users have tags, posts have tags (arbitrary tags).
user should find posts so the tags of the post are a subset of the users tags. Storing user/post tags in a jason datafield and doing a "select where contains/contained by" operation for every post is very expensive especially if you want to compare to 100k of posts. Thats why I need some kind of presortig altough I cannot Memchace to store for example a id list of the relevant posts because every user has a unique combination of tags. Therefore I need some sort of presorting to avoid repeated queries.
Maybe I should have columns for frequently searched tags and one column for a JSON with rare tags and indexing the frequently searched tags?