I understand both index data into Splunk.
Can I have it on a SH cluster ?
Difference is author and support model:
Documentation, if available, is linked to from the app's respective Overview or Details tabs.
Now, keep in mind that a Modular Input currently only scales vertically and will not work on a Search Head Cluster. So, it is recommended to have a Data Collection Heavy Forwarder where such things can run. If you'd like an even more scalable solution, then you should consider using the HEC to send the data to Splunk because then you can more easily horizontally scale the data collection. The blog post Splunking Kafka At Scale goes into rich detail.