What are your heavy forwarders doing, are you using them as intermediary forwarders?
If so, how many universal forwarders are sending how much data through this tier?
You can investigate using parallelIngestionPipelines as documented here to process multiple input streams in parallel.
Sorry for delay.
My architecture is UF > HF > IDX. But I have others data sources that use syslog.
I'll study the suggested article and send the News.
Side note: If you do not need to do any substantial filtering/data masking/routing on your HF tier, consider switching those to UF. Smaller footprint, higher throughput, less network bandwidth, etc.
Main point is: Have at least twice as many intermediary forwarders as you have indexers. You can achieve that logically with multiple ingestionPipelines on at least two intermediary servers (for availability)