Nimbus CPU usage

[Mitchell Rathbun (BLOOMBERG/ 731 LEX)]

When increasing the number of topologies on a worker node, I have noticed that 
the cpu usage for Nimbus also increases. I am guessing that this has to do with 
an increased number of heartbeats to keep track of/respond to. However, the 
supervisor cpu usage does not seem to change. I was under the impression that 
worker processes communicated with supervisor, which in turn communicated with 
Nimbus through Zookeeper. So why does only the Nimbus cpu usage seem to 
increase?