What is the best way to index of documents when aggregating logs?

If they are different formats, then different indices is the best option.