I'm trying to do in one query, the 10 most rented, display a text field (name) of another document that are in the same index but are of different types.
Example:
{
"id": 1,
"id_book": 1
"type": "books"
"name": "My book one"
}
So I guess it was not a real use case but an example as normally a book is pretty much stable once it got printed.
Anyway, it depends. How many instances are going to update? 10? 1000? 10000? 10000000? Depending on that it can time lot of time or nothing.
Would be better to talk about the real use case. But for a use case I have, it takes me around 1 or 2 minutes to index locally 1 million documents (again for my use case). So I'm totally fine having to reindex 1m documents if a value is changing like the book title in your case.
Otherwise, you can look at parent/child feature if you really need it. See the bold text? My advice is to keep things as simple as possible. Parent/child feature comes with some drawbacks so I'd use that only on last resort IMHO.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.