Optimize array query match with operator $all in M

Question

In a collection of 130k elements with the structure:

{
  "tags": ["restaurant", "john doe"]
}

There are 40k documents with "restaurant" tag but only 2 with "john doe". So the next queries are different:

// 0.100 seconds (40.000 objects scanned)
{"tags": {$all: [/^restaurant/, /^john doe/]}}

// 0.004 seconds (2 objects scanned)
{"tags": {$all: [/^john doe/, /^restaurant/]}}

It's there a way to optimize the query without sorting the tags in the client? The only way I can imagine now is putting less frequent tags at start of the search array.

Answer 1

I found a request feature for this in mongodb team JIRA:

https://jira.mongodb.org/browse/SERVER-1000

I implemented a stadistic system to put tags with more cadinality at the end of the array.