Is there is known difference in performance on N1QL vs get/upsert for one document ? As I upsert my doc and after it need to query it for values I wonder if it’s not better to use one way for it and go N1QL for both.
If you operating on single document and know the docId, no filtering needed you can use get/upsert, You can also consider sub-doc option.
If you need to search by condition (no docIds) and do predicate (processing) N1QL is better option.