Production Cluster Configuration for small clusters


We just launched a small cluster (around 10gb of data). We have been playing around with the JVM and Druid configuration for the past couple of days, but we can’t seem to get it right. Is there anyone, more well versed in this topic, who could suggest a proper configuration for a small cluster? Any help would be greatly appreciated.


Hi Tk,

Druid configs mostly depend on the Hardware configs being used and tuning them is an iterative process.

you can find answers to some of the common tunings here -

If you still face issues in tuning the cluster, you can post your runtime.props, jvm args and hardware configs to provide more details.