A benchmark of performances between three different stream processors: Kafka Streams, Apache Flink and Spark Streaming.
This benchmark uses Java for simulating a CPU intensive workload that is sent to three different stream processors:
- Kafka Streams
KafkaProcessor
- Apache Flink :
FlinkProcessor
- Spark Streaming:
SparkProcessor
We used JMX to collect the performance metrics and we stored them in Prometheus, Grafana is used for displaying results.
Below are example of results for Kafka Streams with Grafana.