I had the same issues a while back. Once I deployed Splunk On Splunk to all of our indexers I was able to pinpoint exactly what sourcetype/source/host was pushing us over our indexing limit. When in the S.O.S. app, I go to Indexing -> Distributed Indexing Performance. At the top of that page is a section for Real-Time measured indexing rate and latency. You can run that against source, host, sourcetype, index or Splunk server. There are several other sections within the app that will help find the root cause.
https://apps.splunk.com/app/748/
... View more