I am dynamically creating application-specific spreadsheets within the bucket, where “type” is the name of the spreadsheet. I have to allow an unbounded number of spreadsheets to be created with search and query.
Since I don’t know upfront which field to index. I am statically creating fields “f1”, “f2”, … “fn” and indexing them upfront while mapping application data to these fields in the application layer.
Additionally, I have to pre-create the indexes per data type, for example, 5 indexes for number, 5 indexes for string, etc. and fit my spreadsheet into this structure.
So the question is how many indexes I can create per bucket? Assume fixed-latency architecture with all the updates to all the indexes to complete within 1 sec on a high-end datacenter with all the bells and whistles including memory-optimized indexes of enterprise edition with two replicas.
Inverting this question, in 1 sec how many indexes can be updated?
Also, are index updates triggered with “doc” updates or “field” updates. Hopefully, field since I want to take advantage of “mutateIn” field level APIs.
Many thanks for your answer.