What is the best way to modify a json document ? Is it using insertOrReplace or DocumentMutation ? Is one method better performing than the other one ?
Conceptually the "mutation" would be more efficient as the size of the bytes on the network and the work to do on the cluster would be smaller.
This is especially true if the document is large.
So you can user both depending of the size of the documents and the business logic of your application.
Usually Mutation would be "cleaner", but "InsertOrReplace" is a easier to implement, especially if you are building a user interactive application.
Retrieving data ...