, allocate with care Memory CPU storage everywhere Persistent volumes keep your data safe Tuning parameters set the perfect pace Three two one, replicas must be odd Ticktime heartbeat, sync limit's your rod Configure your Zoo, make it bulletproof Strimzi's foundation underneath your roof [Bridge] Tick time sets the heartbeat rhythm Two thousand milliseconds keep your system Sync limit controls the follower pace Init limit for startup's grace Max client connections set the door Auto purge keeps your disk from being sore Four letter words for monitoring health These parameters are your cluster's wealth [Verse 3] JVM heap size needs careful consideration Too little memory brings poor performance frustration Garbage collection tuning keeps your latency low Parallel collector helps your throughput grow Network threads handling all requests I O threads serving clients at their best Session timeout balancing client needs Recovery time when leadership succeeds [Chorus] Configure your Zoo, allocate with care Memory CPU storage everywhere Persistent volumes keep your data safe Tuning parameters set the perfect pace Three two one, replicas must be odd Ticktime heartbeat, sync limit's your rod Configure your Zoo, make it bulletproof Strimzi's foundation underneath your roof [Outro] From sizing to storage to parameters fine Your ZooKeeper cluster will work every time In Kubernetes pods with Strimzi's might Configure it well and sleep through the night
← 3 Deploying Your First Kafka Cluster | 1 KRaft Performance Characteristics →