I have an index with 20 000 000 documents, each document having ~5 nested objects.
As you can imagine, the update takes a long time, sometimes around 2 minutes.
I was thinking on spliting the big index in multiple indices having documents on one year interval.
I will have 2 aliases: index_write -> the current index and index_read -> all the indices.
So, the update time it will improve.
Now, about the query time, I am wondering how the search on aliases does work? It searches all the indices in paralel time? or sequentially?
Also, I was wondering if it's more time-efficient to search on an alias pointing to small indices, or to search on a big index?