Performance Use Case - Memory Optimization

The service under test was REST implementation with other activities like Mapper, Render JSON, Parse JSON and Invoke REST API. The implementation had multiple mapper activities with iterator and accumulate output. The max heap of the appnode was set to 4 GB for these tests. The testing was focused on analyzing the memory usage of appnode under load.

Memory usage with bw.engine.enable.memory.saving.mode set to false

The following snapshot shows the JVM snapshot of the AppNode under heavy load with memory saving set to false.

Memory usage with bw.engine.enable.memory.saving.mode set to true

The following snapshot shows the JVM snapshot of the AppNode under heavy load with memory saving set to true.

Performance findings

  • Enabling memory saving reduced the heap usage of the application under heavy load.
  • Enabling memory saving did not degrade the performance of deployed services in terms of latency and throughput.
Note: The improvements showcased above should be used as reference. The performance impact of enabling memory saving mode may vary based on service implementation, workload, and payload on the system.