A host's global status is refreshed every time a new report comes in, which could be tens of thousands or more of refreshes every half hour for a large setup using puppet.
This means that the report handling process should be as fast as possible to prevent a bottleneck.

https://github.com/theforeman/foreman/pull/3447#issuecomment-235512793 includes a suggestion for a possible optimization, but more testing and benchmarking should be done to identify other possible bottlenecks.