Dynamic index mappings and autogenerated searches


I hope this is the correct category (please point me in the correct direction if it is not).

I want to give my users the ability to add custom data into an ES index and then search based on that custom data. I will be creating some kind of query language based on the ES index.

Hopefully that's enough information for that part as my question relates more to the size of index mappings and whether it's preferable to have one larger index or many smaller indexes.

I came up with two potential solutions:

  1. I can put all data into one index, there's an index mapping limit of 1000 and there's a recommendation to not increase this. Despite that, I would probably need to increase the index mapping limit to 10k to be future proof.

  2. I can create an index per user and keep the mapping limit of 1000.

Which of these solutions is preferable or are there other options available?


Welcome to our community! :smiley:

It depends on how many users you expect.
You could always denormalise things and consider using something like this one in the docs.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.