3 Capacity Planning

Kafka on Kubernetes with Strimzi · 5:06

Listen on 93

Lyrics

[Verse 1]
When planning Kafka clusters on Kubernetes ground
Three factors multiply to keep your data sound
Retention times throughput and replication too
Storage requirements calculated just for you
Days times gigabytes times copies that you make
Equals total storage space your cluster needs to take

[Chorus]
ReTention Throughput Replication multiply
Storage sizing made simple reach up to the sky
Two gigs CPU four gigs RAM per broker start
Network bandwidth doubles when replicas depart
Plan your partitions wisely scale your Kafka heart

[Verse 2]
Each broker needs resources to handle all the load
Two CPU cores minimum to walk the data road
Four gigabytes of memory baseline for the game
More producers more consumers fan the resource flame
Scale up with your workload don't leave your brokers dry
Monitor and adjust as traffic volumes fly

[Chorus]
ReTention Throughput Replication multiply
Storage sizing made simple reach up to the sky
Two gigs CPU four gigs RAM per broker start
Network bandwidth doubles when replicas depart
Plan your partitions wisely scale your Kafka heart

[Bridge]
Network flows between the brokers replication stream
Double your producer bandwidth living the dream
Partition count guidelines keep performance tight
Ten thousand max per cluster keeps everything right
Balance across your brokers spread the load around
Strimzi makes it easier Kafka safe and sound

[Verse 3]
Start with topic size and growth rate that you see
Multiply by replicas two or maybe three
Cross-cluster traffic patterns network load design
CPU spikes with compression memory aligns
Test your calculations in development first
Before production traffic puts you to the test

[Chorus]
ReTention Throughput Replication multiply
Storage sizing made simple reach up to the sky
Two gigs CPU four gigs RAM per broker start
Network bandwidth doubles when replicas depart
Plan your partitions wisely scale your Kafka heart

[Outro]
Capacity planning victory Strimzi shows the way
Kafka clusters running strong every single day

← 3 Distributed Tracing | 3 Cruise Control for Rebalancing →