The telemetryd adapters currently run in the OpenNMS JVM. In order to increase the speed at which flows can be processed you currently need to either A) Scale vertically by adding more resources to the host running the OpenNMS JVM or B) Run multiple instances of OpenNMS in parallel. A) is not always feasible, and B) introduces a lot of complexity.
Another approach would be to allow the adapters to run outside the OpenNMS JVM, and allow these to be scaled out independently.
The goal of this epic is to evaluate the feasibility and determine the best approach for building a new Karaf container on which the telemetryd adapters can run.
Initial POC and research is available here: https://github.com/OpenNMS/opennms/blob/jw/stooge-poc/STOOGE.md