Here is the link to Splunk Doc dedicated to the topic:
HF does not demand super-dome compute power and hence, it makes more sense to have a dedicated HF to filter data and give more headroom to the Indexer for its core function.
You may want to completely edit (or perhaps better - completely resubmit) this question. And don't use the code text tagging for the entire text so that you can have a reasonably viewable question that can contain the bold, etc. I'm not complaining or anything, but it is very hard to figure out what it is you want with it formatted as it is.
I agree with Cary's response.
Also its not mandatory to use HF if you want to filter raw logs. Indexer can very well do that. HF occupy one additional box & may become bottleneck in large deployment.