Assuming you have multi-site indexer cluster across 2 sites, what's your site search and replication factor? How many copies do you want to store in each site when a new data comes in? Look at your failover and storage considerations.
e.g. any data coming in site 1 can be stored in 2 indexers (or more if you want) in the originating site and another 2 copies in the other site. This will ensure equal amount of data in each site and fault tolerant up-to 2 nodes in a site.
Look at the examples in https://docs.splunk.com/Documentation/Splunk/7.2.3/Indexer/Sitereplicationfactor#Examples and your use cases, available storage to determine/adust Site SF/RF. Generally, distributing the copies equivally in both sites should give you a better performance. Also, if the users are across the world, you may also need to look at balancing them to both sites equally
https://docs.splunk.com/Documentation/Splunk/7.2.3/Indexer/Sitesearchfactor
