I started a demo by docker compose with this distribute example .
I didn’t change anything except that I added the option to receive jaeger_json from kafka under distribute option.I allocated 10GB and 6 cpu for all of the containers.
All traces has only one span in kafka.
But it performed not well,it only consume about 1k spans per seconds from kakfa.
I found the Memory and cpu usage is not very high.
I tried to add a few distributors, but still only one of them did well.
I tested reading kafka without processing it through a script, it can be read through very high speed, so kafka should be no problem.