You could go into Splunk's own metrics like this:
index=_internal component=metrics TERM(group=thruput) | timechart per_second(ev) by splunk_server
index=_internal component=metrics TERM(group=thruput) | timechart avg(instantaneous_eps) by splunk_server
The two use slightly different fields, and would produce vastly different results if your server isn't on all day.
Alternatively, you could also go into the actual data like this:
| tstats count where index=* OR index=_* by _time splunk_server | timechart per_second(count) by splunk_server
