Bulk Indexing performance on AWS ES service

How large are the documents? How many nested objects/levels? How complex are the scripts performing the update? Which version of Elasticsearch are you using?