Tuning Topologies
Also available as:

Tune Batch Indexing Kafka Partitions

The first batch indexing variable that should be determined is the minimum number of Kafka partitions required.

Use the following formula to determine the minimum number of Kafka partitions for the enrichment toplogy
Num Partitions=Max( t/p,t/c)
  • t is the desired throughput
  • p is the maximum throughout using a single producer
  • c is the maximum throughput using a single consumer
You can estimate the minimum number of partitions requres based on the original calculations performed for enrichment topologies. However, this may or may not be suitable as the message size increases as it progresses through Metron topologies. This increased message size can affect throughput. For this reason, we recommend that you perform the same steps to calculate the value for the batch indexing topic.